Spinit Clinical Chemistry Assay

Filipe da Conceicao Fernandes Thomazfilipe.thomaz@ist.utl.pt

Instituto Superior Tecnico, Lisboa, Portugal

October 2014

Abstract

Point of Care (PoC) detection equipments - medical exams that are conducted near the patient- have seen a strong development over the last few years. The development of a spectrophotometerof small dimensions, low cost of production and compatible with technologies of low scale fluidpropulsion would open a world of assays anda analysis that could be done within the same system.A first prototype was built based on an emitter and receiver LEDs to determine the concentrationof total haemoglobin from a small amount of blood (< 15µL) and compatible with the spinit R©equipment from Biosurfit. Total errors below 4% have been obtained for twenty-four venous andcapillary blood samples. A second prototype was build using the same technology allied to atotal internal reflection based technology to increase the optical path from 0.2mm to l = 7.5mm.Glycated haemoglobin (HbA1c) was tested by this prototype and a total CV of CV = 2, 4% anda total error of 3, 8% for thirty-four venous blood samples tested, with results in seven and a half minutes.

Keywords: Point of care, Glycated Haemoglobin, Clinical Chemistry, Spectrophotometry

1. Introduction

Medical exams such as blood and urine tests areof extreme importance for the doctors to establisha diagnosis to the patient. Nowadays results aregiven to the patient at least a couple of days afterthe sample collection. [1, 2] If this exams couldbe made with a minimum amount of time spentand the closest possible to the patient, wherethe sample is collected, the waiting time wouldbe minimized, as well as costs related to theillness and its treatment, less amount of samplewould be collected and the validity of the results inthe time of diagnostic would be more accurate. [1,3]

Point of care Testing is a major area of devel-opment nowadays, comprising all of the examsthat can be executed near the patients bed. [3, 4]Microfluidic is a crucial and important area ofdevelopment since it is required to handle thepropulsion of small amounts of fluid, and howto handle mixing and detecting in the microscale. [3, 5, 6]

Various companies have developed working PocTproducts. Lateral flow tests are a major classof point-of-care testing, consisting on a narrowmembrane that makes use of capillary action whenadding the sample to move the fluid without userintervention. There has been industry interest in

trying to improve their performance over the lastdecade but without significant progress. [4]

It is the goal of point-of-care testing to linka patient to a result, from sample collecting,through pre-treatment, anaylte-specific reaction,signal production, signal detection and reportingof result, and do this with the best cost-to-benefitratio possible. [4, 7, 8] Since very few diagnosticsrely only on one specific marker or test there isa need of an integrated PoC product that can byitself perform the majority of blood tests. Thiscan be very challenging as technical requirementscan vary greatly for each component or assay step,so thinking of integration and a design capable ofperforming various blood tests - disease markers,clinical chemistry and hematology - is imperative inorder to develop a successful real-world product. [4]

iStat Corp. developed a micro-sensor technologythat detected a range of blood chemistries, coagu-lation and cardiac markers. It relies on pneumaticactuation and capillary action to handle drops ofwhole blood, and performs simultaneous assays inless than 2 minutes. [4, 9]

Abaxis has taken a disc-like approach. It testssmall-molecule and protein markers for metabolic,lipid, liver and renal diseases. Focus Diagnostics

1

detect nucleic acid signatures for respiratorypathogens and both use centrifugal forces to movefluids. [4, 10,11]

Biosite uses a test strip with textured mi-crostructures to test for a range of protein andsmall-molecule markers for cardiovascular disease,drugs of abuse and waterborne parasites. Howeverit still requires a pre-treatment of centrifugation toseparate plasma from red blood cells. [4]

There are other companies that have workingand approved PocT systems, most of them actingthe fluid by pneumatic forces, but most of thosesystems were designed for one or few different testsand are of extremely complex - or even impossible- integration with other tests using the similarapproach. [4, 6]

Biosurfit is a portuguese company that uses cen-trifugal microfluidic to move fluids with the goal tocreate a single machine (spinit R©) that can be ableto perform the vast majority of blood tests withonly a small amount of unprepared blood sample(5µL) in less than 10 - 15 minutes. (Figure 1) CRPtesting is already commercial, and a blood countsystem with Leukocyte counting and Hematocritmeasurement has also been developed. [12] Thenext step is to implement in the same format anduniversal CD system a biochemical analysis forclinical chemistry parameters.

Figure 1: Spinit machine by Biosurfit.

Colorimetric detection by spectrophotometry hasbeen used in clinical chemistry applications to de-tect some biochemical markers in addition to vari-ous blood protein panels. [6] Spectrophotometry isbased on reading the amount of light that passesthrough a certain solution. Beer-Lambert Lawstates that the relation between the emitted and

detected light intensity that travels through a sam-ple of distance l is dependent on the concentrationc of the elements in the sample that absorbs in theemitted light’s wavelength,

A = − lnI

I0= ε(λ) l c (1)

being A called the absorbance and ε(λ) theabsorptivity, or extinction coefficient, that dependson the wavelength of the emitted light. I is theintensity passing through the sample and I0 is theintensity that would arrive to the detector if noabsorbing sample was in the light’s path [13]

Working with a portable device as using aCD-like format with a thickness of 1.2mm, theoptical path will be reduced drastically and hencereducing the measured absorbance. A beam oflight passing vertically through the solution in aCD format won’t have a sufficient optical pathin order to have a measurement with sufficientresolution and precision. [6, 14, 15] Various ideashave been thought and tested to increase lopt . Op-tical waveguides have been proposed to guide thelight in order to increase the optical path. [16, 17]Hollow waveguides have an intrinsic loss inverselyproportional to the cube of the waveguide’s length,which is problematic in their use for microfluidic.Using square hollow waveguides an efficiency ofη = 60% was achieved over a distance of 3.2 cm,with a detection limit (considered to be a signal-to-noise ratio S/N = 3) of 200µM fluorescein usinga 50µm path length and a simple photocell detector.

Also integrated micro-lenses have been testedto detect fluorescent species in microfluidicswith a successful detection of a 10-nM Cy5 withS/N = 21 dB. [18] There have been some attemptsto use optical fibers [19, 20] and total-internal-reflection in silicon micro mechanics [21].

All these approaches mentioned above requirehowever complex chip designs and high-accuracyalignment of the optical componentes, making ithard to be competitive and the low-cost detectionmethod that is required. [22]

Most recently it has been tested the use ofmicro-ring resonators to perform absorption spec-troscopy. [23] This ring resonator is designed sothat, at resonance, light circulates many timeswithin the ring and thus leading to a largerenhancement in the interaction between light andthe cladding liquid and increasing the optical pathseveral orders of magnitude than the circumferenceof the ring resonator itself (of about 5mm). Itsmeasurements have shown good agreement between

2

commercial spectrophotometer and the absorbanceusing the micro-ring device for a less than 0.2µL ofN-methylaniline.

MTEK has however used and tested a much sim-pler approach and successfully integrated it into alab-on-a-chip CD. The guidance of the optical beamis achieved by total internal reflexion (TIR). [14,15]When a light beam changes from a media 1 to a me-dia 2 with index of refraction n1 < n2, according toSnells Law well have a critical angle

θc = arcsinn1n2

(2)

where for angles θ > θc all the light is reflected. As-suming a refractive index of n2 = 1.5 for polymersusually used in clinical diagnosis, and being n1 = 1for air, we have from equation 2 a critical angleθc ≈ 41. Two triangular (V-shaped) prisms can beused as reflectors, positioned with an inclination of45 with the surface, one on each side of the testchamber. This way the light beam initially sentvertically is reflected by the first prism and thelight would path horizontally the whole length ofl = 10mm of the detection chamber, before beingsent back vertically to a detector by the otherprism with the same inclination. (Figure 2)

Figure 2: Representative image of the TIR method.E and D are the emitter and detector-LED and Lis the optical path the light travels.

Glucose, alcohol and total haemoglobin concen-tration were measured using this approach on adisc, with excellent results. [14, 15,24]

Light Emitting Diodes (LEDs) are nowadaysapplied in modern optoelectronics as they areextremely cheap, stable, robust, small in sizeand have a long lifetime as they are low-poweredand intensive light sources. [25–27] They covera broad spectral range from UV to NIR andhave a narrow emission spectra, allowing highselectivity in many kinds of determinations. [25]

LEDs are significantly cheaper than photodetec-tors and photodiodes and they can also be usedas light detectors when applied in reverse mode. [25]

LED-photodiodes are considerable less sensitivethan commercial photodiodes, so direct measure-ments without amplification of the photocurrent isdifficult, however it was shown that very preciseand accurate measurement of the photocurrent ispossible using a simple threshold detector and atimer circuit. [27,28] Knowing that light intensity isproportional to a discharge current on the detectorLED (ilight) and neglecting the detector-LEDproper small escape current (idark), one can relatethe detector discharge time (td) with the sampleabsorption [27], since

td =Q

ilight + idark≈ Q

ilight. (3)

where Q is a constant (accumulated charge).From equations 1 and 3, the final expression canbe written as follows [27]

ln t = ε(λ) c l + ln t0 (4)

where t is the discharge time measured by thedetector LED, for a certain ∆V (Vt − V0), whenthe light passes through the desired sample and t0the discharge time provoked by the emitter LED’sintensity without being absorbed.

As stated before, Biosurfit is a company whosegoal is to develop a machine that can perform var-ious blood assays. Biosurfit’s disc format permitschambers of only 200 µm thickness, so the opticalpath is clearly not enough to have a good signalto noise ratio in absorbance measurements. Thegoal of this project is to develop a clinical chem-istry assay based on the two technologies describedbefore, TIR and PEDD, that can robust enoughto detect clinical chemistry parameters with the re-quired performance criteria and to be integrated inthe microfluidic and disc design and constraints ofspinit R©. To study the performance of the PEDDtechnology with clinical chemistry detection withinthe disc thickness constraint a smaller prototypewill be built without the TIR technology, by mea-suring Total Haemoglobin concentration. The fi-nalized prototype with both the PEDD and TIRtechnologies will be assessed by the study and mea-surement of HbA1c using a standard immunoturdi-metric assay.

2. Total Haemoglobin ConcentrationPrior to developing a system that can allow us to in-crease the optical path in order to have a higher res-olution in absorbance measurement using a PEDD

3

system, it was developed a slightly simpler proto-type which would give us a proof of concept of ourmethod allied with a better and easier understand-ing of its behavior, knowledge that we would bene-fit a lot from when transitioning to a more complexsetup.

2.1. Implementation

An electronic board designed specifically for oursetup was used, connected and powered via an USBcable. It works with a timer circuit that countsthe amount of time that a certain receiver LED -connected to the board - takes to drop from 5 Vto a certain threshold voltage. It communicateswith the computer as a Serial RS232 port sothe software Termite 3.0 for Windows was usedfor such communication. The output time is inmicroseconds.

Testing discs were designed and manufacturedin Biosurfit. They comprised of several chambers- ten in the first version of the disc, forty-two onthe second - with a circular shape - 6mm radiusfor the first, 4.5mm for the second. This chamberdesign was cut into a a three-layered bi-adhesivetape, of t = 225µm to approach the 200µm fromspinit R© and two PMMA discs were glued to thetape, one on each side. A drill was used to opentwo holes in each of those chambers for liquid entryand air exit. The second version of the discs canbe seen in figure 3.

Figure 3: Final [Hb] disc with the three layers ofbi-adhesive tape with cuts working as chambers.

Two different LED as emitter and receiver wasused. The first PEDD comprised of a AND180HYPLED from AND Optoelectronics, with λ = 589nm,a luminous intensity of 9.75 cd and a viewing angleof 6◦ as both emitter and receiver. At this wave-length the relation between the most-absorbentand least-abosrbent Hb fractions is minimal.The second PEDD comprised of the same LEDas a receiver but a LED545-06 from Roithner,with λ = 545nm, output power of 8mW and a

half-viewing angle of 4◦ as an emitter. At thiswavelength we have a triple isobestic point forHHb, HbO2 and HbCO, while MetHb variationsare basically neglected.

RBC lysis was accomplished with a 4% saponinsolution in a 0.9% NaCl solution. A dilution of12 µL of blood with 4µL of the saponin solutionlysed the cells in a maximum of three minutes,depending on the Hb concentration. The lysedblood is inserted inside one of the chambers andplaced between a vertically aligned PEDD system.Ten measurements are instantaneously taken andan algorithm uses the average to retrieve the TotalHb concentration.

All the data was treated using Microsoft Excelfrom Microsoft Company.

2.2. Results

Non-lysed blood samples were primarily tested butthe discharge times varied drastically and withoutreproducibility with time due to RBC aggregation.(Table 1)

t (s) [Hb] = 14, 5 g/dL (µs)

60 756

180 734

240 725

300 704

360 695

720 652

Table 1: Discharge time (µs) over time (s) for undi-luted blood sample with [Hb] = 14.5, Emitter anddetector LED: 589nm.

As for the absorbance in the 589nm the max-imum discharge time was extremely far from thesaturation discharge time of the receiver LED -around 2500 µs. Beer-Lambert Law (eqn. ??)states that by increasing t0, that is, decreasing theintensity read by the receiver LED without anysample, would not only increase the discharge timesbut also increase our range. Since the distancebetween LEDs was fixed t0 could be increased bydecreasing the tension applied to the emitter. Evenwith the lowest tension required for the LED topower on (around 2.4 V ) we read discharge timesof maximum 600µs for the high concentrationsamples - quite far from saturation. The LED wasthan powered at 2.7 V , so that it’s not at the limitof the threshold voltage for the LED to power on,and yet maximizing our range.

Figures 4 and 5 show a calibration curve made

4

with eight whole blood samples with a dilu-tion of 1:1/6 (blood:saponin solution) and 1:1/3(blood:saponin solution) respectively for the 589PEDD. It can clearly be seen that diluting 1:1/6is not enough for a full lysis and results are notcoherent with each others. When started usingdilution of 1:1/3 all of the blood samples lysed inless than two minutes, and the calibration curvecan be seen as linear with a correlation factor ofR2 = 0.9769.

Figure 4: The logarithm of the discharge timesagainst [Hb] (g/dL) for eight whole blood samples.Dilution: 1:1/6

Figure 5: The logarithm of the discharge timesagainst [Hb] (g/dL) for eight whole blood samples.Dilution: 1:1/3

Using a dilution of 1:1/3 (blood:saponin solu-tion) several blood samples were analyzed. Forthe 589 nm PEDD, forty-two whole blood sampleswere analyzed, ten being fingerstick and 32 beingvenous blood, spanning from 8.5 to 17 g/dL ofTotal Haemoglobin. The regression curve relatingour results with a reference laboratory can be seenin figure 6.

Using the same dilution the 545 nm PEDD wastested. The voltage powering the emitter LEDhad to be raised to 3.7 V as in this wavelengththe absorbance is higher than in the 589 nm, andour discharge times were too close to saturation.

Figure 6: Spinit* [Hb] (g/L) against ReferenceMethod [Hb] (g/L) for a LED with λ = 589nm

Twenty-four samples were tested and the regres-sion curve relating our results with a referencelaboratory can be seen in figure 7.

Figure 7: Spinit* [Hb] (g/L) against ReferenceMethod [Hb] (g/L) for a LED with λ = 545nm

The 545 nm LED as an emitter provided awider range due to the higher absorbance. Totalanalytical error was below 4% for each sampletested. The main disadvantage of this LED is thefact that it is insensible to MetHb variations, butthis variations are extremely rare, as the normalphysiological conditions of MetHb are of 0-1%.

3. Absorption Module3.1. ImplementationThe board remained the same as in the previousprototype with a smaller change that would allowus to read higher discharge times. Also outputtime is dt = 0.542µs. For the TIR approach a disclayout with prisms was designed and manufacturedby Axxicon. They consist of six chambers ofthickness t = 200µm at the same radius, 24◦ fromeach others. This chambers have a channel thatis 7.5mm long, from now on referenced as ouroptical path. The prisms are 0.75mm wide and200µm thick, with 45◦ walls facing our opticalpath. (Figure 8)

5

Figure 8: Chamber sketch of our prism discs. f1 andf2 represent the walls the light crosses to measureabsorbance and p1 and p2 represent the prism-likestructures.

Disc assembly is quite different from has beenpreviously done. Another disc layout was designedand manufactured in Biosurfit. This disc layoutcomprised of chambers, also 200µm thick, withabout 1.5 times the area of our prism disc cham-bers. These chambers would allow the user toinsert the liquid and than send it to the prism discchambers at a desired time when centrifuging. Achannel would connect both chambers. Since bothdiscs are assembled with a dry bonding film inbetween, the prism disc chamber layout was cut inthe dry film so that the liquid could connect fromone disc chamber to the other. Assemblage withdry film was made using a hot-rolling process.

The LED used was the B5B-436-30 from Roith-ner, with λ = 660nm, output power of 3.5 cd and aviewing angle of 8◦. Both LEDs were attached toa disc tray and immobilized, and manually alignedwith Thorlabs Manual Stages until the emitterLED was above the inner prism and the receiverabove the outer prism of the prism discs. A boxwas built to place the disc tray inside so that noouter light could affect our results. A steppermotor is placed in the disc tray, and allows ourdiscs to slowly rotate.

A software was developed that would find theminimum discharge time value for each chamberand measurement. The sample would be insertedinside the bigger area chambers and by centrifug-ing, it would be led to the prism discs chamberwith the optical path.

As for the HbA1c study, a standard immuno-turbidimetric assay was performed. 20µL of

blood were diluted with 1000µL of an hemolysantreagent comprised of mostly water. Than 2µLof this dilution was added to 70µL of a latexsolution in water (< 0.13%). Both hemoglobinand HbA1c have the same unspecific absorptionrate to latex. This mixture was left incubating forone minute at room temperature, and than 25µLof a monoclonal-polyclonal antibody pair reagentwas added. The monoclonal antibody is a mouseanti-HbA1c and the polyclonal a goat anti-mouse.The monoclonal antibodies will than bind solely tothe HbA1c beads and the polyclonal will enhancethe aggregation of the latex beads connected tothe HbA1c. The higher the percentage of HbA1c,more HbA1c is around each latex bead, and denserand bigger aggregates are formed. This mixture isincubated at room temperature for 30 seconds or 1minute and placed inside the disc chambers.

3.2. Software

The Absorption Software is the software that allowsour method to maximize range and resolution. Itis divided in five modules with three different mainfunctions, from which the most important are the’measure’ and ’finder’ functions.

The ’measure’ function uses the stepper motorto find the minimum discharge time. It records adischarge time, moves a small step to the centerof the prism, and records a discharge time again.This process is repeated until a minimum is foundand the discharge times start to increase. Whenthe discharge time read is a security margin higherthan the minimum gathered value, this minimumis confirmed as an absolute minimum, and not alocal minimum. This process can be repeated au-tomatically, with the stepper going back and forth,to take various minimum values or to monitordynamic reactions.

The ’finder’ function uses the PEDD systemitself as a sensor to find the desired chamber. Ituses the detected discharge time to understand ifthe PEDD system is above a prism or not. Whenabove a prism, the stepper takes a step of about thedistance between two chambers, and this process isrepeated until no more chambers are found. Whenno more chambers are found, the system possessesa positional awareness and can rotate backwardsautomatically to the desired chamber.

A ’log’ file is automatically generated reportingall measured minimums, all errors - if any - andother relevant parameters. All of the accepted min-imum values are placed under an ’accepted’ file.

6

3.3. ResultsVarious tests were performed to assess the qualityand performance of this prototype. Figure 9 showsthat each chamber is different from one another,but our software is being able to reproduce theresults within different measurements.

Figure 9: CD21 discharge times measured for eachchamber using an early version of the AbsorptionSoftware.

A blood plasma sample was diluted in PBS andthe different dilutions placed inside different cham-bers. Figure 10 shows the results. An identicalset-up was employed using a bead solution dilutedin water, and the results can be seen in figure11. Both graphics show that our system is indeedmeasuring absorbance.

Figure 10: Discharge times (dt) for a plasma sampleand dilutions - %Plasma is the percentage in volumeof plasma within the solution with PBS.

Since different chambers have different behaviors,a study was made to assess if we could relate thesame chamber in different discs. Various watersamples were placed inside chambers six of differentdiscs, and than the relation between that chamberair value - that is, minimum discharge time withno sample inside - and the sample value wascompared. The exact same thing was made in thesame chambers for a highly diluted latex sample -1µL of a 10% latex 0.1µm solution diluted in 1000µL of water (L1). Figure 12 shows that there is

Figure 11: Discharge times (dt) for a latex reagentand dilutions - %Latex is the percentage in volumeof the latex reagent within the solution with water.

a relation between the air and sample values, andboth the water and the latex samples were alwaysdistinguishable.

Figure 12: tair (dt) against tsolution (dt). The yel-low line represents a line of slope 1 that passesthrough the L1 solution average point (±SD) clos-est to any Water point. The grey line represents aline of slope 1 that passes through the Water solu-tion average point (±SD) closest to any L1 point.Blue and green lines represent the slope 1 lines thatbetter fit to the L1 and Water average minimumdischarge time measured.

Using a standard immunoturbidimetric assay kitfrom Futura System S.r.l. for HbA1c determina-tion, the reaction was studied in order to asses thequality of our instrument for diagnosis purposes.Using two different protocols, the first providingresults in 14 minutes (3 min protocol) and thesecond providing results in less that 8 minutes(1.5 min protocol), HbA1c was measured and theregression lines relating our equipment results withthe reference method can be seen in figure 13 andfigure 14.

The 3 min protocol provided a pooled CV - withthirty samples measured in duplicate - of 3.3%,whereas the low HbA1c sample have a constanthigh CV. Three reagent lots were used to retrievethis thirty samples results, but only one calibration

7

Figure 13: Comparison between spinit PoC HbA1cprototype (%) and HPLC laboratory referencemethod (%) for the 3 minute protocol.

Figure 14: Comparison between spinit PoC HbA1cprototype (%) and HPLC laboratory referencemethod (%) for the 1.5 minute protocol.

curve was built, using the first eight measuredsamples.

The 1.5 min protocol provided a pooled CV -with sixteen samples measured in duplicate - of2.4%. Low HbA1c samples gave much lower im-precisions. A few samples had CVs over 3%, butalways when two different operators performed theduplicate. With the same operator preparing thesample manually, CVs fell around 2%. Total Erroris below 4%.

4. Conclusions

Both Total Haemoglobin concentration and HbA1cprototypes revealed themselves robust, accurate,and with the necessary range, precision and perfor-mance to measure the desired parameters withinthe specified criteria. The TIR system alongsidethe developed system and software allows us toincrease drastically our optical path and measurethe light absorbance with a high resolution andprecision. The PEDD system allows us to create alow-cost instrument with multiplexing capabilitieswithout apparent loss of range. Different testingson our system, such as with blood plasma, latexbeads solutions and HbA1c reveal the system assuitable not only to HbA1c measurements, but to

any clinical chemistry parameters.

The Total Haemoglobin prototype directly trans-formed a discharge time read into [Hb], meaningthat test results happened in a matter of twominutes due to sample lysis.

Imprecision, total error and CV for the HbA1cvalues were on or below the specified analyticalperformance criteria (CV < 3% and maximum biasallowed of ±0.75% HbA1c). Being a prototype,with still manually prepared samples and smallincubation times compared to the laboratoryprotocols, it is expected that when integrated intoa product with blood preparation automatized,this imprecision will lower even more.

The critical issue of the idea, that is, thealignment, that could lead to low light intensitytraveling through the sample, was totally solvedand it is believed that the stepper-like approachand the algo- rithm are stable, robust and flexibleenough to implement in another spinit R© withoutany main problems or concerns.

The algorithm itself is also flexible enough so thatmultiple LEDs can be used as emitter and detector,so that an array of different wavelength LEDs canbe implemented inside the spinit to test for a widevariety of clinical chemistry parameters. (Figure15)

Figure 15: SolidWorks sketch of a possible im-plementation of various PEDD system for multi-ple clinical chemistry parameters analysis withinspinit R©.

The promising results obtained are leadingBiosurfit to continue the presented work for a set ofproducts for commercialization. An HbA1c assaytest is a very important point-of-care test, sincepatients are advised to take to test at least eachthree to four months, and point of care systems

8

have been struggling to deliver an assay with theoptimal performance for HbA1c.

Many other parameters, such as lipids, cardiacparameters, electrolytes, among others, are gener-ally measured by colorimetric methods, meaningthat their study and measurement can be per-formed by the prototype described in this thesis,since, I reinforce, the software and experimentalsetup are robust and flexible enough so that theemitter and receiver LED, the position of suchLED and the machine can be changed, and theprototype will work the same. No different reagentsor different methods are needed to be tested anddeveloped, meaning that small fine tuning of theparameters and study of the reactions and proto-cols are the necessary work for a new assay, makingit faster to develop new high performance products.

AcknowledgementsThe author would like to thank all of BiosurfitS.A. personnel, specially to Joao Garcia da Fonseca,Jorge Almeida, Mauro Ribeiro, Nuno Reis, SandroBordeira and Rui Fonseca for all the hard work, sup-port and help given throughout the development ofthis thesis, indispensable to the extremely positiveresults and conclusions of this work.

References

[1] G. P. Zaloga, Evaluation of Bedside TestingOptions for the Critical Care Unit, Chest,1990, 97, 185S

[2] D. M. Baer, R. E. Belsey, The Evolving Reg-ulatory Environment and Bedside MetabolicMonitoring of the Acute Care Patient, Chest,1990, 97, 191S

[3] D. Janasek, J. Franzke and A. Manz, Scal-ing and the design of miniaturized chemical-analysis systems, Nature, 2006, 442, 374380.

[4] C. D. Chin, V. Linder, S. K. Sia, Commercial-ization of Microfluidic Point-of-care DiagnosticDevices, Lab on a Chip, 2012, 12, 2118-2134

[5] J. W. Hong and S. R. Quake, Integrated Nano-liter Systems, Nat. Biotechnol., 2003, 21, 11791183.

[6] R. Gorkin, J. Park, J. Siegrist, M. Amasia, B.S. Lee, J.-M. Park, J. Kim, H. Kim, M. Madouand Y.-K. Cho, Centrifugal Microfluidics forBiomedical Applications, Lab Chip, 2010, 10,1758-1773

[7] A. Manz, N. Graber, H. M. Widmer, Minia-turized Total Chemical Analysis Systems: A

Novel Concept for Chemical Sensing, Sens. Ac-tuators, 1990, B1, 244-248

[8] A. Manz, J. C. T. Eijkel, Miniaturization andchip technology. What can we expect?, PureAppl. Chem., 2001, 73, 1555 1561

[9] K. A. Erickson and P. Wilding, Evaluationof a Novel Point-of-Care System, the i-STATPortable Clinical Analyzer, Clin Chem, 1993,39, 283287

[10] C. T. Schembri, T. L. Burd, A.R. Kopf-Sil, L.RShea and B. Braynin, Centrifugation and capil-larity integrated into a multiple analyte whole-blood analyzer, J. Autom. Chem., 1995, 17,99-104

[11] C. T. Schembri, V. Ostoich, P. J. Lingane, T.L. Burd, and S. N. Buhi, Portable Simultane-ous Multiple Analyte Whole-Blood Analyzerfor Point-of-Care Testing, Clin. Chem., 1992,38, 1665-1670

[12] Biosurfit, S.A. www.biosurfit.com

[13] D. F. Swinehart, The Beer-Lambert Law, J.Chem. Edu., 1962, 39, 333

[14] M. Grumann, J. Steigert, L. Riegger, I. Moser,B. Enderle, G. Urban, R. Zengerle and J.Ducre, Sensitivity enhancement for colorimet-ric glucose assays on whole blood by on-chipbeam-guidance, Biomed. Microdev., 2006, 8,20914

[15] J. Steigert, M. Grumann, T. Brenner, L. Rieg-ger, J. Harter, R. Zengerle and J. Ducre, Fullyintegrated whole blood testing by real-time ab-sorption measurement on a centrifugal plat-form, Lab on a Chip, 2006, 6, 10401044

[16] E. Verpoorte, A. Manz, H. Ludi, A. E. Bruno,F. Maystre, B. Krattinger, H. M. Widmer, B.H. van der Schoot and N. F. de Rooij, A siliconflow cell for optical detection in miniaturizedtotal chemical analysis systems, Sens. Actua-tors, B, 1992, 6, 6670

[17] B. Splawn and F. Lytle, On-chip absorptionmeasurement using integrated wave guides,Anal. Bioanal. Chem., 2002, 373, 519525.

[18] J. Roulet, R. Vo lkel, H. Herzig, E. Verpoorteand N. de Rooij, Fabrication of multilayer sys-tems combining microfluidic and microopticalelements for fluorescence detection, J. Micro-electromech. Syst., 2001, 10, 482491

[19] Z. Liang, N. Chiem, G. Ocvirk, T. Tang, K.Fluri and D. J. Harrison, Microfabrication of a

9

planar absorbance and fluorescence cell of inte-grated capillary electrophoresis devices, Anal.Chem., 1996, 68, 10401046

[20] N. Petersen, K. Mogensen and J. Kutter, Per-formance of an in-plane detection cell with in-tegrated waveguides for UV/VIS absorbancemeasurement on microfluidic separation de-vices, Electrophoresis, 2002, 23, 35283536

[21] E. S. Yeung and R. E. Synovec, Detectors forliquid chromatography, Anal. Chem., 1986, 58,12, 1237A1256A

[22] J. Ducre, S. Haeberle, S. Lutz, S. Pausch, F.von Stetten and R Zengerle, The CentrifugalMicrofluidic Bio-Disk Platform, Journal of Mi-cromechanics and Microengineering, 2007, 17,S103-S115

[23] A. Nitkowski, L. Chen and M. Lipson, Cavity-enhanced on-chip absorption spectroscopy us-ing microring resonators. Optics Express, 2008,16, 11930-11936

[24] J. Steigert, M. Grumann, M. Dube, W.Streule, L. Riegger, T. Brenner, P. Koltay,K. Mittmann, R. Zengerle, J. Ducree, Directhemoglobin measurement on a centrifugal mi-crofluidic platform for point-of-care diagnos-tics, Sensors and Actuators A, 2006, 130-131,228-233

[25] L. Tymecki, M. Pokrzywnicka and R. Koncki,Paired emitter detector diode (PEDD)-basedphotometry an alternative approach, The An-alyst, 2008, 133, 1501-1504

[26] M. O’Toole, K-T Lau, D. Diamond, Photomet-ric detection in flow analysis systems using in-tegrated PEDDs, Talanta, 2005, 66, 1340-1344

[27] K-T Lau, S. Baldwin, M. O’Toole, R. Shep-herd, W.J. Yerazunis, S. Izuo, S. Ueyama, D.Diamond, A low-cost optical sensing devicebased on paired emitter-detector light emittingdiodes, Analytica Chimica Acta, 2006, 557,111116

[28] K. T. Lau, S. Baldwin, R. L. Shepherd, P. H.Dietz, W. S. Yerzunis, and D. Diamond, Novelfused-LEDs devices as optical sensors for col-orimetric analysis, Talanta, 2004, 63, 167-173

10