Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Microdialysis is a well established technique for continuous sampling of small, water-soluble molecules within the extracellular fluid space in vivo. It was used initially for the recovery of neurotransmitters from the brains of experimental animals (Bito et al., 1966). Since then it has been adapted for use in many other tissues including skin (Petersen et al., 1994; Andersson et al., 1995; Krogstad et al., 1996; Kellogg et al., 1998; Clough, 1999), adipose tissue (Lindberger et al., 2001), muscle (Tegeder et al., 1999; Newman et al., 2001) and the gastrointestinal tract (Iversen et al., 1997). It has been used to assess the transcutaneous delivery of xenobiotics and their pharmacokinetic disposition, and to explore the efficacy of percutaneous delivery devices such as iontophoresis (Stagni et al., 1999). Microdialysis has also provided information about the physiology of the tissue space including the release of endogenous vasoactive molecules such as histamine (Petersen et al., 1994), nitric oxide (Clough, 1999), eicosanoids (Rhodes et al., 2001), and neuropeptides (Schmelz et al., 1997) under basal conditions and following dermal provocation. Microdialysis has advantages over other sampling techniques in that it is minimally invasive and is well tolerated by human volunteers.

The principle of passive diffusion that determines the process of dialysis is similar to that governing the exchange of small hydrophilic solutes across the microvascular wall. It can be described by the Fick equation, J = DA dC/dx, where the solute flux across the membrane, J, is proportional to the diffusion coefficient of the solute, D, the area of diffusion, A, and the concentration gradient dC/dx. The ability to dialyze a substance (i.e., recover it from or deliver it to the tissue) will therefore partly be determined by the physicochemical properties of the solute, including its size and charge, and the nature of the dialysis membrane, its composition, pore area, and molecular mass cutoff. The solute concentration gradient across the membrane will be in part determined by the rate of clearance of the molecule from the tissue space. This will depend on its rate of removal in the dialysate (i.e., probe perfusion rate) and on tissue-related, physiological factors such as local blood flow and metabolism.

There have been several attempts to model the process by which a solute is recovered by dialysis and, consequently, to relate the concentration of the solute in the dialysate to that in the tissue space (Benveniste and Huttemeier, 1990; Kehr, 1993; Bungay et al., 2001). However, modeling the impact of physiological factors such as local tissue blood flow on this process has proved more difficult (Bungay et al., 1990). Our aim in this study was to investigate the impact of changes in local blood flow on the recovery of a small, diffusible molecule from the extravascular tissue space of the skin by microdialysis in vivo and to incorporate these parameters into a simple model of microdialysis. We have used sodium fluorescein as a small (mol.wt. = 376), water-soluble molecule that is readily distributed within the extravascular space. Because of its low toxicity and lack of known pharmacological effects, it has been used safely in humans in vivo, including in studies performed in the skin (Jager et al., 1997; Stagni et al., 1999). Sodium fluorescein also has the advantage that it can be reliably assayed at very low concentrations, and thus a rapid temporal analysis of changes in the local concentration of the solute is possible using relatively small volumes of dialysate collected over brief periods of time.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Subjects. Five healthy, nonsmoking volunteers (three men, two women; average age 24 ± 1 years) were recruited into the study. Volunteers with dermatologic problems, allergic skin disease, or cardiovascular disease, or those taking prescribed medication were excluded. Subjects were asked to refrain from alcohol and caffeine-containing beverages for 12 h before entering the study. All were acclimatized to 22-23°C for 30 min before the start of the study, which was performed with the subject lying supine, with the arm at heart level. The study was performed according to the declaration of Helsinki and approved by the local research ethics committee (LREC 060/98). Informed, signed consent was obtained from all subjects.

Microdialysis. Linear flow microdialysis probes (Focus 90H Hemophan Hollow Fiber Dialyzer; National Medical Care, Rockleigh, NJ) with a 5-kDa molecular mass cutoff and external diameter of 0.22 mm were inserted into the skin of the volar forearm under topical lignocaine anesthetic (EMLA; Astra, Kings Langley, UK) using two 30-mm-long 23-gauge hypodermic guide needles inserted as close to one another as possible to run for 20 mm, just below the skin surface, as described previously (Clough, 1999). The probes were threaded through the guide needles, which were then removed, leaving the probes lying within the mid-dermis. Three pairs of probes were inserted, each pair separated by at least 30 mm. The average depth of each probe and the separation between each pair of probes, assessed at the end of each experiment at three sites along the length of the probes (midpoint and 5 mm from entry and exit points) using two-dimensional dermal ultrasound (Dermascan; Cortex Technology, Hadsund, Denmark), were found to be (mean ± S.E.M.) 0.58 ± 0.08 mm (n = 30) and 1.26 ± 0.12 mm (n = 15), respectively (Fig. 1).

View larger version (169K): [in this window] [in a new window]   Fig. 1.   Ultrasound image of the skin showing the position of two dialysis probes inserted to lie in the mid dermis at a depth of approximately 0.6 mm.

1

Assay of Sodium Fluorescein. The concentration of fluorescein in the perfusate and dialysate samples was assayed using a fluorescence plate reader (Cytofluor 4000 Microplate Fluorescence Reader; Biosearch Technologies Inc., Novato, CA) at an excitation wavelength of 485/20 nm and emission of 530/25 nm. Samples were diluted with PBS up to 200 µl as appropriate, and fluorescein was quantified in 100-µl aliquots using a calibration curve constructed for each experiment over the range 5 to 1000 ng · ml¹.

Manipulation of Local Skin Blood Flow. To investigate the effects of local blood flow on the delivery of fluorescein to the tissue and recovery from it by microdialysis, blood flow in the vicinity of the probes was manipulated using either the vasoconstrictor noradrenaline (NA) or the vasodilator glyceryl trinitrate (GTN). NA (0.005 mg · ml¹ in PBS) (Levophed; Abbott Laboratories, North Chicago, IL) was added to the probe perfusate of one of the probes (retrieval probe) in a randomly selected pair, and perfusion continued for 30 min before the start of perfusion of the second (delivery) probe with the fluorescein-containing perfusate. NA perfusion was maintained throughout the experiment. GTN, which has been shown to cause local vascular damage leading to hemostasis if delivered by microdialysis (Boutsiouki et al., 2001), was delivered to a second site using a patch (Transderm-Nitro 5; Novartis Pharmaceuticals UK Ltd., Surrey, UK) applied for 30 min before the start of fluorescein-probe perfusion. Perfusion of one of the probes beneath the GTN patch (retrieval probe) with PBS was initiated on application of the patch 30 min before the start of perfusion of the second (delivery) probe with the fluorescein-containing perfusate. The third pair of probes was used as control, with one of the pair (retrieval probe) perfused with PBS for 30 min before the introduction of fluorescein to the second probe. Skin blood flow was measured at all sites before and at the end of the 30-min perfusion period with NA or PBS, or in the case of GTN application, after removal of the patch, using scanning laser Doppler imaging (Clough, 1999).

Evaluation of in Vitro Dialysis Probe Efficiency for Sodium Fluorescein. The dialysis efficiency for the 5-kDa probes for fluorescein was estimated as described previously (Clough, 1999). Essentially, individual probes were immersed in a 20-ml bath containing PBS and perfused with 1 mg · ml¹ fluorescein in PBS at a rate of 5 µl · min¹ for 120 min. The relative loss of fluorescein from the probe was calculated from the relationship (C_out  C_in)/(C_bath  C_in), where C_in and C_out are the concentration of fluorescein in the perfusate and dialysate collected at 120 min, respectively, and C_bath is the final bath concentration. The in vitro fractional recovery of fluorescein was 0.28 ± 0.04 (n = 6) at 25°C.

Statistical Analysis. Results from the five volunteers are expressed as mean ± S.E.M. with all subjects acting as their own control. Statistical comparisons have been made using a Student's t test for paired data, ANOVA, or a nonparametric Kruskal-Wallis test where appropriate. A value of p < 0.05 has been taken as significant.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Manipulation of Skin Blood Flow. The effects of NA on local blood flow were visible within 15 min of the start of perfusion as an area of blanching extending up to 2.5 mm either side of the probe and encompassing the second probe in the pair. Scanning laser Doppler imaging of the site showed NA to have reduced mean blood flux over the two probes by ~50%, from an initial value of 230 ± 50 PU (perfusion units) to one of 124 ± 29 PU, 30 min after the start of NA perfusion (n = 6) (Table 1). Blood flux did not differ significantly from this value for the duration of the experiment (i.e., >120 min). The value of blood flux recorded at the NA-perfused sites was similar to that measured as "biological zero" during arterial occlusion (200 mm Hg, 4 min; P. Boutsiouki and G. F. Clough, unpublished observation). Application of the GTN patch resulted in an area of erythema which developed over the two probes within 30 min and extended as far as the patch margins (~20 × 25 mm) but not beyond them. Mean blood flux within this area at 30 min, measured after brief removal of the patch, was 433 ± 54 PU (n = 6). It showed no further increase for up to 120 min of patch application. The GTN-induced increase in blood flux was approximately 75% of that induced by thermal warming to 43°C. At the control sites there was a small but not significant reduction in mean blood flux in the area above the probes during the initial 30-min PBS perfusion period to a value of 200 ± 70 PU (n = 6) immediately before perfusion of the second probe with sodium fluorescein. No further changes in blood flux were observed during dialysis of sodium fluorescein.

Delivery of Sodium Fluorescein to the Tissue. At all sites there was an initial rapid loss of fluorescein from the fluorescein-containing delivery probe over the first 10 min of perfusion. At the control sites, loss reached a steady state of 750 ± 85 ng · min¹, 20 to 40 min after the start of perfusion (Fig. 2). Neither the initial rate of loss of fluorescein nor the steady-state loss was significantly affected by changes in local blood flow at the GTN-treated sites (980 ± 245 ng · min¹) or at NA-treated sites (1160 ± 280 ng · min¹) (Fig. 2). The total amount of fluorescein delivered to the tissue over the 120 min of the experiment calculated from these losses was 0.11 ± 0.02 mg, 0.10 ± 0.03 mg, and 0.09 ± 0.025 mg for PBS, GTN, and NA, respectively. The in vivo extractions for sodium fluorescein calculated from the initial rate of loss of fluorescein from the delivery probes, assuming C = 0, were 38 ± 4, 33 ± 4, and 37 ± 4% for PBS, GTN, and NA, respectively (n.s., n = 5), which were not significantly different from that estimated in vitro (28 ± 4%).

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Effects of blood flow on the loss of sodium fluorescein from a 5-kDa, 20-mm linear flow microdialysis probe in the skin of healthy human volunteers under conditions of normal blood flow, vasodilatation with glyceryl trinitrate (GTN patch), or vasoconstriction with noradrenaline (NA, 0.005 mg · ml1 in PBS). Probes were perfused with sodium fluorescein (~1 mg · ml1 in PBS) at a rate of 5 µl · min1. Loss from the probe perfusate is expressed relative to the concentration of fluorescein in the perfusate ([NAF]in) Data are mean ± S.E.M., n = 5.

Recovery of Sodium Fluorescein from the Tissue. The rate of retrieval of fluorescein by the second probe appeared to be slower than that of its loss from the delivery probe (Fig. 3). The concentration of fluorescein in the retrieval probe rose very slowly over the first 20 to 30 min of infusion. Thereafter it rose more rapidly, with an exponential approach to a level at which it remained constant for the remainder of the experiment. With PBS infusion, the steady state was reached between 80 and 90 min after the start of infusion, but both the time taken to reach a steady state and the steady-state concentration in the retrieval probe were influenced by the local blood flow. At vasoconstricted sites, the steady-state concentration in the retrieval probe was 2.5 times greater than at control sites (4.0 ± 0.7 and 1.8 ± 0.7 µg · ml¹, respectively) (p < 0.001, repeated measures ANOVA), and it was not reached until 100 to 120 min after the start of infusion. The total amount of fluorescein recovered over the 120 min in the presence of NA was 1.25 ± 0.24 µg compared with 0.60 ± 0.27 µg at control sites (p < 0.001, ANOVA). Local vasodilatation reduced the recovery of fluorescein by ~50% to give a steady-state concentration of fluorescein in the dialysate at 40 to 50 min after the start of perfusion of 0.9 ± 0.3 µg · ml¹ (p = 0.05). The total amount of fluorescein retrieved at the GTN-treated sites (0.38 ± 0.12 µg) was significantly less than that at the NA-perfused sites (p < 0.001) but did not differ significantly from that retrieved at the control sites.

View larger version (19K): [in this window] [in a new window]   Fig. 3.   Effects of blood flow of the retrieval of sodium fluorescein from the skin of healthy human volunteers under conditions of normal blood flow, vasodilatation with glyceryl trinitrate (GTN patch), or vasoconstriction with noradrenaline (NA, 0.005 mg · ml1 in PBS). Probes were perfused with PBS at a rate of 5 µl · min1. Data are mean ± S.E.M., n = 5.

Evaluation of tissue concentration surrounding the dialysis probes. If we assume that the clearance of a substance from the tissue by microdialysis is equivalent to the clearance by a single large capillary, it may be described by the Renkin equation:

(1)

where PS = permeability-surface area product for the dialysis tube and F is the perfusion rate = 5 µl min¹.

(2)

1

1

1

1

1

1

1

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Microdialysis is the sampling method of choice to measure the concentration of small water-soluble molecules at discreet locations within the skin, in vivo. In general, it has proved an efficient and reliable tool. However, there is increasing awareness that recovery of solute from the tissue space by dialysis is subject to many variables, including mass transport processes between the blood and tissues (Bungay et al., 1990; Singh and Roberts, 1993; Borg et al., 1999; Sun et al., 2001). Indeed, recent studies in our own laboratory in which we used microdialysis to assess the distribution and clearance of an organophosphorus compound in human skin in vivo showed that local vasodilatation, caused by the compound itself, had a significant impact on dialysis recovery from the tissue space and, by implication, on the local and systemic distribution of the compound (Boutsiouki et al., 2001). We attempted to investigate this further in the present study by focusing on the influence of local blood flow on dialysis probe extraction through the correlation of recovery of sodium fluorescein with altered blood flow in human skin in vivo.

We anticipated that local vasodilatation would be associated with an increase in both the initial rate of loss and the steady-state loss of fluorescein from the delivery probe as a result of an increase in the volume of distribution of fluorescein and the maintenance of a larger concentration gradient across the dialysis membrane. However, we detected no significant difference in the concentration of fluorescein in the outflow from the delivery probes at any site. It is possible that changes in the rates of delivery of sodium fluorescein to the tissues did occur with changes in blood flow but that they were too small to detect with our fluorescence assay.

We did find, however, that manipulation of local blood flow influenced the recovery of sodium fluorescein and that a local vasoconstriction was associated with an increased extraction of fluorescein by the retrieval probe. We propose that this is due to a reduction in clearance of tracer from the interstitial pool by the blood and a subsequent increase in the interstitial concentration of solute in the vicinity of the retrieval probe. GTN-induced vasodilatation and the subsequent increased removal of tracer from the pool was associated with a lower recovery. It seems possible that a decrease in blood flow increased the volume of the pool of distribution of fluorescein around the delivery probe in which steady-state concentrations were reached in addition to increasing the steady-state concentration itself. Similarly, when blood flow was increased by GTN, more sodium fluorescein was cleared from the immediate vicinity of the delivery probe and the size of the pool, as well as the steady concentration, were reduced. These reciprocal changes of pool volume with blood flow may account for the differences in the time it took to reach a steady state that varied inversely with blood flow and in parallel with C_T. The rate of attainment of a steady state might be expected to be directly proportional to the sum of clearances into and out of the tissues via the blood flow and the dialysis probes and to be inversely proportional to the volume of the steady state pool. If the volumes of these pools between the delivery and recovery probes were in the range of 0.1 to 0.2 ml, the predicted rates of approach to a steady state would be of a similar order of magnitude to those observed.

Further analysis of the data suggests that the apparent drop in interstitial concentration (C_T) between the delivery probe (~400 µg · ml¹) and the recovery probe (3.6, 6 0.4, and 14.5 µg · ml¹ for GTN, PBS, and NA, respectively) appears to be very large. The binding of sodium fluorescein to proteins in the tissues may possibly contribute this large fall in C_T. Whereas the binding of sodium fluorescein to plasma proteins has been reported by Hodge and Dollery (1964), more recent work by Adamson et al. (1994) and Fu et al. (1998) suggests that the binding of freshly made up sodium fluorescein to tissue proteins is negligible. Alternatively, the gradient of concentration between the probes may be exaggerated by the clearance of solute from the intervening tissue by the microcirculation, even in the presence of NA.

To gain some insight into how blood flow influenced our findings, let us consider that the solute diffuses laterally from the delivery probe into the tissues where it is also carried away by the microcirculation. We can formulate the rate of change of solute concentration, dC/dt, in a very thin slice of tissue of thickness, dx, in terms of the solute diffusion coefficient in the tissue (D) and the clearance of solute by the blood (qE), where q is the blood flow per unit volume of tissue and E, the extraction. Thus, the rate of change of concentration equals the difference between the rate of diffusion of solute into and out of the tissue slice minus the clearance of solute from the slice into the circulating blood.

Thus:

(3)

In the steady state, dC/dt = 0, and therefore:

(4)

The general solution to eq. 4 is C = Ae^kx + Be^kx, and A_e^kx is one solution. The particular solution which fits the boundary condition, C = C_T(D) when x = 0, is:
which may be rearranged to give an expression for qE/D:

(5)

If C(x) is taken to be the value of C_T immediately outside the recovery probe and C_T(D) is the value immediately outside the delivery probe, their values can be calculated using eq. 2. Since x, the distance through the tissue between the delivery and the retrieval probes, is equal to 0.1 cm, eq. 5 can then be used to estimate qE/D under control conditions (PBS), after vasodilatation induced by GTN and after vasoconstriction induced by NA. These estimates are given in Table 1 along with the corresponding laser Doppler estimates of blood flow. Also shown in Table 1 are the ratios of the clearances and the blood flows.

Because the term on the right-hand side of eq. 5 is squared, the estimates of qE/D are sensitive to small errors in x. Thus, if x = 0.11 cm rather than 0.1 cm, qE/D during PBS infusion would be 1410 rather than 1672. Similarly if x = 0.09 cm instead of 0.1 cm, qE/D would be 2106. Despite these uncertainties, it seems we may conclude that although clearance increases and decreases with blood flow, the changes are not linear but approximate to a relation of the form:

(6)

where F = laser Doppler value for blood flow. This is shown in Fig. 4. Assuming that D is unaffected by blood flow and does not vary significantly in the tissue between the delivery and the recovery probes, eq. 6 is just the type of relation between clearance and blood flow that we might expect when F is varied and permeability-surface area product (PS) for the exchange vessels remains constant. This agreement between the theoretical relationship and our paired estimates of qE and F is pleasing, particularly as it would be reasonable to expect that PS might fall with vasoconstriction (NA) and rise with vasodilatation (GTN). Evidence for small changes of PS with NA and GTN is suggested by an analysis of the ratios of qE/D and F for the control and vasoconstricted (NA) states and for the control and vasodilated (GTN) conditions. These calculated changes in microvascular PS, however, are small and could arise from variations of the mean values of qE/D and F of less than 5% (giving rise to larger changes in their ratios), i.e., well within the error of their estimation. It is therefore reasonable to conclude that the changes in PS with GTN and NA are relatively small in human skin, and the clearances of sodium fluorescein by these capillaries and venules follow the changes in blood flow in the manner described by Renkin (1959).

View larger version (12K): [in this window] [in a new window]   Fig. 4.   Relation between clearance of tracer from the tissue space between the delivery and the recovery probes and blood flow estimated using scanning laser Doppler imaging (see Table 1).

Extraction of [¹⁴C]ethanol and ³H₂O from microdialysis probe perfusate has been used recently to measure total blood flow in rat muscle and to assess its distribution between nutritive and non-nutritive microvascular beds (Newman et al., 2001). However, to our knowledge, this is the first time that microdialysis probes have been used in humans in vivo to both deliver and recover solute from the cutaneous tissue space in an investigation of the impact of blood flow on the distribution of solute within the interstitial space. Our findings are consistent with microdialysis theory and confirm that a viable microvasculature can have a significant impact on the dialysis extraction of small, freely diffusible solutes. They demonstrate that clearance of solute from the extracellular tissue space by the blood may increase more than 4-fold under the conditions of reduced vascular flow used in our experiments in vivo. Furthermore, they also provide evidence that it is a change in flow rather than the nature of the exchange surface (PS) that is most influential in the clearance of solute by the superficial microvascular bed in the skin. The changes in skin blood flow that we induced experimentally are similar in magnitude to those seen under normal physiological conditions. Thus, our findings will have important implications for the design and interpretation of studies in which microdialysis is used to assess the pharmacodynamic profile of xenobiotics in vivo, where local tissue blood flow may be subject to change. They may also have implications for the optimization of transcutaneous drug delivery through manipulation of local blood flux to facilitate (vasodilatation) or limit (vasoconstriction) distribution.