Eur J Appl Physiol (2004) 92: 312–320 DOI 10.1007/s00421-004-1099-0
O R I GI N A L A R T IC L E
N. Desvigne Æ J. C. Barthe´le´my Æ F. Bertholon J. P. Gay-Montchamp Æ D. Freyssenet Æ F. Costes
Validation of a new calibration method for human muscle microdialysis at rest and during exercise
Accepted: 3 March 2004 / Published online: 20 April 2004 Springer-Verlag 2004
Abstract Microdialysis presents the unique possibility to measure metabolite concentrations in human interstitial fluid. During exercise, the recovery of these metabolites should be precisely monitored since it is known to increase greatly with muscle blood flow. The loss of ethanol, perfused at low concentration, can be accurately measured and reflects the changes in dialysis conditions. We evaluated whether using the relationship determined in resting metabolic conditions between the loss of ethanol, as reference substance, and the recovery for lactate or glucose would allow us to calculate precisely the concentration of these substances and their variations during exercise. Using the new catheter calibration method (slope method), the error of estimation of lactate and glucose in vitro was limited to )0.6 (5.8)% and )0.7 (6.2)%, respectively. In resting human muscle, the slope method proved to be as accurate as an established calibration technique (‘‘no net flux method’’) to evaluate interstitial lactate concentration [1.82 (0.58) vs 1.83 (0.47) mM, respectively]. During dynamic knee-extension exercise or light neuromuscular electrical stimulation, the estimated interstitial lactate and glucose concentrations varied differently, but their time course changes remained consistent with their respective plasma values. We conclude that, after an initial calibration step, the slope method allows accurate measurement of interstitial muscle metabolites and it could be used to monitor rapid metabolic changes during exercise. Keywords Ethanol Æ Interstitial fluid Æ Lactate
N. Desvigne Æ J. C. Barthe´le´my Æ F. Bertholon Æ D. Freyssenet F. Costes (&) Laboratoire de Physiologie GIP-E2S, C.H.U. de Saint Etienne–Hoˆpital Nord, 42055 Saint Etienne cedex 2, France E-mail:
[email protected] J. P. Gay-Montchamp Laboratoire Central de Pharmacologie et Toxicologie, C.H.U. de Saint Etienne–Hoˆpital Bellevue, 42055 Saint Etienne cedex 2, France
Introduction Microdialysis provides a unique opportunity to identify substrates and measure their concentration changes in the muscle interstitial environment (de la Pena et al. 2000; Henriksson 1999). However, a major methodological limitation lies in the difficulty of knowing the actual interstitial substance concentrations as well as their temporal profile while the recovery for such substances varies greatly in vivo in response to a change of environmental conditions (MacLean et al. 1999). Therefore, several calibration methods of microdialysis probes have been described to resolve this experimental drawback. One possibility is to find out the point where the dialysate and the perfusate concentrations are the same using successive different concentrations of the perfusate [’’no net flux (NFF) method’’; Lo¨nnroth et al. 1989] or to obtain a complete recovery by using a very slow perfusion flow rate (‘‘zero flow method’’; Bolinder et al. 1992; Hagstrom-Toft et al. 1997). However, both processes require the metabolism to be stable and they cannot be applied during exercise. In order to monitor fast changes, an internal reference method, using the labelled molecule of interest, has recently been developed (Langberg et al. 1999a, b, c; MacLean et al. 1999). While very efficient, this method also has some limitations. Indeed, it is sometimes difficult to obtain the labelled substance of interest and the use of radioactive elements for human research purposes is questionable. Also, it has been established that this method leads to a slight overestimation of probe recovery (Arner 1999). Recently, Strindberg and Lo¨nnroth (2000) proposed an alternative method. Their endogenous reference method relies on the observation that the ratio of probe recovery between a substance to be studied and a reference substance calculated in vitro equals the ratio obtained in vivo [’’probe recovery ratio (PRR) method’’]. This method also assumes a constant plasma to interstitial gradient for a reference substance such as
313
urea or inulin (Sjo¨strand et al. 1999, 2000; Strindberg and Lo¨nnroth 2000). The in vitro probe recovery of the reference substance is calculated from the dialysate concentration of that substance and its interstitial concentration is obtained from the plasma value and the supposed constant interstitial–plasma gradient. The recovery of the substance of interest can then be calculated using the in vitro PRR and the probe recovery for the reference substance. However, the constant plasmatic–interstitial gradient for the reference substance, assumed in this method, could introduce bias measurements during exercise. Then, we reasoned that the probe recovery of the reference substance should be precisely measured with the dialysis condition occurring during exercise, such as the increase of blood flow. These characteristics have been extensively demonstrated for ethanol (Hickner et al. 1992, 1994, 1997; Radegran et al. 1998; Stallknecht et al. 1999), the loss of which can be precisely calculated from its outflow/inflow ratio. We hypothesized that the determination of an individual probespecific relationship between the loss of ethanol and the recovery of lactate or glucose will improve the precision of interstitial substance measurements as compared with the use of an averaged PRR. In the present paper, we thus evaluated the precision of a new calibration method (‘‘slope method’’) as compared with the PRR method to provide accurate estimations of interstitial lactate and glucose. Finally, we found in human muscle that the slope method yields reliable measurements of lactate and glucose during different conditions: rest and increased local blood flow with (dynamic exercise) or without (light isometric transcutaneous neuromuscular electrical stimulation) significant metabolic changes.
Methods Microdialysis catheters The same types of catheters were used for all in vitro and in vivo studies. Microdialysis probes were designed as described by MacLean et al. (1999). Briefly, a microdialysis fibre obtained from an artificial dialysis kidney (Tricea High Flux Cellulose Triacetate Dialyser Model TCA-110G, Baxter, Ill.) was glued at both extremities to nylon tubes (Standard G, 0.5 mm ID, 0.63 mm OD, Portex, UK). The cut-off of the dialysis membrane was not precisely known but it was possible to dialyse molecules as large as 20 kDa. A suture (Vicryl 6/0 Ethicon, Brussels, Belgium) was also attached to the inside of the tubes to increase the strength of the probe. The actual microdialysis probe length was 3 cm. Before use, the probes were sterilized (Sterrad system: low-temperature hydrogen peroxide gas plasma). The microdialysis probes were attached to a perfusion pump (Harvard Apparatus PHD 2000 Series, Mass.).
Calibration of the probes using the PRR method Ten different probes were used in vitro to calculate the mean ratio between the loss of the substance of reference, ethanol, and the probe recovery of lactate or glucose. They provided 54 dialysates which were divided into two sets: a first series of 36 dialysates (calibration set) allowed us to calibrate the probes and a second set of 18 dialysates (validation set) were used to evaluate the accuracy of the calculation of lactate and glucose concentrations using the averaged ratio established from the calibration set. The probes were perfused with our standard solution (0.9% NaCl solution plus 5 mM ethanol). They were immersed in a bath containing 0.1% bovine serum albumin (Sigma, Saint Quentin Fallavier, France) in phosphate-buffered saline (Sigma), which were added to obtain a fluid similar to in vivo dialysis conditions. During the course of the experiment, bath lactate and glucose concentrations varied either simultaneously or separately by adding 1–5 mM lactate (Aguettant, France) and 4.5–7.5 mM glucose (Renaudin, France) sequentially to the bath. In order to obtain a large range of probe recovery for lactate and glucose, and for the loss of ethanol, the probes were perfused at different flow rates, from 2 to 15 ll min)1. During the experiments known as the ‘‘calibration set’’, only the perfusion rate varied, while lactate and glucose concentrations in the bath were kept constant. By contrast, during the experiments known as the ‘‘validation set’’, after a dialysate has been collected, we sequentially changed one or several parameters (perfusion rate, bath lactate and/or glucose concentration), in order to obtain various dialysis conditions. Lactate and glucose concentrations were measured in each dialysate of the calibration set obtained with the different combinations of bath substance concentration and perfusion flow rate. Recovery for lactate (RecLac) and recovery for glucose (RecGlu) were calculated as follows: [substance]dialysate/[substance]bath. Similarly, ethanol concentration was measured in each dialysate (D) and perfusate (P), and loss of ethanol (LossEth) was calculated on the same set of calibration dialysates as follows: 1)([Eth]D/[Eth]P). Applying the PRR method, we then calculated the recovery for each substance in the dialysate of the validation set using LossEth of this dialysate and the averaged recovery for substance (RecSubstance)/ LossEth ratio calculated from the calibration set, as follows: RecSubstanceD = (LossEthD)(RecSubstance/ LossEth)calibration set. The errors obtained in the estimation of the bath values were expressed as percentages of the difference between the calculated values and the true values in vitro. Sample analysis All the microdialysis samples were weighed and then immediately frozen for subsequent analyses. Lactate and
314
glucose concentrations were enzymatically determined in duplicate for a 25-ll sample using a StatPlus 2300 analyzer [Yellow Springs Instruments (YSI), Ohio]. Ethanol was measured by gas phase chromatography (HewlettPackard 5890 series II) with a 1-ll sample; for each sample, four successive measurements were performed and averaged.
was sampled from a catheter inserted in an antecubital vein, during the course of every dialysate. The blood samples were centrifuged, and plasma was separated and frozen until assayed. Plasma lactate and glucose were measured in duplicate using a YSI analyzer. Reference method for in vivo probe calibration
Recalculation of the in vitro data using the slope method We reanalyzed the data obtained during the first in vitro experiment to calibrate the probes using the slope method. From the couples [LossEth; natural logarithm of RecLac (LnRecLac)] we calculated the relations linking LossEth with RecLac for a particular probe. In order to simplify the reading of the relationship between ethanol and lactate for further calculations of lactate concentration, we chose to transform the initial exponential relationship into a linear relationship by log transformation. The relationships between probe recoveries were calculated by least square regression analysis for each probe of the calibration set. Then, using the 18 dialysates of the validation set, we calculated RecLac from LossEth and the previously determined relationships. This resulted in a new estimation of the concentration of lactate and glucose from any dialysate and finally we compared these calculated concentrations to the true bath concentrations. The same procedure was applied to the relationships for RecGlu and LossEth.
Application of the slope method in human muscle Subjects and procedures of probes installation We performed microdialysis in the vastus lateralis of eight healthy young subjects [age 22.6 (2.3) years, height 180.8 (4.7) cm, body mass 70.4 (8.0) kg]. The subjects gave their informed written consent to participate in this protocol, which was approved by the local ethics committee (DGS 2001/0158). Two microdialysis probes were inserted longitudinally into the vastus lateralis muscle of the subject’s right limb. The distance between the entrance and exit sites of the probe on the skin was 7 cm; each probe was separated laterally from the other by 1.5 cm. The skin and subcutaneous tissue were anaesthetized with a local injection of 1 ml xylocaine (1%). The catheters were inserted into the muscle using an 18-gauge cannula (1.2·90 mm, Becton Dickinson) parallel to the muscle fibre orientation. The subjects rested in a semi-recumbent position during the equilibration period (first 60 min) following probe insertion and during the calibration period (70 min). During the equilibration period the probes were perfused at a flow rate of 5 ll min)1 with our standard solution (0.9% NaCl plus 5 mM ethanol) without dialysate collection. Blood
For in vivo studies in resting conditions, muscle interstitial lactate concentration ([Lac]i) estimated with the slope method was compared with that calculated with a reference calibration procedure (NNF method). However to obtain such a reference value, the original NNF method was modified in such a way that the perfusion rate was changed at each lactate concentration. This allowed us to obtain the muscle interstitial lactate concentration at rest as well as the different LossEth–RecLac couples, which were necessary to characterize the probes using the slope method. Since recovery changed at each concentration perfused, we obtained for lactate a curvilinear relationship between the lactate concentration in the perfusate ([Lac]P) and that in dialysate)perfusate ([Lac]D)[Lac]P). Then, we calculated the polynomial equation that fitted the values ([Lac]2 D)[Lac]P=c+b[Lac]P+a[Lac]P ) (Fig. 2A), where a and b are the coefficients and c is the intercept of the regression. When D–P=0, the corresponding P value reflected the [Lac]i in the muscle, which could thus be calculated as follows: [Lac]i=()b+(b)4ac))/()2a). In vivo, the modifed NNF (mNNF) gave the actual muscle concentration of lactate and it allowed us to also calculate RecLac corresponding to each perfusion flow rate, determined from the equation: ([Lac]D)[Lac]P)/ ([Lac]i)[Lac]P). The different RecLac values were then used to determine the LossEth–RecLac relationship. The validity of the mNNF method had been preliminary confirmed in vitro, showing a minimal error in the estimation of bath lactate concentration [)0.92 (2.49)%, n=24].
Isometric transcutaneous neuromuscular electrical stimulation (NMES) protocol In three subjects, microdialysis was performed during electrical stimulation exercise. At rest, [Lac]i values estimated with the slope method were compared with values obtained from the mNNF method performed simultaneously to the probe calibration step of the slope method. We perfused five combinations of lactate concentrations (standard solution plus 1–4 mM and 0 mM lactate) and flow rates (4–6 ll min)1 and 3 ll min)1). We used the couples (LossEth; LnRecLac) obtained during the mNNF calibration procedure (Fig. 2A) to calculate the corresponding relationships. Then, from the measured LossEth in each dialysate, we calculated RecLac (Fig. 2B) and then [Lac]i from the dialysate concentration.
315
Microdialysis continued during a 45-min period of light NMES applied to the vastus lateralis by two skin electrodes. A Compex stimulator (Compex-Sport, Switzerland) yielded a constant current with the form of rectangular biphasic pulse (0.4 ms) at a frequency of 14 Hz and an intensity of 30 mA (time-on 8 s, time-off 2 s). Our standard solution was perfused at a flow rate of 4 ll min)1 and a dialysate was collected over 12 min every of 15 min of NMES exercise. Dynamic exercise protocol In five additional subjects, microdialysis was performed in the vastus lateralis muscle at rest and during a 45-min period of dynamic exercise. The subjects performed knee extension of the right thigh on a Krogh ergometer (Andersen et al. 1985). The frequency of cycling was 60 min)1 and, after a 5 min warm-up period, the power output was set at 60% of the predetermined maximal sustained workload. At the initial resting phase, muscle microdialysis was performed as described above. For the resting calibration phase we used five different combinations of lactate concentrations perfused (standard solution plus 0.3– 5 mM and 0 mM lactate) and flow rates (3.5–6 ll min)1 and 3 ll min)1). After the calibration step, our standard solution was perfused at a rate of 5 ll min)1, and a dialysate was collected over 12 min of every 15 min. Statistics Lactate concentrations obtained with mNNF and slope methods were compared using paired Student’s t-tests. For both exercise protocols, the changes in lactate and glucose concentrations or LossEth from rest to exercise were analyzed by ANOVA for repeated measurements and by Scheffe´’s test when appropriate. P
