Clinical Chemistry 43: 369-378, 1997;
(Clinical Chemistry. 1997;43:369-378.)
© 1997 American Association for Clinical Chemistry, Inc.
Fluid elements—a concept for automation of diagnostic tests
Henk Van Damme,
Thea Van Velthoven,
Erik Kaelen and
Eduard Pelssersa
Organon Teknika, Boseind 15, 5280 AB Boxtel, The Netherlands.
a Author for correspondence. Fax +31 411 654427; e-mail epelssers{at}am.otbc01.umc.akzonobel.nl
 |
Abstract
|
|---|
Constructs consisting of a channel, a membrane, and an absorber are
designed for autonomously carrying out various liquid-handling
functions of analytical tests. These so-called fluid elements can be
used to set up various circuits for conducting several kinds of
analytical tests. To demonstrate the feasibility of this concept, we
constructed such a circuit and used it to perform, with two handling
steps, an ELISA of hepatitis B surface antigen. The detection limit of
the assay was comparable with those of state-of-the-art ELISAs for
screening blood, and results could be obtained within a total test time
of 20 min. We anticipate that this concept of automation may also serve
as a basis for new, highly simplified immunoanalyzers.
Key Words: indexing terms: ELISA • hepatitis B surface antigen • immunoanalyzers
|
Introduction
|
|---|
Numerous assays have been developed for qualitative and
quantitative detection of various analytes used in the biochemical
diagnosis of human and animal disorders. Assay developers are
continually trying to make these assays easier to perform and thus more
available for use by nontechnical personnel in a wide variety of
environments, e.g., doctors' offices, clinics, patients' homes,
crisis centers, emergency rooms, ambulances, blood banks, and hospital
laboratories. Toward achieving these improvements, various different
concepts of automation have been applied to diagnostic tests. For
example, continuous-flow analyzers (1)(2) have
been used as clinical chemistry analyzers but are not suitable for
immunological tests because of those tests' stringent requirements
with respect to carryover. Especially for immunologically based tests,
robotics is used in instrumental analyzers that process separate
cuvettes.
Often, the physical procedural steps previously carried out by
technicians are copied by robotics; for single tests, however, robotic
automation is obviously unsuitable. Consequently, different concepts of
automation have been developed. For example, in a dipstick pregnancy
test (Predictor; Chefaro International, Rotterdam, The
Netherlands), a porous carrier containing all reagents is wetted
by urine, and the analytical reactions are conducted during the
subsequent transport of the urine by the capillary action of the porous
carrier; particles incorporated in the carrier are used to create a
physically detectable signal. Performance of these tests is analogous
to the action of a continuous-flow analyzer: movement of liquid through
a defined track progressing in one direction.
Here we introduce a new concept for transporting liquid in a timely
manner in various directions; the resulting liquid flow is more
complicated than that achieved in a dipstick test. The concept is based
on so-called fluid elements, constructs that can be used to conduct
various well-defined liquid-handling functions (i.e., processes
typically taking place during execution of an analytical test).
Connecting these elements in various ways yields different circuits,
each capable of carrying out different types of analytical tests.
To demonstrate the feasibility of the concept, we constructed a circuit
of fluid elements to conduct an ELISA that has a relatively complicated
test protocol. In the circuit of elements presented here, the main
parts of the transport functions of the assay are carried out
autonomously by the fluid elements. In contrast, ELISAs in
microtiter plate format require the various handling steps to be
carried out manually (e.g., pipetting liquids and moving the plates to
the incubator, the washer, and the reader). An important advantage of
the manual test format is its design to have no substantial increase in
the number of handling steps with respect to a dipstick test but to be
more sensitive through the use of an enzyme label. The further
development of fluid elements may allow a circuit of elements to be
used as a disposable "cuvette" in an immunoanalyzer and thereby
largely obviate the need for robotics.
|
Theory
|
|---|
liquid transport in the fluid element
In principle, a fluid element consists of a channel bordered
partially with a semipermeable membrane, with an absorbent material
(absorber) placed on the other side, as schematically depicted in Fig. 1
. The element has three gates, each of which can be used to
connect it to a circuit of elements. Beside the entrance and the exit
of the channel, a third gate is formed by an air vent coupled directly
to the absorber area.
The liquid flux of the liquid entering the channel
(Q1) is governed by the (applied) hydrostatic pressure
(
gz, where is the liquid density, g the
gravitational acceleration, and z the height of the liquid
column) and the capillary pressure of the channel such that:
 | (1) |
The term 2
cos
adv/D is the
capillary pressure difference at the liquid front according to the
Laplace equation for a slit geometry (where the depth D is
much smaller than the width W). The surface tension of the
liquid is represented by , and adv is the advancing
contact angle at the liquid front. The term (12
µh(t)/D3W) is
the hydrodynamic resistance of a slit according to the Poiseuille
equation, with h(t) the distance between the
entrance and the position of the liquid front at time t; µ
is the viscosity of the liquid; and Rcir is the
hydrodynamic resistance of the fluid circuit in front of the element.
In the first stage, when the liquid has not yet reached the membrane,
the contact angle will adapt to the value of the advancing contact
angle adv. In the second stage, the membrane is
wetted and subsequently so is the absorber. During this transitional
process, no substantial liquid flow into the absorber is developed and
the liquid front continues to progress into the channel. Only
thereafter is the third stage reached, in which a liquid flux
Q2 through the membrane into the absorber is established by
the capillary action of the absorber. The aspiration power of the
absorber can be characterized by the initial value of Q2
(see Fig. 1
caption).
If the absorber has a low aspiration power, the liquid front will
continue to progress through the channel and, consequently, more
membrane surface will be wetted. Therefore, the hydrodynamic resistance
to flow into the absorber will decrease and Q2 will
increase. If in this case Q2 matches Q1, the
liquid front will come to a halt, with 
adv. For
an absorber with medium aspiration power, the initial flux
Q2 will decrease the pressure at the meniscus of the liquid
front in the channel; consequently, Q2 will decrease and
Q1 will increase until both are matched at the same value.
An absorber with a large aspiration power will decrease the pressure at
the liquid site of the meniscus to even less than the capillary
receding pressure, and the liquid front in the channel will start to
retreat; consequently, the membrane surface available for liquid
transport into the absorber will decrease, and the hydrodynamic
resistance for liquid flow into the absorber will therefore increase.
As a result, the liquid front in the channel again will halt, and
Q1 and Q2 will be regulated to exactly the same
value, with rec. In the cases where
Q1 = Q2, the value of the fluxes during the
filling of the absorber can be calculated as:
 | (2) |
where heq is the distance between the
entrance and the position of the stationary liquid front. The actual
aspiration power of an absorber is determined by several factors: the
capillary pressures in the absorber, the flow resistance of the
absorber (depending on porosity and shape), and the flow resistance of
the membrane (depending on the pore size, pore density, and surface
area used for transmembrane flow). After the absorber is saturated with
liquid, the fourth and last stage is reached and the liquid front in
the channel starts to move again toward the exit of the element with a
flux Q3 = Q1, according to Eq. 1
. The time to
fill the absorber, 
Tabs, can be calculated
by:
 | (3) |
where Vabs is the volume of the absorber
and Por is the porosity factor of the absorber (fraction of volume
available for liquid). The time to fill the channel,
Tch, can be derived from the Washburn
equation (3), as adapted for a slit geometry:
 | (4) |
where hele is the length of the channel in
the element. The time delay, T, of liquid entering the
element and exiting the element is a summation of
Tabs and Tch. The
time taken by the wetting process in the second stage is neglected
because, in the experimental setups, this time is short with respect to
the time needed to fill the absorber.
separation process in the element
One of the applications of the element is to separate species of
different sizes. The membrane separates species contained in the liquid
by size exclusion. This process is conducted by leading a liquid
sample, containing the species to be separated, into the element and
subsequently adding a wash liquid to the element. After the sample is
absorbed, residual layers of the liquid sample will still remain in the
channel. These nonremoved residual layers contain all components in the
concentrations originally present in the liquid sample, and these
components will migrate into the wash liquid only by passive diffusion.
The fraction of the components 
left in these residual layers after
a wash time of t can be correlated to the thickness of the
layer and the diffusion coefficient (Dc) of the
component under consideration (4):
 | (5) |
After the absorber is saturated, the wash liquid will change
direction from migrating into the absorber toward migrating in the
direction of the exit. From that moment on, the components diffusing
into the wash liquid will exit the element, and the concentration in
the liquid, being proportional to , will therefore decrease
exponentially with time according to Eq. 5. Both Dc
and are intrinsic parameters of the assay and can be improved only
by changing the assay conditions. In the design of the element, the
surfaces used mutually by assay mix and wash liquid should be minimized
and sufficient time should be allowed for diffusion to occur (this time
is proportional to the size of the absorber).
liquid transport by the element
Another application of the element is to transport liquid samples.
During liquid uptake by the absorber, the air originally present in the
absorber is displaced. When the air vent is connected to a chamber
where a liquid sample is positioned, this liquid sample can be
transported to another location by the air pressure building up between
the element and the chamber. The minimum pressure
(Pbar) to start displacing a liquid sample
depends on the pressure barrier caused by the hysteresis effect of the
capillary pressure at the advancing and receding meniscus of the liquid
sample. For a chamber with circular cross-section, this pressure
barrier can be calculated as follows:
 | (6) |
where rec is the receding contact angle and
Drec is the diameter of the chamber at the
position of the receding meniscus; adv represents the
advancing contact angle and Dadv is the diameter
at the position of the advancing meniscus; and ß represents the
inclination angle of the surface of the chamber. In many cases, a
chamber with equal cross-sectional diameters will be used, such that
Dadv = Drec and ß = 0.
However, in the case Dadv <<Drec, the pressure barrier changes into a
pressure that forces the liquid sample to migrate spontaneously. The
maximum pressure (Pmax) that can be built up by
the absorber equals the pressure at which the liquid is pressed out of
the largest pores of the absorber:
 | (7) |
where absrec is the receding
contact angle of the absorber, and Dmax is the
diameter of the largest pores of the absorber. In the case where
Pmax >> Pbar, then
Q4, the flux of the liquid sample, equals Q2.
supporting elements
Beside the fluid elements, three supporting elements are used.
These elements perform the functions of interfacing the circuit of
elements to the outside world: providing liquid sample, providing wash
liquid, providing a hydrostatic pressure, and rendering a physically
detectable signal.
|
Materials and Methods
|
|---|
materials
The membrane is a track-etched membrane of poly(ethylene
terephthalate) (Cyclopore; Whatman SA, Louvain La Neuve, Belgium),
which has well-defined orthogonal transmembrane pores; its
hydrophilicity is regulated by a proprietary method. Descriptive data
provided by the manufacturer include: average pore density, 3.6 x
107 pores/cm2; thickness, 11 µm;
transmembrane flux, 103
mL • min-1 • cm-2 at 10 psi (~69
kPa); and average pore diameter, 0.68 µm.
The glass-fiber absorber was purchased from Whatman, Maidstone, UK.
Using a porometer (Coulter, Mijdrecht, The Netherlands) and Coulter
Porofill as test liquid, we determined the pore distribution. Measured
characteristics are: average pore size, 3.8 µm; pore size range,
2.3–10.9 µm; porosity, 0.95; and thickness, 0.8 mm.
The channel is positioned in an injection-molded polycarbonate material
(Lexan R144; General Electric, Bergen op Zoom, The Netherlands). No
grease or other lubricants were used during injection-molding, and the
parts were cleaned with absolute ethanol (Merck, Darmstadt, Germany)
and dried at ambient temperature before use. Dimensions of the channel
are 2 ± 0.02 mm wide and 100 ± 5 µm deep; the channel
length is a few centimeters, depending on the specific element. The
cover part, in which the absorber-cavity is placed, is made and treated
in the same manner.
elisa reagents
The solid phase consists of 811-nm-diameter monodisperse red
polystyrene latex particles (laboratory sample AEX7 red, as prepared by
corporate research of AKZO Nobel, Arnhem, The Netherlands) to which 2.5
mg/m2 F(ab')2 fragments of a monoclonal
antibody (laboratory sample OT IgG 2D; Organon Teknika, Boxtel, The
Netherlands) against hepatitis B surface antigen (HBsAg) have been
adsorbed. The final concentration of latex in the assay is 0.2 g/L. The
conjugate consists of a sheep-anti-HBsAg polyclonal antibody (Organon
Teknika) conjugated to horseradish peroxidase (HRP) from Boehringer
Mannheim (Mannheim, Germany). The final conjugate concentration in the
test is 12.5 µg/mL. On average, 4.4 HRP molecules are bound to one
antibody molecule. Test samples were made by adding to pooled normal
human serum (a mixture of at least 10 individual sera) HBsAg secreted
from the primary liver carcinoma cell line PLC/PRF/5 produced in
protein-free hollow fiber culture (PLC-HBsAg; Organon Teknika).
The substrate compounds are stored in a matrix of nylon tortuous
membranes, pore size 0.2 µm (MSI Europe, Bergen op Zoom, The
Netherlands). This matrix is used in a supporting element, in which the
substrate reaction is carried out. The tortuous membranes are wetted by
dipping into a substrate solution, after which excess liquid is removed
by pressing the membranes between two rollers. This matrix is further
dried under preheated nitrogen (30 °C) for 20 min. Membranes are
stored under nitrogen at 4 °C in sealed aluminum sachets containing
dry silica gel. Two membranes are used per element—one impregnated
with a substrate solution containing 3,3',5,5'-tetramethylbenzidine
dihydrochloride (pro analyze; Fluka, Neu-Ulm, Germany), the second
impregnated with a substrate solution containing hydrogen peroxide.
methods
Assembly of elements (and circuits of elements).
Membranes are sealed on the channel by a stamp that is heated to
238 ± 1 °C and exerts a pressure of 200 ± 10 kPa for
3 ± 0.1 s. Contact between absorber and membrane is
maintained reproducibly by placing the absorber against the membrane
with a slight pressure. This pressure is provided by placing the
absorber in a cavity 0.7 ± 0.02 mm deep (the noncompressed
absorber is 0.8 ± 0.02 mm thick). The injection-molded parts, one
containing the cavity and the other containing the channel, are
connected by heat-sealing at certain positions or by using nuts and
bolts (Fig. 2
).
View larger version (58K):
[in this window]
[in a new window]
|
Figure 2. Left: a bound/free element (middle);
a combination of a bound/free element and an output element, called a
detection circuit (top); and an older version of the
detection circuit (bottom); right: integrated
circuits.
In the top and middle devices (left
panel), the membranes are heat-sealed onto the channel; the
plastic parts are also connected by heat seals. In the older detection
circuit (bottom device), the membrane and parts are
connected by double-sided adhesive tape but the function is identical
to that of the topmost device. In the integrated circuits in
the right panel, the membrane is heat-sealed to one of the
plastic parts, and the plastic parts are connected by nuts and bolts.
|
|
The dimensions of the experimental elements are optimized for
performing the functions to be carried out (see Figs. 3
and
4). The fluid element that provides for the separation of the
bound and unbound conjugate molecules is called the bound/free element.
The fluid element transporting the assay mix by air pressure is called
the pressure element, and the fluid element that delays the liquid
transport and thereby controls the incubation time of the assay is
called the timing element. The supporting element that accepts the
sample or assay mix is called the input element. The supporting element
that provides the substrate reaction is called the output element,
and the supporting element providing hydrostatic pressure is called the
hydrodynamic power source.
View larger version (26K):
[in this window]
[in a new window]
|
Figure 3. The bound/free element, the pressure element, the output
element, the modified bound/free element of the integrated circuit, and
the detection circuit (combination of a bound/free element and an
output element).
The modified bound/free element has separate entrances for sample
liquid (the assay mix containing the sample) and wash liquid. The
surface of the air vent of the output element is made hydrophobic to
prevent leakage of liquid but to allow air to escape during filling
with liquid. Except for the side view of the modified bound/free
element of the integrated circuit, all diagrams are at the scale
shown.
|
|
Measurement of physical parameters.
The latex recovered
from a bound/free element is defined as the percentage of solid phase
recovered in the first 40 µL of liquid exiting the element with
respect to the amount initially added. The efficiency of the separation
of bound and unbound conjugate molecules in a bound/free element is
defined as the percentage of unbound conjugate molecules that is not
recovered in the first 40 µL of the liquid exiting the element. The
values for the physical parameters were experimentally determined;
details are available upon request.
Test of assay reagents and procedure.
We tested the
reagents independently of the fluid elements as follows. We mixed 10
µL of latex solid phase (8.5 g/L) with 90 µL of normal human serum
containing 10 mL/L normal sheep serum in a 1.5-mL Eppendorf micro test
tube (Eppendorf, Hamburg, Germany). After adding 100 µL of conjugate
solution and mixing, we added 200 µL of test sample, mixed, and
incubated the assay mix solution for 5 min at ambient temperature. The
reaction was stopped by adding 4 µL of sheep-anti-HBsAg
polyclonal antibody (Organon Teknika), 30 g/L. Using this much
excess antibody ensures that all free binding sites on the HBsAg will
be occupied by IgGs and that no more conjugate molecules can bind to
the antigen during the rest of the procedure. Subsequently, we washed
the latex solid phase by mixing it with 1 mL of wash buffer [50 mmol/L
glycine, pH 9 (Janssen Chimica, Geel, Belgium), and 1 g/L bovine serum
albumin HR (Organon Teknika)]. We then separated the particles by
centrifugation (15 000g, 5 min) and redispersed the pellet
in 1 mL of wash buffer by vortex-mixing. After repeating this wash
procedure for a total of four times, we resuspended the solid phase in
80 µL of water, pipetted 7.5 µL of this dispersion into a
microtiter well, and added 100 µL of substrate solution from the
Hepanostika Uni-form II 1.0 kit for HBsAg (Organon Teknika) to the
well. After incubation for 5 min, the substrate reaction was stopped by
the addition of 100 µL 1 mol/L H2SO4 (Baker,
Deventer, The Netherlands). To measure the absorption of the product at
450 nm, we used a Model 510 microtiter plate reader from Organon
Teknika, Turnhout, Belgium.
Assay performed with circuits of fluid elements.
The
integrated circuit performs an assay as follows (Fig. 5
): The assay mix, prepared as described above, is incubated for
3 min, after which 20 µL of it is injected into the input element.
Immediately thereafter, water (for laboratory use, ISO 3696:1987) is
added to the hydrodynamic power source. Subsequently, the mixture is
incubated for 2 min in the incubation chamber of the input element,
this remaining incubation time being controlled by the combined
timing/pressure element. The mixture is then transported to the
bound/free element by the combined action of the pressure element and
the hydrodynamic power source. After the bound/free separation process
is completed, the latex solid phase is carried autonomously by the
liquid flow to the output element. Here, the substrate components
dissolve and, in the presence of the bound conjugate molecules,
participate in the enzyme–substrate reaction; the color formed is
observed by eye. The timing carried out by the circuit can easily be
changed to 5 min, in which case the incubation outside of the circuit
is not needed.
View larger version (42K):
[in this window]
[in a new window]
|
Figure 5. A schematic representation of the operation of the
integrated circuit, consisting of one combined timing/pressure element
and one modified bound/free element.
For ease of viewing, the circuit is unfolded into one plane; in the
actual device, the channels and elements are located in two levels. In
addition to the fluid elements, an input element, a hydrodynamic power
source, and an output element are used. The input element consists of a
chamber into which the assay mix is deposited. Alternatively, this
chamber can contain dry reagents, which are resuspended during
injection of the sample.
|
|
The single bound/free element is operated by adding the assay mix to
the element, followed immediately by addition of the wash liquid (water
for laboratory use, ISO 3696:1987). At first, the liquid is carried
through the device by capillary action and later, after wetting, by
gravity.
|
Results
|
|---|
performance of individual fluid elements and reagents
Bound/free element.
The liquid front of the assay mix
entering the element travels ~10-15 mm in the channel beyond the
start of the membrane (typical flux in these experiments, 0.5 µL/s),
which corresponds to 0.5–1.5 s before the membrane and absorber are
wetted and the aspiration power of the absorber becomes available;
after this time, the liquid front comes to a halt. Without a latex
solid phase in the assay mix, this position is ~5 ± 1 mm from
the start of the membrane; presumably, the particles increase the
hydrodynamic resistance of the membrane by blocking a number of pores.
The latex recovery result is denoted in Table 1
. If the membrane was not made hydrophilic, latex recovery
substantially decreased. Including salt in the wash liquid also
worsened the latex recovery, as did a decrease in the amount of latex
solid phase initially applied in the assay mix.
We determined the amount of unbound conjugate in the aliquots exiting
the element and plotted the log-transformed values as a function of the
time at which the aliquots exited the element (Fig. 6
, curve A). The data points show a linear decrease as function
of time, as predicted by Eq. 5. By applying an estimated value for
Dc, the diffusion coefficient of the conjugate molecules in
serum, we could calculate the thickness of the residual layer ()
from the value of the slope of curve A. Extrapolating the data of Tyn
and Gusak (5) gave an estimated Dc of 1.8
x 10-11 • m2 • s-1 and a
calculated of 90 µm. However, the layer thickness of the residual
layer is not equal at all positions. In particular, the fact that the
channel has a cross-section of 2 x 0.1 mm makes it likely that
the residual layer at the corners is much thicker than in the middle
part of the channel. Therefore, the calculated value of the layer
thickness will be an average, depending on the size of the measured
time window. Moreover, conjugate molecules located in the membrane
pores and in the absorber can diffuse back into the channel, also
influencing the calculated value of .
Decreasing the surface area mutually used by the assay mix and the wash
liquid decreased substantially the amount of unbound conjugate
molecules finally leaving the element (Fig. 6
, curves B and C). For
comparison reasons, we include the tabulated separation efficiency of
the bound/free element in Table 1
. Track-etched membranes, having a
smooth surface, provided the best results. When we used tortuous
membranes, which have many dead spaces and a relatively rough surface,
the wash efficiency dropped by 10-fold (results not shown).
Pressure element.
Measurement of the pressure necessary
to displace a water or serum sample in a 2-mm-diameter Lexan chamber
showed it to be <0.100 kPa. The maximum pressure buildup produced by
filling the absorber with water was measured as 6.500 (±10%) kPa.
Because the pressure to drive the displacement is much greater than the
pressure barrier, one can apply Eq. 2
as a good approximation for
determining the liquid flux of the liquid sample being displaced.
The advancing contact angle of water on Lexan was found to be
85° ± 2°. The receding contact angle was difficult to
measure; therefore, we used Eq. 6 to derive the receding contact
angle: 61°. From these values we could predict the pressure barriers
(Pbar) in various geometries. In principle, Eq. 7
can be used to calculate the receding contact angle of water in the
absorber. However, during filling of the absorber with water,
components of the absorber dissolve and thereby influence the surface
tension value. The value of the pore sizes, as determined with the
porometer, can also be influenced by this effect.
Because no separation process is conducted with this element, the
membrane component is omitted. In this case, the retreat effect
(described above in Liquid transport by the element) can be
easily observed. The water progresses into the channel to a position
5 ± 1 mm past the starting point of the absorber. After this, the
absorber is wetted and the full aspiration power (unrestricted by any
membrane) is available, such that the liquid front in the channel
retreats to a position 1 ± 0.5 mm from the start of the absorber.
Timing element.
The time to fill an element was measured
in a small circuit of a hydrodynamic power source connected in series
to a combined timing/pressure element (see Fig. 4
). The hydrodynamic
power source included a vertical cylinder (where a water column
introduces a hydrostatic pressure) and a membrane restriction. The
delay time (T) for the element was measured as ~70 s.
To calculate T with the help of Eqs. 2–4
, one must know
the hydrodynamic flow resistance
(Rcir) of the circuit
in front of the element. This resistance is mainly determined by the
membrane restriction, and the resistance of the channels connecting the
elements can be neglected. The hydrodynamic resistance of the membrane
restriction can be directly derived from the transmembrane flux and the
surface area of the membrane (3.9 mm2). Furthermore, as
mentioned above, the liquid retreats during the filling of the
absorber; hence, the receding contact angle has to be used in Eq. 2
.
The values of constants and variables used in these calculations are
listed in Table 2
. Note that Eq. 2
is valid for a slit where D <<
W. In the experimental setup, D =
W/20 and the different surfaces making up the channel can
distort the meniscus. Given that one wall of the channel is the surface
of the absorber and three walls are Lexan, we used estimates of the
average advancing and receding contact angles. The resulting calculated
T was between 52 and 74 s, which is comparable
with the experimentally measured value.
View larger version (26K):
[in this window]
[in a new window]
|
Figure 4. The timing circuit is a combination of a hydrodynamic
power source and a combined timing/pressure element; it is used to test
the timing element.
|
|
The T in the liquid flow introduced by elements can be
used to control the incubation time of an ELISA performed as shown in
Fig. 5
.
Output element.
As perceived by eye, the amount
of free HRP still producing a color in the element within 5 min was 4
pg.
Reagents.
Testing the reagents independently of
the fluid elements and circuits (see Materials and Methods)
yielded a detection limit for HBsAg of 1 IU/mL. The test samples used
were calibrated against standards provided by the Paul Ehrlich
Institute, Langen, Germany. The dose–response curve is depicted in
Fig. 7
.
View larger version (9K):
[in this window]
[in a new window]
|
Figure 7. Dose–response curve of the HBsAg assay tested
independently of the fluid elements and circuits (see Materials
and Methods).
|
|
performance of fluid element circuits
The detection circuit (Fig. 3
) is a combination of a bound/free
element and an output element. The detection limit was monitored and is
tabulated in Table 1
. In this case the liquids are added in the same
way as to a single bound/free element, and color formation is observed
by eye.
The integrated circuit contains several elements and is capable of
carrying out an ELISA almost fully autonomously (Fig. 5
). The circuit
pictured uses an adapted bound/free element in which the assay mix
(containing the sample) and wash liquid enter through separate
channels, minimizing the area of mutually used surface. The resulting
separation efficiency exceeds that of a bound/free element with only
one entrance (see Table 1
). Also, the detection limit is improved,
probably because of the increased separation efficiency.
Because the liquid sample and wash liquid are not in physical contact
with each other in the pressure element, the wash liquid can still be
used in the bound/free element. Also, because the wash liquid is
released by the pressure element only after the liquid sample is
deposited in the bound/free element, no extra timing for transport of
the wash liquid has to be built in.
|
Discussion
|
|---|
The detection limit for material added to normal human serum, as
detected with the integrated circuit, was similar to that for the
Hepanostika kit for screening blood for HBsAg. The screening of
clinical samples, including sero-conversion samples, with the
integrated circuits is the subject of further study. In any event, the
number of handling steps is much smaller than in the blood screening
assay, and the total test time with the integrated circuit is fivefold
less (Table 1
). The larger surface-to-volume ratio of the dispersed
solid phase, in comparison with a microtiter well as solid phase,
probably accounts for this shorter test time with an almost equal
detection limit. Also, the disperse nature of the solid phase decreases
the diffusion distances during incubation of the assay mix.
Remarkably, the detection limit of the integrated circuit is even
better than that of the reagents tested independently of the
circuit. Possibly the separation efficiency is higher than when testing
the reagents separately. Unfortunately, separation efficiencies
>99.9988% could not be determined accurately, so no definite
conclusions can be drawn with respect to this.
The detection circuit is generic in the sense that different analytes
can be detected with the circuit. The integrated circuit can be used in
several ways. When the assay mix is added to the circuit, the circuit
is generic. When the reagents are self-contained in the chamber of the
input element, the circuit is dedicated to one type of analyte. When
different reagents for various analytes are stored in the chamber,
multianalyte testing is possible. Alternative circuits for carrying out
an ELISA are also possible, one theoretical example being shown in Fig. 8
. Additional types of analytical tests can also be carried out
but are beyond the scope of this report.
View larger version (28K):
[in this window]
[in a new window]
|
Figure 8. An alternative circuit for conducting an ELISA, based on
the use of a combined timer/pressure element and a combined bound/free
+ pressure element.
In this case, the output element is used only for storing the
impregnated plates and resuspending the substrate components. The
reaction producing visible results is executed in the combined
bound/free + pressure element.
|
|
The demonstrated feasibility of the concept makes it possible to
further develop manual tests based on fluid elements for application in
less-sophisticated laboratories or in point-of-care situations. The
small size of the assay components, the few handling steps, and simple
operation could make the test accessible to nontechnical personnel;
even the use at home by patients could be considered. Existing manual
tests cannot offer a detection limit comparable with that of current
blood screening assays in combination with a short test time and few
handling steps.
Future developments in instrument design could benefit from the
concept. The complexity of conducting an ELISA is almost fully
controlled by the circuit of fluid elements. If such a circuit of
elements could be considered as a disposable "cuvette," it could be
used in combination with an instrument of relatively simple
design—i.e., an instrument that would add the sample to the cuvettes
and would read the results by optical or electrochemical means. The
complex control software and complex mechanical robotics typical for
current immunoanalyzers could be omitted and thereby decrease the need
for servicing. The size of the instrument could decrease substantially
and perhaps even utilize a different design: parallel processing of
cuvettes. This might open a product line wherein the customer could
determine the capacity of the instrument used by simply changing the
number of slots for positioning such cuvettes in the
instrument.
In conclusion, the fluid element as an automation concept has
some potentially very beneficial features: flexible design, low
detection limit, and rapid, autonomous operation. The concept of
creating circuits by combining fluid elements makes it likely that
flexible different designs can be set up. Assay detection limits, as
determined with analyte-supplemented material, are comparable with
those for state-of-the-art blood screening assays—a necessary feature
if fluid element assays are to be used for screening blood. The
detection limit for clinical samples was not determined but is the
subject of further study. Rapid test times, especially convenient in
point-of-care testing and in an emergency situation, of 20 min can be
obtained for the integrated circuit vs 90 min for a typical
state-of-the-art blood screening assay. Moreover, because a circuit of
elements can perform a test almost fully autonomously, without moving
parts and without an external power source, one would expect a test
system using these circuits to be highly reliable. We think that the
features of the fluid element circuit concept—i.e., no moving parts
and only two interfaces to the outside world (input and output)—could
combine very well with the miniaturization technology currently used
for chip fabrication and could yield an even more exciting future
involving new diagnostic formats.
|
Acknowledgments
|
|---|
We acknowledge B. Krutzer and S. Vos for supplying the
reagents and carrying out experiments with the detection circuit; R.
Hoeben and H. van der Linden for technical support; T. Beumer for
sharing his knowledge about wash processes and for support in
conducting several of the fundamental experiments; and W. Carpay for
stimulating discussions about fundamentals and methods.
|
References
|
|---|
-
Skeggs L. An automatic method for colorimetric analysis. Am J Clin Pathol 1957;28:311-322.
-
Ruzicka J. Flow injection analysis, from test tube to integrated microconduits. Anal Chem 1983;55:1040A-1053A.
-
Hiemenz PC. Principles of colloid and surface chemistry,
2nd ed. New York: Marcel Dekker, 1986:338pp..
-
Beumer T, Stoffelen E, Smits J, Carpay W. Microplate washing: process description and improvements. J Immunol Methods 1992;154:77-87.
[Medline]
[Order article via Infotrieve]
-
Tyn MT, Gusak TW. Prediction of diffusion coefficients of proteins. Biotechnol Bioeng 1990;35:327-338.
The following articles in journals at HighWire Press have cited this article:
|
|
|
|
 
C. P. Price
The Evolution of Immunoassay as Seen Through the Journal Clinical Chemistry
Clin. Chem.,
October 1, 1998;
44(10):
2071 - 2074.
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
) and C (
) show results
for when this distance is 5 and <0.5 mm, respectively. For this study,
(B and C) the position of the liquid front was controlled by a pump
that added liquid through an extra hole at the position of the
membrane. Each line refers to a single experiment.