Analytica Chimica Acta 1205 (2022) 339782 Contents lists available at ScienceDirect Analytica Chimica Acta journal homepage: www.elsevier.com/locate/aca Biomedical point-of-care microanalyzer for potentiometric determination of ammonium ion in plasma and whole blood Antonio Calvo-Lo  pez a, Beatriz Rebollo-Calderon a, Aida Ormazabal b, Rafael Artuch b, n Alonso-Chamarro a, Mar Puyol a, * Javier Rosell-Ferrer c, Julia a Group of Sensors and Biosensors, Department of Chemistry, Autonomous University of Barcelona, Edifici Cn, 08193, Bellaterra, Barcelona, Spain b Clinical Biochemistry Department. Institut de Recerca Sant Joan de D eu, CERCA and CIBERER-ISCIII, Passeig de Sant Joan de D eu, 2, 08950, Esplugues de Llobregat, Barcelona, Spain c Research Centre for Biomedical Engineering (CREB), Universitat Polit ecnica de Catalunya (UPC), Campus Diagonal Sud, Edifici H., Av. Diagonal 647, Barcelona, Spain h i g h l i g h t s g r a p h i c a l a b s t r a c t  Low-cost Point-Of-Care analytical microsystem with integrated gas- diffusion, dialysis membranes and potentiometry.  Direct whole blood and plasma ammonium monitoring.  Physiological and pathological blood ammonium levels are covered with accuracy and precision.  Simple and automatic analytical method for ammonium determination. a r t i c l e i n f o a b s t r a c t Article history: Some inborn errors of metabolism and other diseases can result in increasing blood ammonium Received 12 November 2021 (hyperammonemia episodes), which can cause serious neurological complications in patients or even Received in revised form death. Early diagnosis, follow up and treatment are essential to minimize irreversible damages in brain. 24 March 2022 Currently, adequate analytical instrumentation for the necessary ammonium bedside determination is Accepted 26 March 2022 not available in all health centers but only in clinical laboratories of reference hospitals. We therefore have developed a low cost and portable potentiometric Point-of-Care microanalyzer (POC) to address Keywords: this problem. It consists of a cyclic olefin copolymer-based microanalyzer, the size of a credit card and Blood ammonium Hyperammonemia working in continuous flow, which integrates microfluidics, a gas-diffusion module and a potentiometric Monitoring detection system. The analytical features achieved are a linear range from 30 to 1000 mmol L1 NHþ 4, a Point-of-Care analyzer detection limit of 18 mmol L1 NHþ 4 and a required sample volume of 100 mL, which comply with the Ion selective electrodes medical requirements. Plasma and blood samples are analyzed with no significant differences observed between ammonium concentrations obtained with both the proposed microanalyzer and the reference method. This demonstrates the value of the developed POC for bedside clinical applications. © 2022 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). * Corresponding author. E-mail address: mariadelmar.puyol@uab.cat (M. Puyol). https://doi.org/10.1016/j.aca.2022.339782 0003-2670/© 2022 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). A. Calvo-Lo bal et al. pez, B. Rebollo-Calderon, A. Ormaza Analytica Chimica Acta 1205 (2022) 339782 1. Introduction In this work, a cyclic olefin copolymer (COC)-based continuous flow potentiometric POC microanalyzer is presented to monitor the A large group of inborn errors of metabolism (IEM), which may amount of ammonium ion in plasma and blood of patients diag- impair urea cycle, a liver metabolic pathway in charge of ammo- nosed with IEM or other diseases causing hyperammonemia. The nium elimination, present the accumulation of high levels of microanalyzer integrates monolithically microfluidics, a gas- ammonium in blood as main pathophysiological mechanism [1,2]. diffusion module and a potentiometric detection system in a This fact is clinically known as hyperammonemia [3,4]. Within a credit-card size single substrate. large group of diseases, primary disorders of the urea cycle, different organic acidemias and other metabolic conditions 2. Experimental affecting liver function, which can cause a secondary dysfunction of the urea cycle, are included. Urea cycle disorders (UCDs) are IEM of 2.1. Reagents and materials amino acids due to genetic defects in different urea cycle enzymes and transporters, which are involved in the conversion of ammo- POC microanalyzer prototypes were fabricated using layers of nium/ammonia into urea [5e9]. These defects result in episodes of COC from Tekni-Plex (Erembodegem, Belgium) in different grades hyperammonemia, the clinical outcomes of which may range from and thicknesses: COC 6013 foils of 400 mm-thick and COC 8007 foils subtle neurological manifestations during childhood/adulthood to of 25 and 50 mm-thick. hyperammonemic coma at few days after birth or during the evo- A hydrophobic gas-diffusion membrane made of polyvinylidene lution of the disease. Similarly, any chronic or acute liver disease fluoride (PVDF) with a 0.45 mm-pore diameter (Millipore, USA) was that leads to liver failure may end up with hyperammonemic en- used in order to separate the analyte from blood or plasma matrix. cephalopathy as one of the most severe consequences, since the A hydrophilic dialysis membrane made of polycarbonate (PC) detoxification of ammonium/ammonia through the urea cycle oc- with a 0.05 mm-pore diameter (Whatman, USA) was used in order curs in liver [10]. The range of usual ammonium concentrations in to protect the gas-diffusion membrane from adsorption of proteins healthy individuals ranges from 11 to 50 mmol L1 in adults, and up and other lipophilic compounds present in blood matrix. to 75 mmol L1 in newborns (according to some authors). Values A carbon-based ink (ELECTRODAG PF-407A, Acheson, France) beyond 200 mmol L1 are considered as severe episodes of hyper- was used as a conductive support for the ISE. An Ag/AgCl paste ammonemia that can cause irreversible neurological damage. (C2030812P3, Gwent Group, UK) was used as reference electrode. When patients present grossly elevated ammonium values (be- For the optimization and evaluation of the POC microanalyzer, tween 500 and 1500 mmol L1) mortality is a common complication analytical grade reagents were used. All solutions were prepared by [7,11e15]. Early and fast diagnosis and thus, rapid and controlled weighing out and dissolving the corresponding salt or compound in treatment are therefore essential to minimize the impact of double distilled water. hyperammonemia on neurological functions. This is done by re- Standard solutions of ammonium chloride (Acros Organics, striction of proteins intake, using drugs to enhance ammonium Belgium) were prepared by successive dilutions of a 100 mmol L1 elimination or carrying out dialysis or hemofiltration in acute cases stock. A KCl (Sigma Aldrich, Barcelona, Spain) solution (0.1 M) was [8,11,16]. used to keep constant the potential of the reference electrode. 4-(2- There are different commercial clinical equipment and kits to hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) (Fisher determine ammonium ion mainly in plasma samples. Among them, Scientific, Spain) buffering solution (1 mM) adjusted to pH 7 with those that are based on enzymatic (using the glutamate barium hydroxide (Sigma Aldrich, Spain) was used as acceptor so- dehydrogenase-catalyzed reaction) and spectrophotometric lution. NaOH (0.1 M) (Sigma Aldrich, Spain) with ethylenediamine methods are the most commonly used [17]. [e] [21] However, tetraacetic acid (EDTA) (103 M) (Panreac, Spain) were used as a despite they meet some of the requirements to take adequate basic solution that once mixed with samples or standards, acted as medical decisions, they still show some of the following different donor solution. constraints: lack of portability (large, bulky and heavy instrumen- ISE-membrane reagents were nonactine, bis(1-butylpentyl) tation), high cost, use only by skilled professionals in 3rd level adipate (BBPA), polyvinyl chloride (PVC) and tetrahydrofuran (THF), Hospitals or limited working range and precision. This issue makes all of them obtained from Fluka (Barcelona, Spain). these analyzers not suitable for regular and routine control of ammonium in blood to diagnose or follow-up diseases associated 2.2. Fabrication of POC microanalyzer with hyperammonemia episodes out of reference hospitals (for instance in secondary hospitals or health centers). Moreover it Modular and monolithic prototypes were developed during the cannot be implemented in developing countries, where even optimization stage for different purposes. The first was used to analytical kits are inaccessible due to their high price. For these evaluate the influence of the sample matrix on the integrity of reasons, portable, low cost, easy-to-use and automatic Point-of- different elements of the microsystem, so that if it was necessary to Care (POC) analyzers with high robustness, high throughput sam- replace any (malfunction or configuration changes) this could be ple analysis and low sample and reagents consumption are needed done easily without discarding the rest. The later was the final with great immediacy to allow precise bedside ammonium deter- version of the POC microsystem, used once the compatibility of all mination in plasma or whole blood. elements with the sample matrix was verified. Among the different analytical techniques to determine A multilayer approach was employed for the manufacture of ammonium ion, potentiometry and especially ion selective elec- prototypes. It consists in the lamination of different machined COC trodes (ISEs) are a good option to develop POC microanalyzers due layers with different glass transition temperatures (Tg). COC 8007 to their robustness, good selectivity, wide working range and foils with Tg ¼ 75  C were used as sealing layers. Blocks of two COC compatibility with flow techniques in which a high level of inte- 5013 foils with Tg ¼ 130  C, bonded previously using a sealing layer gration, automation, autonomy and high speed of analysis are between them, were used as fluidic structural layers, where all the achieved, without the need of using unstable enzymes or colori- designed patterns were machined. The fabrication process is metric reagents. In addition, the utmost selectivity can be reached described in detail elsewhere [25e28]. It briefly consists of four by the use of gas-diffusion membranes to separate the analyte from main steps: prototype design using CAD software, pattern the complex blood matrix [22e24]. machining using a computer numerically controlled (CNC) 2 A. Calvo-Lo bal et al. pez, B. Rebollo-Calderon, A. Ormaza Analytica Chimica Acta 1205 (2022) 339782 micromilling machine (Protomat C100/HF, LPKF, Spain), integration of different elements (such as electrodes and PVDF and PC mem- branes) and final lamination at 102  C and 4 bar to seal the microanalyzer using a thermo-compression press (Francisco Camps, Granollers, Spain). The modular prototype consisted of three separated compo- nents (Fig. 1): a micromixing, a gas-diffusion and a detection module. The size of each one was 27.0  34.0  1.5 mm, 32.0  56.0  6.0 mm and 26.5  34.0  2.0 mm, respectively. The dimensions of the microfluidic channels both for micromixing module and detection module were 0.4 mm-wide and 0.3 mm- height. The dimensions of the meander in the gas-diffusion module (Fig. 1.C) were 1.0 mm-wide and 0.1 mm-height, in order to maxi- mize the contact area between the ammonia generated and the gas-diffusion membrane. With the aim to allow an easy exchange of the PVDF/PC membranes when necessary, an aluminum holder was fabricated (Fig. 1.B and C). The diameter of the detection chamber (Fig. 1.D c) was 3.3 mm and the dead volume of the detection Fig. 2. Picture of the monolithic POC microanalyzer; a) fluidic connections; b) module was 22 mL. microfluidics; c) PVDF (above) þ PC (below) membranes; d) reference electrode; e) The monolithic prototype of the microanalyzer had the same detection chamber; f) ISE. elements as the modular version but in a single integrated device (Fig. 2). Its size was 50.8  66.6  4.5 mm and it weighed 14 g. The electrode during operation [29]. This reference channel acts as a dimensions of the microfluidic channels, where the gas-diffusion flowing liquid junction by merging with the main channel after the takes place and the detection chamber were the same as in the detection chamber. modular prototype. The dead volume of the detection chamber was The solid inner support for the ISE, made of a carbon-based ink, 15 mL and the total dead volume of the monolithic microanalyzer was placed by filling a bas-relief machined onto the COC layer. It was 110 mL. was cured at 80  C for 30 min. The Ag/AgCl paste, acting as the The principle of operation of the POC microanalyzer is very reference electrode, was screen-printed in a specific place over the simple. A carrier solution (Fig. 3 A or B channel d), where the reference channel using a screen-printer machine (DEK 248, DEK, sample/standard solution containing ammonium is injected, con- Spain) and then it was cured at 80  C for 30 min. verges with a NaOH solution (c) in a Y-shape confluence point. They The PC dialysis and PVDF gas-diffusion membranes were cut, get mixed along a serpentine micromixer obtaining ammonia. The washed with deionized water, dried and placed correctly in the gas- mixed stream reaches the gas-diffusion module, where a bottom diffusion chamber. meander-shape channel allows ammonia to diffuse, through the The ammonium selective polymeric membrane was prepared by binomial formed by the PC and PVDF membranes, to an acceptor weighing out, mixing and dissolving 1% nonactin, 65.5% BBPA and solution on the upper side. This acceptor solution (b), composed by 33.5% PVC in THF (3 mL) [30]. This cocktail was deposited dropwise HEPES at pH 7, collects and converts ammonia to ammonium. on the surface of the conductive support of a carbon-based ink Finally, a volume of ammonium solution arrives to the detector, inside the detection chamber [25,27], which was defined between causing a signal variation on the measured potential difference, the ISE membrane surface and the microchannels. The protocol which is proportional to its concentration. To keep constant the consisted in the repeated deposition of 1 mL of membrane cocktail potential of the reference electrode, an auxiliary KCl solution and evaporation for 5 min until the cavity of the detection chamber (0.1 M) is continuously pumped at 0.1 mL min1 over the reference was filled. Fig. 1. Modular microsystem prototype. A) Micromixing module: a) inlets and b) outlet; B and C) gas-diffusion module: a) inlet, b) outlet, c) microfluidic channels and d) PVDF/PC membranes; D) Detection module: a) inlets/outlet, b) reference electrode, d) reference electrode connection, c) Ammonium ion selective electrode (ISE) and e) ISE electric connection. 3 A. Calvo-Lo bal et al. pez, B. Rebollo-Calderon, A. Ormaza Analytica Chimica Acta 1205 (2022) 339782 Fig. 3. A) Experimental setup for the modular POC microanalyzer. B) Experimental setup for the monolithic POC microanalyzer. In both cases: (a) KCl (0.1 M); (b) HEPES (1 mM) at pH 7; (c) NaOH (0.1 M) þ EDTA (103 M); (d) H2O; (e) NHþ4 standard solution or sample; (P) peristaltic pump; (V) six-way injection valve; (M) micromixing module; (G) gas-diffusion module; (W) waste; (1) reference electrode; (2) ISE; (3) gas-diffusion membrane. Brownish color: microchannels under the PVDF membrane; Blue color: microchannels over the PVDF membrane. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) 2.3. Experimental setup samples for 10 min at 3000 rpm at 10  C. The remaining 54 blood samples were centrifuged by the same way to obtain the plasma Fig. 3.A shows the experimental assembly used with the fraction. These were analyzed in parallel using both methods. modular prototype. Once the proper functioning of the analytical Results obtained by the potentiometric POC microanalyzer were system was verified with a complex matrix, the monolithic proto- validated by comparison with those obtained by the reference type was developed and tested using the experimental setup method implemented in the clinical laboratory of HSJD. The later is shown in Fig. 3.B. In both cases, the continuous flow system setup based on an enzymatic method by means of an automated spec- consists of an external peristaltic pump (Minipuls 3, Gilson, Wis- trophotometric procedure in an Architect ci8200 automated consin, US) with Tygon® tubing (Ismatec, Wertheim, Germany) of analyzer (ABBOT, Park, Illinois, EE.UU) [31]. The analysis of the 1.14 mm internal diameter and a six-way injection valve (Hamilton samples by both methods was simultaneously done. The reference MVP, Reno, US). Teflon tubing (Scharlab, S. L., Cambridge, England) method is accredited by ENAC agency following the ISO 15189 of 0.8 mm internal diameter was used to connect the external el- norm, and it is subjected to external and internal quality control ements to the microsystem. A customized potentiometer (6016 4- schemes. These data are available on request. electrodes, TMI, Barcelona, Spain) and its software were used to signal acquisition and data processing. 3. Results and discussion 2.4. Plasma and blood samples collection and preparation 3.1. Optimization and analytical features of the potentiometric POC microanalyzer Before starting this study, the permission from the Ethics Committee of HSJD was obtained in order to validate the POC The main goal of this work was the development of a simple, system using patient's samples. These were remnant blood speci- selective, robust, portable and low cost POC microanalyzer for the mens from routine determinations of ammonium performed by the bedside monitoring of ammonium ion in plasma or whole blood of reference method in the HSJD laboratory. Once ammonium is patients with hyperammonemia. This application sets out some analyzed, theses samples are usually discarded after 1 week. We specific requirements such as enough working range to cover the collected an aliquot of 500 mL in an anonymized way. Thus, no ammonium physiological and pathological intervals, good baseline sensitive information about patients or their identity was stored, all stability, limited sample and reagents consumption, high automa- the experiments were conducted in the HSJD laboratory, and no- tion potential and avoiding sample pretreatments and operational data were transferred to external Hospital data repositories. skills. Keeping all this in mind, different parameters and variables 77 venous blood samples were collected in tubes containing that could influence the microsystem performance were evaluated heparin or EDTA (three of them in both tubes to compare the and optimized. anticoagulant influence over the ammonium determination), as The composition of the polymeric membrane used for the indicated in protocols of HSJD for routine ammonium analysis. In 26 ammonium selective electrode was based on a previous work [25]. of these blood samples, ammonium was immediately measured by The fabrication process of the carbon conductive support for the the developed POC microanalyzer in whole blood and in parallel in ammonium ISE was improved regarding epoxy-graphite paste- plasma samples by the reference method, which is done only in based electrodes used in other works [25,32]. Herein, a commercial plasma, to obtain contrast results. Plasma fraction was obtained screen-printable carbon ink was employed. A bas-relief was from the resulting supernatant after centrifugation of blood mechanized onto the COC substrate to enhance adhesion between 4 A. Calvo-Lo bal et al. pez, B. Rebollo-Calderon, A. Ormaza Analytica Chimica Acta 1205 (2022) 339782 materials and simplify the deposition process, which will allow an A basic solution composed by NaOH (0.1 M) þ EDTA (103 M) easier future industrialization of the microsystem manufacturing was used as donor solution to generate ammonia from ammonium. process. Carbon ink was cured at 80  C for 30 min to ensure the EDTA was added to avoid the precipitation in form of hydroxide of development of proper electrical properties by removing the some metals presents in blood matrix, a fact that could block or organic components. Since the thickness of the conductive support damage the microfluidic channels or the PC and PVDF membranes. layer was 300 mm, the conductive material was not damaged, when Regarding the main interfering compounds of the ammonium the THF-sensor cocktail was deposited dropwise. Analytical fea- ISEs present in plasma or blood, such as sodium or potassium ions tures such as sensitivity, baseline stability and detection limits [35], none of them are capable of diffusing through the PVDF gas- obtained with ISEs fabricated with this procedure were comparable diffusion membrane, therefore the proposed analytical micro- to those obtained using epoxy-graphite paste. system performs the highest selectivity versus ammonium ion. Regarding lifetime of the microsystem with whole blood sam- Flow rate of the carrier, donor and acceptor streams were tested ples, previous works revealed a progressive deterioration of the from 300 to 800 mL min1 and the sample injection volume was PVDF gas-diffusion membrane with successive injections of the varied from 25 to 500 mL. Best results, as a compromise between the same whole blood sample [32]. A peak decrease and a positive drift expected analytical and operational features stated before, were of the baseline potential were continuously observed until total obtained using a flow rate of 650 mL min1 for carrier, donor and signal loss, even with standard materials. This phenomenon could acceptor streams, and a sample injection volume of 100 mL. be related to the adsorption of lipophilic compounds present in the Fig. 4 shows, as example, the recording of the signal and the whole blood into the surface of the hydrophobic PVDF membrane. corresponding curve for one calibration. The obtained Nernst This fact, eventually leads to a series of surface modifications that equation (n ¼ 7; 95% confidence) was E ¼ 297 (±3) þ 63 (±1) log trigger a change in the membrane permeability, allowing NaOH to [NH4 þ] with r2 ¼ 0.9993. Linear range was from 30 to permeate to the acceptor channel. In order to protect the gas- 1000 mmol L1 NHþ 4 and the limit of detection, according to the diffusion membrane from this undesired event, different hydro- IUPAC [36], was 18 mmol L1 NHþ 4 . As it can be seen, the obtained philic dialysis membranes, regenerated cellulose (RC) and poly- linear range included the target concentrations corresponding to carbonate (PC), with different pore size were tested. These normal values of ammonium ion in blood, the cut-off value to membranes were integrated into the gas-diffusion module be- identify hyperammonemia in adults (50 mmol L1) and in children tween the donor microchannel and the PVDF membrane. In this (75 mmol L1), and typical values of acute cases of hyper- way, cross of the lipophilic compounds through the hydrophilic ammonemia (>200 mmol L1), so that the analytical features of the dialysis membrane was considerably minimized, while ammonia microanalyzer were ideal for the desired application. Moreover, it is pass was not blocked. With this configuration, signal peaks worth noting that the upper limit of the linear range is defined by decreased a bit but were big enough to comply with the required the highest concentrated standard solution used during the cali- limit of detection and PVDF membrane lifetime, and thus, the bration, which was chosen according to the concentrations most whole microsystem lifetime was enhanced dramatically. Best re- commonly present by patients. In the event that an ammonium sults in terms of microsystem lifetime, easy integration, detection concentration in a sample was punctually higher than the limit and analysis time were provided by a PC dialysis membrane maximum established in the calibration curve, the POC analyzer with 0.05 mm pore size. would provide valid results because the ISE presents a linearity two A countercurrent flow mode and a ratio 2:1 between donor and orders of magnitude above this limit. acceptor streams inside the gas-diffusion module were used as Repeatability studies were performed by successive injections previously optimized [25,32]. (n ¼ 10) of three standard solutions of 30, 100 and 1000 mM NHþ 4 , in Chemical and hydrodynamic parameters were also evaluated order to test the performance of the microdevice in the lower, using a univariate optimization procedure in order to achieve a medium and upper part of the expected range of ammonium in real compromise between baseline signal stability, expected sensitivity, samples. RSD of the signals were 4, 2 and 1%, respectively. Control proper working range, maximum analysis throughput and mini- materials and calibration solutions provided by Sant Joan de De u mum reagents and sample consumption. All optimized parameters, Hospital (HSJD), were tested to determine method accuracy including tested intervals and optimal values, are shown in Table 1. obtaining differences smaller than 6%. Reproducibility was also Other improvements in lifetime were directed to the buffering tested from 5 calibration experiments performed along 1 month. solution. A TRIS solution is commonly used in ammonium deter- Sensitivity determined as the mean slope of the 5 calibration curves mination procedures using ISEs [22,23,25]. However, as a primary was 57 mV with RSD value of 8% and the mean of the y-intercept for ammine in solution, it decomposes easily to ammonium as a result the Nernst equation was calculated as 333 mV with a RSD of 8%, of bacterial degradation and thermal instability [33,34]. In order to thereby demonstrating the intra and inter-day precision of the prevent the use of biocides or other preservatives and to simplify whole system. Reproducibility between POC devices was calculated the required chemistry, HEPES, phosphate or sulfate buffer solu- as the difference in the analytical response of 8 different ISEs. RSD tions were tested. Best analytical features were obtained using below 2% were obtained in both the mean of y-intercept and slope. HEPES (1 mM) at pH 7, being even slightly better than the ones A maximum sampling rate of 10 samples h1 was obtained, taking obtained with TRIS. as a reference the most concentrated standard solution (1000 mM), which takes longer to analyze. As the lower the ammonium con- Table 1 centration in the sample the higher the sampling rate, improved Optimization of chemical and hydrodynamic parameters. sampling rates are foreseen when real samples are analyzed in an Parameter Tested interval Optimal value automated POC, since not extremely high values are expected in the Injection volume (mL) 25e500 100 routine follow-up of the associated diseases. Microsystem lifetime Channel flow rate (mL min1) 300e800 650 was calculated to be about 3e4 weeks, depending on the number of Buffer solution: samples to be analyzed and the matrix (plasma or blood). [HEPES] (mM) 0.5e50 1 All results showed the potentiality and robustness of the whole pH 7e8 7 [NaOH] (M) 1$104 e 0.1 0.1 experimental setup for the proposed application. 5 A. Calvo-Lo bal et al. pez, B. Rebollo-Calderon, A. Ormaza Analytica Chimica Acta 1205 (2022) 339782 Fig. 4. Signal recording (A) and calibration curve (B) for a POC microanalyzer calibration using standard solutions of NH4Cl with NHþ 4 concentrations of 20 mmol L 1 (a), 30 mmol L1 (b), 40 mmol L1 (c), 50 mmol L1 (d), 100 mmol L1 (e), 250 mmol L1 (f), 500 mmol L1 (g) and 1000 mmol L1 (h). 3.2. Analysis of plasma and blood samples with a intercept of a ¼ 3 ± 3, a slope of b ¼ 0.98 ± 0.02 and a correlation coefficient of 0.9900. Different samples of plasma and whole blood from patients with These data show that the plasma matrix do not alter or damage different degrees of hyperammonemia, provided by HSJD, were the analytical microdevice nor any of its parts, demonstrating thus analyzed using the proposed POC microanalyzer. the viability of the POC microanalyzer presented for the analysis of First of all, plasma samples with a simpler matrix than that of ammonium in plasma. whole blood were analyzed. Once verified the technical feasibility Fig. 6 shows the results obtained from the analysis of 26 whole of the POC microsystem, we proceed with whole blood samples blood samples. From the statistical data treatment, a systematic analysis. error of approximately 12% underestimation was obtained with the Three plasma samples were pretreated with heparin and EDTA POC in comparison with the reference method. This could be as anticoagulant. Differences between results obtained with the explained for two reasons. 1) On the one hand, as the reference POC analyzer and those obtained with the reference method were method is only prepared for plasma samples, whole blood centri- less than 3% in all cases, thus demonstrating that it is indistinct to fugation must be previously done to obtain the plasma. This pro- use samples pretreated with heparin or EDTA. cess is likely to release ammonium from the red blood cells to the Fig. 5 shows the comparative test of both the developed POC plasma, which could justify the higher concentration values found. microanalyzer and the reference method for 80 plasma samples. 2) On the other hand, the complex whole blood matrix could alter From the statistical data treatment, it can be concluded that results the ammonium diffusion pattern with respect to the one existing obtained with the proposed POC microanalyzer were not signifi- when performing the calibration process with aqueous standard cantly different from the ones measured with the reference method solutions. This fact would cause a lower percentage of diffusion of according to the paired t-test (tcalc ¼ 1.372, ttab two-sided ¼ 2.000, ammonia through the PVDF membrane, which would lead to tcalc < ttab) and to the graphical representation of the results of each slightly lower concentrations than expected. It is worth mentioning method, obtaining a regression equation (n ¼ 80; 95% confidence) that despite these slight differences do not alter the medical deci- sion related with blood ammonium concentrations (since the pathologic threshold range applied in practice is set to avoid these problems and to prevent possible false negatives), we further studied this phenomenon to improve the POC performance. To evaluate the first hypothesis, two aliquots of the same blood sample were centrifuged. The one to be measured as plasma was analyzed by the reference method analogously to the rest of the samples, and the one to be measured with the POC microanalyzer was vortexed before being analyzed as whole blood. The results obtained with 3 samples did not show significant differences be- tween analyzing whole blood directly or whole blood previously centrifuged and vortexed, so this hypothesis was discarded. Taking into account the results obtained, we can conclude that matrix affects the diffusion profile temporarily, consequently reducing the amount of diffused analyte. In order to avoid this situation, blood matrix standard solutions could be used instead of aqueous ones. However, this would undoubtedly increase the cost per analysis, limit the reagents stability and cause an overestimation of 12% in the results of plasma samples analyzed. With a view of obtaining simple and low-cost microanalyzers to be able to be used in sec- ondary hospitals, health centers or in developing countries, and taking the underestimation in blood samples results as a systematic error, we considered to apply a signal correction by adding the Fig. 5. Comparative study between the results for the analysis of plasma samples average of the differences between both methods (12% for the 26 (n ¼ 80) obtained by the developed POC microanalyzer and the reference method. 6 A. Calvo-Lo bal et al. pez, B. Rebollo-Calderon, A. Ormaza Analytica Chimica Acta 1205 (2022) 339782 Fig. 6. A) Comparative study between the results for the analysis of whole blood samples (n ¼ 26) obtained by the developed POC microanalyzer and the reference method. B) The same comparative study with the correction factor applied to the POC microanalyzer results. analyzed blood samples), even if its origin is not currently exactly avoiding the centrifugation process, thus saving time, which makes known. Corrected results showed no significant differences ac- the proposed device much more competitive and attractive than cording to the paired t-test (tcalc ¼ 1.152, ttab two-sided ¼ 2.064, current ammonium analytical systems. tcalc < ttab) and to the graphical representation of the results of each Results and analytical features show that after an industriali- method, obtaining the following regression equation (n ¼ 26; 95% zation stage, the potentiometric POC microanalyzer assesses the confidence): [NHþ þ 4 ]HSJD ¼ 0.97 (±0.05) [NH4 ]POC þ 1 (±3); R : 2 commercialization potential in clinical applications for the bedside 0.9840. In order to obtain a highly robust and accurate method, a determination of ammonium in blood and plasma as well in pri- much more extensive study will be carried out in the near future to mary or secondary care centers as in health centers in developing obtain a statistically representative correction factor with a large countries, as alternative of current more expensive equipment. enough sample population. In order to test the repeatability of the microanalyzer with samples, a plasma sample and three whole blood samples were CRediT authorship contribution statement analyzed per quintuplicate. The results show satisfactory results Antonio Calvo-Lo  pez: Conceptualization, Microsystem design even with complex matrices, with a mean ammonium concentra- tion of 108 ± 5, 44 ± 4, 66 ± 3 and 68 ± 6 mM and RSD values of 2, 3, and fabrication, Methodology, development, Investigation, Raw 2 and 4% respectively. data analysis, Formal analysis, Validation, and, Writing e original To sum up, obtained results showed that the proposed analytical draft. Beatriz Rebollo-Calderon: Microsystem fabrication, Inves- POC microsystem is suitable for the determination of ammonium in tigation, Raw data analysis, Formal analysis, and, Writing e review & editing. Aida Ormaza bal: Samples preparation, Validation, and, plasma and is analogous to the reference method in terms of quality of the results and time of analysis, but in addition with a simple and Writing e review & editing. Rafael Artuch: Conceptualization, robust operation. For direct blood analysis, attained results were as Resources, Supervision, Project administration, and, Funding good as for plasma analysis, if a correction factor was applied. In acquisition, Writing e review & editing. Javier Rosell-Ferrer: n Alonso- Electronics integration, Writing e review & editing. Julia this case, the advantages are clearly more significant since centri- fugation process to obtain plasma is not necessary, which means Chamarro: Conceptualization, Resources, Supervision, Project saving time and resources, and allowing bedside measurements administration, and, Funding acquisition, Writing e review & practically immediately. editing. Mar Puyol: Conceptualization, Resources, Supervision, Project administration, and, Funding acquisition, Writing e review & editing. 4. Conclusions Declaration of competing interest A miniaturized and low-cost POC microanalyzer for direct determination of ammonium ion in whole blood and plasma is The authors declare that they have no known competing developed and successfully applied in samples from patients pre- financial interests or personal relationships that could have senting different plasma ammonium values. In comparison with appeared to influence the work reported in this paper. other previously reported POC analyzers [21], we improve analyt- ical characteristics in terms of working range and limit of detection, and regarding the operational mode, it results in simplicity and Acknowledgments ease of automation in continuous mode, with good precision, repeatability and reproducibility. In addition, it is a much smaller The authors would like to thank the Institute of Health Carlos III and cheaper analytical system than common clinical laboratory (ISCIII) by the financial support through project DTS18/00075 and equipment. Moreover, thanks to its reduced dimensions, it con- DTS18/00104, the Spanish Ministry of Science and Innovation sumes few amounts of reagents (which are common reagents through the projects MINECO-FEDER CTQ2017-85011-R, PID2020- indeed) and sample and shows a relatively fast analysis time. Above 117216RB-I00 and PDC2021-121558-I00, and to Catalan govern- all, the fact that direct whole blood can be analyzed, allows ment through project 2017SGR 220. 7 A. Calvo-Lo bal et al. pez, B. Rebollo-Calderon, A. Ormaza Analytica Chimica Acta 1205 (2022) 339782 References biological-fluids, Ann. Clin. Biochem. 31 (1994) 529e543, isi: A1994PR44100002. [20] H.C. van Anken, M.E. Schiphorst, A kinetic determination of ammonia in [1] A.W. El-Hattab, Inborn errors of metabolism, Clin. Perinatol. 42 (2015) plasma, Clin. Chim. Acta 56 (1974) 151e157, https://doi.org/10.1016/0009- 413e439, https://doi.org/10.1016/j.clp.2015.02.010. 8981(74)90223-X. [2] C. Alcalde, L. Aldamiz-Echevarría, F. Andrade, J.A. Arranz, F. Arrieta, R. Artuch, langer-Quintana, N. Benítez Brito, M. Bueno Delgado, [21] R. Goggs, S. Serrano, B. Szladovits, I. Keir, R. Ong, D. Hughes, Clinical investi- S. Balcells, A. Be stico y tra- gation of a point-of-care blood ammonia analyzer, Vet. Clin. Pathol. 37 (2008) J. Campistol, P. Campos, C. Casasnovas Pons, Protocolos de diagno nitos de metabolismo, 2018. https://ae3com.eu/ 198e206, https://doi.org/10.1111/j.1939-165X.2008.00024.x. tamiento de los errores conge [22] P.B. Martelli, J.G. Neto, E.A.G. Zagatto, S.M.B. Brienza, M.C.B.S.M. Montenegro, recursos/. J.F.C. Lima, Sequential analyte removal in flow analysis: determination of ni- [3] C. Bachmann, Mechanisms of hyperammonemia, Clin. Chem. Lab. Med. 40 trogen, phosphorus and potassium in fertilizers, Anal. Chim. Acta 317 (1995) (2002) 653e662, https://doi.org/10.1515/CCLM.2002.112. 239e245, https://doi.org/10.1016/0003-2670(95)00418-1. [4] A.S. Clay, B.E. Hainline, Hyperammonemia in the ICU, Chest 132 (2007) [23] S. Alegret, J. Alonso, J. Bartroli, E. Martinez-fabregas, Flow injection system for 1368e1378, https://doi.org/10.1378/chest.06-2940. €berle, N. Boddaert, A. Burlina, A. Chakrapani, M. Dixon, M. Huemer, on-line potentiometric monitoring of ammonia in freshwater streams, Analyst [5] J. Ha 114 (1989) 1443e1447. D. Karall, D. Martinelli, P.S. Crespo, R. Santer, A. Servais, V. Valayannopoulos, [24] T.S.U. Aoki, M. M, Continuous flow method for simultaneous determina- tion M. Lindner, V. Rubio, C. Dionisi-Vici, Suggested guidelines for the diagnosis of nitrate and ammonia in water, 515e517, Environ. Sci. Technol. 20 (5) and management of urea cycle disorders, Orphanet J. Rare Dis. 7 (2012) 32, (1986) 515e517, 20. https://doi.org/10.1186/1750-1172-7-32. [25] A. Calvo-Lo  pez, O. Ymbern, M. Puyol, J.M. Casalta, J. Alonso-Chamarro, [6] T.S. Raghuveer, U. Garg, W.D. Graf, Inborn errors of metabolism in infancy and Potentiometric analytical microsystem based on the integration of a gas- early childhood: an update, Am. Fam. Physician 73 (2006) 1981e1990. diffusion step for on-line ammonium determination in water recycling pro- [7] R.M. Cohn, Hyperammonemia, bane of the brain, Clin. Pediatr. (Phila). 43 cesses in manned space missions, Anal. Chim. Acta 874 (2015) 26e32, https:// (2004) 683e689, https://doi.org/10.1177/000992280404300801. doi.org/10.1016/j.aca.2014.12.038. [8] B. Lee, G.A. Diaz, W. Rhead, U. Lichter-Konecki, A. Feigenbaum, S.A. Berry, C. Le [26] O. Ymbern, N. Sa ndez, A. Calvo-Lo pez, M. Puyol, J. Alonso-Chamarro, Gas Mons, J. Bartley, N. Longo, S.C. Nagamani, W. Berquist, R.C. Gallagher, diffusion as a new fluidic unit operation for centrifugal microfluidic platforms, C.O. Harding, S.E. McCandless, W. Smith, A. Schulze, M. Marino, R. Rowell, Lab Chip 14 (2014) 1014e1022, https://doi.org/10.1039/c3lc51114f. D.F. Coakley, M. Mokhtarani, B.F. Scharschmidt, Glutamine and hyper- [27] A. Calvo-Lo  pez, M. Puyol, J.M. Casalta, J. Alonso-Chamarro, Multi-parametric ammonemic crises in patients with urea cycle disorders, Mol. Genet. Metabol. polymer-based potentiometric analytical microsystem for future manned 117 (2016) 27e32, https://doi.org/10.1016/j.ymgme.2015.11.005. €lker, D. Karall, J. Ha€berle, R. Posset, G.F. Hoffmann, space missions, Anal. Chim. Acta 995 (2017) 77e84, https://doi.org/10.1016/ [9] S. Nettesheim, S. Ko j.aca.2017.08.043. B. Heinrich, F. Gleich, S.F. Garbade, Incidence, disease onset and short-term [28] A. Calvo-Lo  pez, O. Ymbern, D. Izquierdo, J. Alonso-Chamarro, Low cost and outcome in urea cycle disorders ecross-border surveillance in Germany, compact analytical microsystem for carbon dioxide determination in pro- Austria and Switzerland, Orphanet J. Rare Dis. 12 (2017) 111, https://doi.org/ duction processes of wine and beer, Anal. Chim. Acta 931 (2016) 64e69, 10.1186/s13023-017-0661-x. https://doi.org/10.1016/j.aca.2016.05.010. [10] A. Hadjihambi, N. Arias, M. Sheikh, R. Jalan, Hepatic encephalopathy: a critical [29] N. Iban~ ez-García, M. Baeza, M. Puyol, R. Go mez, M. Batlle, J. Alonso-Chamarro, current review, Hepatol. Int. 12 (2018) 135e147, https://doi.org/10.1007/ Biparametric potentiometric analytical microsystem based on the green tape s12072-017-9812-3. technology, Electroanalysis 22 (2010) 2376e2382, https://doi.org/10.1002/ [11] A.J. Cooper, F. Plum, Biochemistry and physiology of brain ammonia, Physiol. elan.201000133. Rev. 67 (1987) 440e519. http://www.ncbi.nlm.nih.gov/pubmed/2882529% [30] N. Ibanez-Garcia, M.B. Mercader, Z. Mendes da Rocha, C.A. Seabra, 5Cnhttp://physrev.physiology.org/content/physrev/67/2/440.full.pdf. M.R. Go ngora-Rubio, J.A. Chamarro, Continuous flow analytical microsystems [12] W. Bernal, C. Hall, C.J. Karvellas, G. Auzinger, E. Sizer, J. Wendon, Arterial based on low-temperature Co-fired ceramic technology. Integrated potenti- ammonia and clinical risk factors for encephalopathy and intracranial hy- ometric detection based on solvent polymeric ion-selective electrodes, Anal. pertension in acute liver failure, Hepatology 46 (2007) 1844e1852, https:// Chem. 78 (2006) 2985e2992, https://doi.org/10.1021/ac051994k. doi.org/10.1002/hep.21838. [31] L.Y. Ha, W.W. Chiu, J.S. Davidson, Direct urine ammonium measurement: time [13] A.N. Rao, P. Varma, Sumitra, S. Dhanya, Hyperammonemia: diagnostic expe- to discard urine anion and osmolar gaps, Ann. Clin. Biochem. 49 (2012) rience at the metabolism laboratory, Perinatology 9 (2007) 9e13, https:// 606e608, https://doi.org/10.1258/acb.2012.012013. doi.org/10.5580/21c3. [32] A. Calvo Lo pez, Design, Construction and Evaluation of Miniaturized Analyzers [14] Y.-L. Tseng, C.-R. Huang, C.-H. Lin, Y.-T. Lu, C.-H. Lu, N.-C. Chen, C.-C. Chang, for Aerospace, Environmental, Food, Biomedical and Industrial Applications, W.-N. Chang, Y.-C. Chuang, Risk factors of hyperammonemia in patients with Universitat Auto noma de Barcelona, 2017. epilepsy under valproic acid therapy, Medicine (Baltim.) 93 (2014) e66, [33] F. Bougie, M.C. Iliuta, Stability of aqueous amine solutions to thermal and https://doi.org/10.1097/MD.0000000000000066. oxidative degradation in the absence and the presence of CO2, Int. J. Greenh. [15] J. Diaz, P.L. Tornel, P. Martinez, Reference intervals for blood ammonia in Gas Control 29 (2014) 16e21, https://doi.org/10.1016/j.ijggc.2014.07.008. healthy subjects, determined by microdiffusion, Clin. Chem. 41 (1995) 1048. [34] K. Yoshimura, S. Machida, F. Masuda, Biodegradation of long-chain alkyl- LP e 1048, http://clinchem.aaccjnls.org/content/41/7/1048.1.abstract. amines, 1980, pp. 238e241. [16] M.C.C. Machado, F.P. da Silva, Hyperammonemia due to urea cycle disorders: a [35] M.S. Ghauri, J.D.R. Thomas, Evaluation of an ammonium ionophore for use in potentially fatal condition in the intensive care setting, J. Intensive Care. 2 poly(vinyl chloride) membrane ion-selective electrodes: solvent mediator (2014) 2e6, https://doi.org/10.1186/2052-0492-2-22. effects, Analyst 119 (1994) 2323e2326, https://doi.org/10.1039/ [17] M.A. Qadar Pasha, R.B. Ram, M.D. Gupta, A rapid method for plasma ammonia AN9941902323. estimation using an indigenously purified enzyme, Indian J. Clin. Biochem. 15 [36] E. Lindner, Y. Umezawa, Performance evaluation criteria for preparation and (2000) 29e35, https://doi.org/10.1007/BF02873544. measurement of macro- and microfabricated ion-selective electrodes (IUPAC [18] R.J. Barsotti, Measurement of ammonia in blood, J. Pediatr. 138 (2001) Technical Report), Pure Appl. Chem. 80 (2008) 85e104, https://doi.org/ S11eS20. 10.1351/pac200880010085. [19] J.R. Huizenga, A. Tangerman, C.H. Gips, Determination of ammonia in 8