Towards using high-performance liquid chromatography at home

In order to make high-performance liquid chromatography (HPLC) more widely available at home and in small-scale settings, we have simplified two of its most costly modules, namely the pump and the detector. This should make the setup affordable for home or small laboratory use.

Trang 1

Jan Lankelma , Dirck J van Iperen , Paul J van der Sluis

a Department of Molecular Cell Physiology, VU University Amsterdam, O|2 Lab Building, De Boelelaan 1108, 1081 HZ Amsterdam, The Netherlands

b Department of Fine mechanics and Engineering VU - Bèta, VU University Amsterdam, The Netherlands

c Foundation for Chromatography at home, Demonstrator Lab, Amsterdam, The Netherlands

a r t i c l e i n f o

Article history:

Received 2 September 2020

Revised 13 January 2021

Accepted 16 January 2021

Available online 20 January 2021

Keywords:

Chromatography at home

low-cost HPLC

low-cost electrochemical HPLC detector,

low-cost HPLC pump

quantiﬁed self

a b s t r a c t

Inordertomakehigh-performanceliquidchromatography(HPLC)morewidelyavailableathome and

insmall-scalesettings,wehave simpliﬁedtwoofitsmostcostlymodules, namelythepumpand the detector.Thisshouldmakethesetupaffordableforhomeorsmalllaboratoryuse.AmanualHPLCpump wasconstructedsoastoﬁtintoacaulkgunfromalocalhardwarestoreenablingthegenerationof

100-150barofpressure.Inordertolimitthepressuredropduringtherunningofachromatogram,apulse dampenerwasdeveloped.Wefurthermodiﬁedtheelectrochemicaldetection(ECD)systemsoastouse

acheapboron-dopeddiamondelectrodewithanoverlayofthinﬁlterpaper,causinganeluentﬂowover theelectrodebywickingandgravity.Boththepumpandthedetectorareatleasttentimescheaperthan conventionalHPLCmodules

Usingahome-packedJupiter R Proteoreversedphasecapillarycolumnweshowhowthislow-costHPLC

systemgenerates wellresolving chromatogramsafter directinjection offreshurine TheECD did not loseitssensitivityduringregularuseovermorethanhalfayear.Forhomovanillicacid(HVA),whichis

ofmedicalinterest,wemeasuredalineardynamicrangeoftwoordersofmagnitude,adetectionlimit

ofHVAinthe injectedsampleof3 μMand acoeﬃcientofvariation<10%.Thecontribution topeak broadeningbythedetectorwasmuchsmallerthanthecontributionsbytheinjectorandbythecolumn Afterconsumptionoftableolivescontaininghydroxytyrosol(HT),itsmetaboliteHVAinthe correspond-ingurinecouldbemeasuredquantitatively.AnapproachtoquantifyHTintableolives ispresented,as well.Thismethodprovidesanewtoolforinvestigatingphysiologyofoneselforofdearonesathome

1 Introduction

The trend towards self-monitoring for possible life style ad-

justments is supported by new devices and methods Moreover,

health data tracking technologies further enable the quantiﬁed self

movement [1] The ultra-low cost microﬂuidic paper-based analyti-

cal devices ( μPADs) with colorimetric or electrochemical detection

[2-4] are recent developments Here the paper is mostly used for

transport by wicking rather than for separation Among separation

methods HPLC is a powerful tool for the analysis of e.g body ﬂu-

ids, but is generally too expensive for use at home or in small-scale

settings Moreover, sample pretreatment can be time consuming

Urine may be considered as a clear ﬁltrate of blood and can be

injected directly into a reverse phase column Indeed, measuring

metabolite proﬁles in human urine has the potential to monitor an

individual’s general health status [5]

∗ Corresponding author

E-mail address: j.lankelma@vu.nl (J Lankelma)

Significant effort to make HPLC smaller, port able and cheaper have been made [6-9] These developments were aimed at enabling analysis in the field or in a point-of-care system In paral- lel to miniaturization operating costs were lower due to a lower consumption of eluent, which made the new methods greener as well Lower construction costs have been mentioned [ 9, 10], but these portable systems are generally still too complex to construct and use at home In a recent review the history of the development of portable systems and the importance of pumps yielding a pulse-free flow, has been highlighted [11] Most pumps are battery- powered and such batteries may add a considerable weight to the instrument, compromising their portability An approach circum- venting the relatively high power requirement for pumping was the application of high-pressure gas [12] Although relatively inex- pensive, for its use at home we foresee that the safety of high- pressure gas may become an issue Another interesting pumping system is the electroosmotic pump [13] that was used for a portable HPLC by Lynch et al [14] The 5V power of a USB socket

of a laptop could be used to generate a pressure of 1200 bar The

https://doi.org/10.1016/j.chroma.2021.461925

Trang 2

construction of this pump and of its high voltage generating elec-

tronics is still relatively complex Most of the reported miniatur-

ized HPLC systems are still prototypes, with the exception of a few

that have been commercialized [15], but of which we have not yet

found indications of widespread use

When using capillary LC a low eluent ﬂow rate does not re-

quire continuous pumping, as long as the pressure can be held

high by a pulse dampener Quantitative electrochemistry can be

used at very low cost, as proven by millions of glucose meters

that are used by diabetics These two ideas led us to develop a

HPLC system with a highly simpliﬁed pump and detector: capillary

HPLC columns [16]with low eluent ﬂow rates and boron-doped di-

amond (BDD) electrodes for electrochemical detection with a sta-

ble sensitivity, which are essential components integrated in this

new setup Thereby we have built on HPLC developments during

the sixties of the previous century

We have tested home use for the detection of homovanillic acid

in urine We have shown that with relatively cheap instrumenta-

tion the urinary concentration of HVA may yield valuable medical

information that can be obtained at home

2 Materials and methods

2.1 Reagents

Homovanillic acid (HVA), hydroxytyrosol (HT), sodium phos-

phate dibasic dihydrate (Na 2 HPO 4 ) and sodium azide (NaN 3 ) were

purchased from Sigma-Aldrich (Zwijndrecht, The Netherlands)

Sodium chloride (NaCl) and sodium dihydrogenphosphate mono-

hydrate (NaH 2 PO 4 ) were from Merck (Darmstadt, Germany) The

buffer solutions were prepared using demineralized water

2.2 Instrumentation

2.2.1 Overview of the setup

The manually driven pump head was connected by standard

PEEK (polyether ether ketone) HPLC tubing (outer diameter 1/16

inch, inner diameter 0.5 mm) to a 3-way SS-41GXS1 switching

valve purchased from Swagelok (Waddinxveen, The Netherlands)

Through this valve the eluent from a syringe could ﬁll the cylinder

space of the pump head at low pressure This ﬁlling of the pump

head with eluent was facilitated by pulling the caulk gun piston

away (i.e to the right in Fig 1) By switching the 3-way valve,

the pump cylinder space was then connected to a ﬂow-through

manometer via a pulse dampener A high pressure was generated

by squeezing the caulk gun, thereby pushing its piston to the left

in Fig 1 Downstream the manometer a standard HPLC valve had

an external loop of 20 μL, followed by a fused-silica capillary LC

column of which the outlet was directly positioned on the work-

ing electrode of the electrochemical detector The type of the HPLC

valve and the use of timed injection for sub- μL injection volumes

will be elaborated upon below The wooden support construction

was of pressed bamboo Anodic oxidation currents were recorded

in Volts by a VC820 digital multimeter (Conrad Electronic Benelux,

Oldenzaal, The Netherlands) This HPLC setup weighed less than 5

Kg The digital signal was transferred to a laptop running under

Windows XP with Datalyse software ( http://www.datalyse.dk) for

data logging as described before [17] In a single run 30 0 0 data

points could be collected For chromatograms with relatively “slow

peaks” every 10 sec a data point was collected by the software For

rapid peaks, e.g without column for detector testing, every 1 sec a

data point was collected

For HVA we have compared several urine samples with analysis

by HPLC-MS-MS with the help of Dr Martijn van Faassen of the

University Medical Center Groningen HVA- 13 C 6 was used as inter-

nal standard A Spark Holland Symbiosis system was used for on-

Fig 1 Overview of the setup; 1) caulk gun with PEEK pump head, 2) T-valve, 3)

PEEK pulse dampener, 4) manometer, 5) PEEK connecting tubing, 6) HPLC injection valve, 7) capillary column, 8) electrochemical detector, 9) multimeter

line solid-phase extraction and liquid chromatography Mass spec- trometric detection was performed by a Waters Xevo TQ-S micro triple quadrupole mass spectrometer in positive mode The method has been validated according to the Dutch guideline for validation

of analytical methods in medical laboratories by the Dutch Society

of Clinical Chemistry and Laboratory Medicine (NVKC)

2.2.2 Pump head

We developed a PEEK pump head module ( Figs 1, 2) that was machined by Microtherm B.V (Oudkarspel, The Netherlands) The outer diameter of the left-hand part (38 mm) as shown in Fig.2 was such that the module ﬁtted into a standard caulk gun from a local hardware store ( Fig.1) A DS 119 PTFE seal (6 mm x

12 mm, 6 mm long) was obtained from Eriks (Alkmaar, The Nether- lands) A stainless steel (SS 316) piston with a diameter of 6 mm was generating the high pressure when pushing the piston to the left ( Fig.2) The piston was guided by a Teﬂon slider within the shaft for positioning A side branch was welded to the piston for better grasp when pulling it for ﬁlling with new eluent

2.2.3 Pulse dampener

In order to reduce the fall in pressure during the running of

a chromatogram, a pulse-dampener with a PEEK body was de- signed by us and machined at the Technics Campus Den Helder, The Netherlands It consists of two PEEK disks of 24 mm (upper disk) and 10 mm thick (lower disk) that are pressed together by twelve M5 bolts (see Figs 1and 3) with a Viton O-ring (dimen- sions 33 mm x 3 mm; obtained from Eriks, Alkmaar, The Nether- lands) between them In the lower disk a circular groove (width

3 mm and depth 2 mm) contained the O-ring The space created

by the O-ring and both PEEK walls of the chamber formed by the O-ring will be elastically reshaped at high pressures

Trang 3

Fig 2 Construction of the pump head module The left-hand part (outer diameter 38 mm) contained the seal with an internal back-up ring and was connected through a

female VICI Valco ﬁtting to 1/16 inch PEEK capillary tubing Left and right PEEK parts were assembled as indicated by the arrows using two bolts and nuts

Fig 3 Pulse dampener for keeping the pressure suﬃciently constant during a chromatographic run It consists of two PEEK disks, with an O-ring in between, and ﬁrmly

connected to one another with 12 bolts and 12 nuts The chamber thus formed is connected to the eluent by two female VICI Valco ﬁttings connected to two small channels (disk on the right, see arrows)

2.2.4 Electrochemical detector

The electrochemical detector presented here is a modiﬁcation of

a previous design in which liquid transport along the working elec-

trode was driven by wicking and gravity alone [17], rather than by

the pressure generated using the caulk gun The electronics were

modiﬁed for amplifying small currents (Fig S1) The column end

was directly positioned onto a triangular piece of 105 Whatman

lens cleaning tissue with a thickness of approximately 35 μm cov-

ering a part of the electrode We have used a 10 mm x 10 mm

boron-doped diamond electrode obtained from Condias (Itzehoe,

Germany) ( Fig.4), for obtaining a better stability than when using

a glassy carbon electrode [ 18, 19] Downstream the working elec-

trode, the lens cleaning tissue was in contact with a piece of 1

mm thick ﬁlter paper that was itself in contact with ﬁlter paper at

the bottom of a small Petri dish with an outer diameter of 35 mm

An Ag/AgCl reference electrode (middle electrode) and a stainless

steel auxiliary electrode (left behind the reference electrode) were

standing on the wetted bottom paper disk The level of the elu-

ent did not rise during elution because of a hole near the bot-

tom of the small Petri dish through which a cotton wire estab-

lished contact with a strip of ﬁlter paper reaching to the bottom

of the glass vial ( Fig.1) Adherence between the different pieces of

wet ﬁlter paper was achieved through capillary force just by gentle

touching

2.2.5 Capillary columns

For packing capillary LC columns [ 16], fused silica capillaries from Polymicro (Phoenix, AZ, U.S.A.) with an OD of approximately 0.35 mm and an ID of 0.20 mm were used at various column lengths, as indicated below A porous ceramic frit was made using potassium silicate and formamide, followed by polymerization

at elevated temperatures using a procedure modified from Meir- ing et al [20], heating at approximately 1 °C/min to 150 °C, hold- ing for 120 minutes at 150 °C and then cooling down at approximately 2 °C/min Instead of the usual oven of a gas chromatograph, a home gas furnace was used A 4 mm hole was drilled into a refractory brick, ending approximately 3 cm above the gas flame Together with a thermocouple the capillaries were both positioned at the end of this hole After calibration using an oil bath, the readout of the thermocouple was followed in mV and the gas flame was readjusted if necessary to create the right temperature track The columns were packed at approximately 100 bar using an in-house built stainless steel module containing an internal space with 1 ml of a stirred slurry of 50 mg/ml of column packing material (Jupiter 4u Proteo 90A from Phenomenex, Torrance, CA , U.S.A )

in 2-propanol, that was ﬁrst sonicated for 5 min When irregular- ities of the packing were observed at the top of the column after packing at home the corresponding piece was cut off by a ceramic capillary cutter The columns were run using 50 mM phos-

Trang 4

Fig 4 Above, electrochemical detection cell (see also Fig 1 ) Using a lid of a Petri

dish (a) and a pipet tip end, the outlet of the capillary column was positioned on

the upstream corner of a triangular piece of lens cleaning tissue (b) partially cov-

ering the working electrode The reference electrode (c) and the auxiliary electrode

(d) can also be seen

Below, schematic representation The outlet of the capillary column touched the

working electrode (we) that was partly covered by a triangular piece of tissue (pa-

per strip 1); this ﬁlter paper touched a wet 1 mm thick piece of ﬁlter paper (paper

strip 2) that was in contact with the reference electrode (ref) and the auxiliary elec-

trode (aux) at the bottom of the Petri dish (containing a wet 1 mm thick ﬁlter paper

disk) A cotton wire guided the ﬂow to another strip of ﬁlter paper (paper strip 3)

reaching to the bottom of the glass vial In this way the amount of ﬂuid in the Petri

dish stayed constant and was limited to wetting the bottom paper disk

phate buffer (pH 6.82) containing 10 mM sodium chloride for elec-

trical conductivity and for maintaining a stable reference poten-

tial Sodium azide (0.02%) was added for prevention of growth of

microorganisms The capillary columns were operated at ambient

temperature

2.2.6 Sampling valves

A macroscopic VICI Valco HPLC injection valve with a sam-

ple loop of 20 μL (model Cheminert C2-1006D with, in our case,

a stator bore of 0.5 mm) allowed column cleaning and abolish phase collapse [21]by injecting the complete loop volume with 2- propanol/0.1 N nitric acid in a volume ratio 4/1 [22] For small injection volumes timed injection was used, whereby the valve was temporarily switched to “INJECT” and switched back to “FILL” after 10 or 20 sec, creating an injection volume of 0.17 μL or 0.33

μL at a typical eluent flow rate of 1 μL/min The eluent flow rate was estimated by dividing the void volume by the unretained retention time The fraction void volume/total volume (total column porosity) was taken as 0.6 [23] A VICI Valco micro-injection valve (model Cheminert C4-1004-.5 with stator bore of 0.25 mm and an internal loop of 0.5 μL) was used in cases where an exact and fixed injection volume was needed, e.g when the number electrons per oxidized molecule was calculated during oxidation at different flow rates

3 Results and discussion

3.1 Functioning of the pump module and the pulse dampener

A working pressure of 100 bar was used most of the time For building up this pressure, manual compression of one cylinder volume of the pump module (see Fig.2) was suﬃcient How- ever, by strong squeezing of the caulk gun a pressure of 170 bar could also be reached At the start we injected approximately 20

μL of the propanol/nitric acid mixture (for details, see above) in order to abolish phase collapse at zero flow rate and for removing highly retained compounds of previous injections To avoid pollu- tion of the electrode the column was uncoupled from the detector during this procedure The best time for this was late in the evening, since the following morning a series of chromatograms could be started without losing too much time for stabilization In this study isocratic elution with a buffer without organic modifier was used Under these conditions phase collapse is a serious dan- ger when the flow is stopped, e.g by leaving the HPLC valve in

an intermediate position In case phase collapse happened, 20 μL

of the propanol/nitric acid mixture was injected (see above, under 2.2.6), and the column was allowed to stabilize for at least 1

h Phase collapse during the night could be avoided by applying

a starting pressure of 150 bar in the evening before The purpose

of a pulse dampener is to reduce pressure variations Omitting the pulse dampener could lead to a rapid drop in pressure, and un- acceptable peak broadening, as shown in Fig 5 This figure also shows how the pressure evolved for starting pressures of 50, 100 and 150 bar as well as the corresponding effect on the HVA peak when using the pulse dampener To prevent leaking back along the seal in the pump module, the T-valve was set in the filling position during a chromatographic run This meant that at the high- pressure side the T-valve was closed and the eluent under high pressure was not connected to either the caulk gun or to the filling syringe After switching the T-valve, the pressure was raised manually to the starting pressure at the start of a new chromatogram

by squeezing and reading the pressure gauge Using a 15 cm column the pressure did not decrease by more than 20% during one hour Usually, two times squeezing the caulk gun (corresponding

to about 1/3 of the cylinder volume) was enough to restore the pressure for the next one-hour run The eluent flow was pulse-free and resulted in a stable amperometric baseline This work shows the generation of chromatographic peaks at relatively low cost The cost was low not only thanks to the simplified instrumentation, but also consequent to minimal use of eluent, because of the low flow rate The pump is not a black box with an on/off switch, but users generate the pressure by hand and come in direct contact with the pressure generating process Therefore this pump may become an instructive tool, useful for HPLC education

Trang 5

Fig 5 Effect of the pulse dampener on the reduction in pressure (broken lines;

axis on the right) Homovanillic acid (drawn lines; axis on the left) peaks at starting

pressures of 50 (green), 100 (black) or 150 bar (blue) with pulse dampener Without

pulse dampener, the pressure (starting at 100 bar) dropped rapidly (the red broken

line), leading to signiﬁcant peak broadening caused by reduction of the eluent ﬂow

rate (the red HVA peak) HVA concentration 0.2 mM; calculated injection volume

0.28 μL; column length 30 cm

3.2 Functioning of the electrochemical detector

After working for over half a year with the same electrode and

without any electrode cleaning we observed no loss of sensitivity

for HVA at + 0.7 V (vs Ag/AgCl/Cl −) [24] This is an advantage over

a glassy carbon electrode, which needs regular polishing to pre-

vent loss of sensitivity [25] At the potential of +0.7 V used, the

sodium azide that had been added to prevent microbial growth,

did not increase the background current signiﬁcantly The refer-

ence electrode must be close to the working electrode in order

to minimize a so-called iR-drop and maximize the linear detection

range [25] In the present conﬁguration the upper limit of the lin-

ear range was measured to be 0.6 μA (see Fig S2) The detector

was tested further in the linear range below this upper limit Nor-

malized per molecule applied to the column, the integrated peak

area should provide information on the number of electrons trans-

ferred per molecule during the oxidation accompanying the detec-

tion We had aimed for a thin diffusional liquid layer during wick-

ing through the thin ﬁlter paper on the working electrode In this

way we facilitated the diffusion so that most molecules should be

oxidized by the electrode In accordance with coulometric detec-

tion [25] we indeed measured no change in total peak area when

the pressure was lowered down from 100 bar to 20 bar, support-

ing the idea that all molecules were being oxidized For HVA, us-

ing the law of Faraday [25] we measured 3.3 electrons/molecule

under our detection conditions (see Fig S3) This number is high

compared to the number expected on the basis of the number of

OH-moieties per molecule [26] Rapid polymerization reactions af-

ter the ﬁrst one-electron oxidation could lead to reactive interme-

diates reacting further to molecules with more oxidizable groups,

possibly explaining this high number [26] Using this number of

electrons in the data analysis, it might still be possible to quan-

tify HVA without the need for comparison with a standard solu-

tion However, because the number is not an integer and cannot

robustly be explained from a chemical reaction scheme at present,

we prefer calibration through spiking, because unforeseen varia-

tions in the reaction conditions may occur leading to variations in

the net number of electrons per molecule In order to minimize

inﬂuence of ambient temperature ﬂuctuations we advise frequent

comparisons with a standard and to position the setup in a spot

with relatively stable temperature, not close by a heat source

Fig 6 Normalized output signal after quickly touching the working electrode sur-

face (red line) with a tip of a pipette made from a disposable insulin syringe [ 17 ], after a 2 sec timed injection (blue line) with a Vici micro-injection valve (model Cheminert C4-1004-.5 with stator bore of 0.25 mm and an internal loop of 0.5 μL) and (green line) after a 2 sec timed injection with a “macroscopic” HPLC Vici valve (model Cheminert C2-1006D with, in our case, a stator bore of 0.5 mm and an external loop of approximately 20 μL) The valves were connected with the working electrode by low-volume fused silica tubing (length 15 cm, ID 25 μm) The ﬂow rate for all three cases was 1.7 μL/min For these signals the data were collected every second by the Datalyse software

The noise level was about 1 nA (see Fig S4) After injection of 0.5 μL of a HVA solution the detection limit taken to reside at three times the noise level corresponded to a concentration of approximately 3 μM in the injected sample At 0.2 mM and 10 sec

on INJECT, the coeﬃcient of variation between HVA peak heights was smaller than 10% (n = 8) The linear range, expressed as the ratio between the upper limit of the linear range and the detection limit was 10 2 This was in the same range as that reported in

a previous publication on the measurement of HVA in urine with macroscopic HPLC and electrochemical detection at a ﬂow rate of 1.3 mL/min [27]

When comparing different chromatograms, the retention times may differ somewhat (see Fig S5) If necessary, in the future a direct overlay of chromatograms may be accomplished by time warping algorithms [28]

The same electronics were used as described earlier for measuring glucose concentrations using 6 V and 9 V battery power for the amplifier and the multimeter, respectively [17] This amplifier was chosen for its low price and simplicity and has not yet been optimized for HPLC detection Further optimization by choosing low-noise amplifiers might well decrease the detection limit and expand the linear range Preliminary measurements using a new open source amplifier [29] that also allowed other electrochemical modes, such as cyclic voltammetry, confirmed this expectation (data not shown) This will be worked out in a future paper

3.3 Peak broadening

Contributions by injection volume and couplings to peak broadening merit attention when using capillary columns Peak broadening by the coupling between the column and the detector is expected to be minimal, as the end of the fused silica column rests directly on the triangular piece of lens cleaning tissue on the working electrode ( Fig.4) Indeed, introduction of a pulse of HVA directly onto this ﬁlter paper by a tip of a home-made pipette [17]showed a relatively sharp peak (red line in Fig.6), when compared to the peaks by using two different HPLC injection valves with timed injection (only 2 sec on INJECT) For this measurement both valves were connected to the working electrode by a low- volume fused silica capillary (length 15 cm, ID 25 μm) The direct detection peak shape was asymmetrical Peak broadening caused

Trang 6

Fig 7 On the left the normalized injection peak shape after timed injection with the micro-injection valve during 1 sec (red), 2 sec (dark blue), 5 sec (green), 15 sec

(magenta) at INJECT and the complete loop of 0.5 μL (light blue) In the middle the normalized injection peak shape using a normal macroscopic valve (see Fig 6 ) for 2 sec (magenta), 5 sec (red), 10 sec (blue) and 20 sec (green) at INJECT For experimental setup, see Fig 6 On the right the resulting volume dispersion σv values for the micro-valve (blue) and the macroscopic valve (brown)

Fig 8 Chromatographic homovanillic acid (HVA) peaks after using a “normal macroscopic HPLC valve” (as mentioned underFig 6 )) for various times at INJECT (magenta

2 sec, green 10 sec, red 30 sec, dark blue 60 sec, light blue 180 sec); the HVA concentration was 40 μM On the left the primary signal and on the right comparison of normalized peaks The ﬂow rate was inferred by dividing the estimated void volume by the elution time of the unretained peak (t Ro )

by an increased injection volume ( Figs 6 and 7) resulted in less

asymmetry Approximating the peaks by Gaussians, the temporal

dispersion σt was estimated as half the width at 0.6 of the max-

imal height The corresponding volume dispersion σv was calcu-

lated by multiplication with the ﬂow rate The ﬂow rate was in-

ferred from the void volume by division through the elution time

of the unretained peak (t Ro ) As band broadening caused by the de-

tector was much smaller ( σv of less than 0.1 μL) than that caused

by the injection volume of the HPLC valves ( Fig.6), the contribu-

tion of the detector to the peak width was negligible

Next, we looked at the peak broadening caused by the injection

For this, the sample injection volume was varied by using different

INJECT times The resulting injection peaks for the two HPLC valves

are presented in Fig.8

At small injection volumes, the calculated values for σv were

larger than predicted by the empirical formula presented by Prüss

et al [30], but at higher injection volumes these differences be-

came less This can be explained by less inﬂuence of dispersion at

the edges of the peak proﬁle [30] However, it is not our aim to

theoretically dissect peak broadening over a large range of injec-

tion volumes In practice, INJECT times of 10 or 20 sec could be

handled most conveniently

Under the chromatographic conditions used to produce

Fig 8 for up to 60 sec at INJECT, a “normal macroscopic” HPLC

valve could be used, because the chromatographic peak was only

affected for INJECT times > 60 sec Therefore, neither a nanovol-

ume injection valve (corresponding to < 1 sec on INJECT), nor a

splitter system [20]should signiﬁcantly reduce peak broadening in

the present conﬁguration At the highest INJECT time (180 sec) a σt

of 95 sec was found and a corresponding σv of 1.7 μL Again as- suming Gaussian distributions and independence, variances in vol- umetric units of the ﬂuid path up- and downstream from the column can be added to the column variance to obtain the total peak variance [ 31, 32] As below INJECT times of 60 sec σv did not decrease, while the inﬂuence on peak broadening by the injection did ( Fig.7), this σv (0.6 μL) was taken as the peak broadening caused

by the column The calculated injection variance in volume by IN- JECT at 180 sec, according to Prüss et al [30], σv, inj 2 , was added

to the column variance The square root gave a σv for the total peak of 1.1 μL, which was in the same range as the σv of 1.7 μL, measured experimentally

An increase of the INJECT time and thus an increase in the injection volume will lead to higher peaks, with a better signal to noise ratio However, as we have discussed above, increasing the injection volume can increase the peak width For peaks at a lower retention time than that of HVA or for better columns with nar- rower peaks, INJECT times of 60 sec or less can have more impact

on total peak broadening than shown in Fig.8 For such cases an

LC valve with a bore diameter of 0.25 mm is advised

3.3 Applications

The aim of this work is to present a new tool for HPLC analysis

at home or in small laboratories Attractive should be the analysis

of urine without or with a simple sample pretreatment (e.g solid phase extraction) After direct injection of fresh urine of a healthy

Trang 7

Fig 9 Analysis of two urine samples, one (A) just before, and one (B) collected at 1.8 h after taking 10 Kalamata table olives on an empty stomach, indicating the conversion

of hydroxytyrosol from the olives to homovanillic acid (at arrow) Column length 15 cm

68-year old volunteer the present method generated 5-20 peaks

( Fig.9A) As compared to frozen urine, fresh urine samples offer

the advantage that no precipitates have to be removed before in-

jection and that there is no oxidation due to storage We focused

on the effect of table olives as a key component of a Mediterranean

diet HVA was determined after intake of table olives and quercetin

tablets After taking 10 pitted Kalamata table olives on an empty

stomach, a rapid rise of the peak at the retention time of HVA was

detected in the ﬁrst urine collected around 1.8 h after ingesting the

olives ( Figs.9A and 9B) This is in accordance with rapid absorp-

tion and a metabolic conversion of HT from the olives [33] Sub-

sequent urine samples showed a decline of the HVA peak At re-

tention times in excess of 40 0 0 sec the chromatogram of the urine

of Fig.9B showed no signiﬁcant peaks ( Fig.10), so under the pre-

sented chromatographic conditions the system showed to be ready

after about 1 h for the next sample Cleaning of the column by

the propanol/nitric acid mixture (see above for details) was done

about one time per week and columns could be used for at least

half a year enabling HVA measurements, comparable to that shown

in Figs.9A and 9B

The HVA concentration in the ﬁrst morning urine sample calculated with the standard plot (Fig S2) and adjusted for the injection volume, was in the order of 10 μM Stroe et al showed that the urinary HVA concentration is more than two orders of magni- tude higher than the blood concentration [34], presumably by active tubular organic anion transport in the kidney [35] The urine results thereby yield (indirect) information of variations in the low blood concentrations of HVA

The content of HT in table olives was measured under the present chromatographic conditions after the following extraction procedure The fruit ﬂesh of three olives was cut into pieces of 2-

3 mm and immersed in 200 ml of tap water overnight The following morning, after stirring a couple of times, the mixture was filtered through cotton wool and the filtrate was directly injected into our chromatographic system (Fig S7) A second extraction of the residue (under the same conditions) yielded less than 5% of the first extraction, indicating that almost all the HT had diffused overnight into the water

An additional experiment was the determination of HVA after intake of quercetin After taking an oral dose of 500 mg of

Trang 8

Fig 10 The chromatogram of Fig 9 B at a longer time scale, showing the absence of signiﬁcant peaks in the later phase

quercetin and subsequent collection of urine samples from two

healthy volunteers, a peak at the retention time of HVA was

emerged much later (around 12 h, data not shown) than after the

consumption of Kalamata olives This delay may reﬂect the intesti-

nal transit time and conversion of quercetin to HVA by the mi-

croﬂora in the large intestine [36] Therefore, the HVA pattern may

carry quantitative information on the composition of the micro-

biome

The tool we present here does not identify the peaks at the re-

tention times of HVA or HT as HVA or HT with 100% certainty,

even after spiking the sample with pure HVA or HT For our test

compound HVA we have compared urinary concentrations with

those measured by HPLC-MS-MS The agreement between both

methods ( Fig 11) conﬁrms the identity of the measured peak to

be that of HVA

Increasing the length of a home-packed column beyond 15 cm

enhanced separation power, with inherently longer elution times

(Figs S5 and S6)

3.4 Perpectives

In an attempt at simplifying HPLC we have integrated a low-

cost hand pump, a pulse dampener and a battery-powered detec-

tor The estimated total cost of the presented setup amounts to

EUR 2200 for off-the-shelf components, including EUR 10 0 0 for a

commercial HPLC injection valve, and excluding the cost of con-

struction labor For other published comparable miniaturized and

low-weight HPLC systems [11]at least one HPLC sample valve was

needed for sample introduction A refurbished valve could reduce

the price of the total setup signiﬁcantly By comparison, relatively

simple isocratic HPLC pumps by which the same chromatogram

can be created, range from EUR 50 0 0 to EUR 10 0 0 0 Although our

newly constructed modules are relatively simple, instrument con-

struction skills are required The ampliﬁer could simply be con-

structed by any moderately experienced electronics craftsman, or

by using a breadboard for connecting the electronic resistances,

op-amps, etc The PEEK pump module can be made using a lathe,

or ordered like we did Interesting possibilities may exist in the

near future for 3-D printing of the PEEK modules or of the sam-

pling valve Increased demand should lower the price and stim-

ulate commercial activities Increasing quantities of machining of

the components can reduce the price drastically In the setup pre-

sented here the HPLC valve is the most expensive module If higher

y = 0.9896x R² = 0.9925

0 100 200 300 400

MS-MS: HVA (µM)

Fig 11 Comparison of urinary homovanillic acid concentrations between our HPLC-

ECD method and an established HPLC-MS-MS method A linear trend line was obtained using Excel R For two samples with HVA concentrations of 352.7 μM and 279.5 μM (measured with MS-MS) we measured average HVA concentrations of 364.3 μM and 259.0 μM, respectively; the standard deviations were 8.7 μM (n = 3) and 16.3 μM (n = 5), respectively Samples of two human volunteers were collected after taking Kalamata olives, mucuna pruriens and quercetin and without taking any- thing (after overnight fasting)

separation power is needed, peak broadening caused by the injection should be reduced Then, a valve with a small stator bore may

be needed This will be less available than e.g a second-hand wide bore valve

For detecting compounds that are not electrochemically active,

a ﬂorescence detector or a recently published prototype of a UV LED based detection cell [8] could serve The use of alternative commercially available detectors for capillary LC, also of a con- tactless conductivity detector, would add EUR 20 0 0 or more to the cost As HPLC is modular, our pump module or electrochemical detector may also be used to replace a pump or detector in another portable system In case a gradient elution will be needed one may

Trang 9

A daily dose of 5 mg of HT and derivatives has been recom-

mended by EFSA [41]for reduction of oxidation of LDL cholesterol

[42] A large variation in the reduction of oxidation of LDL choles-

terol by HT and derivates in humans has been reported [43-46]

Also the content of HT of table olives may vary widely [ 47, 48]

Comparison with the peak height of a HT standard we found for

some Kalamata table olives that just one olive would already be

enough to obtain 9 mg of HT, while we found in Thassos table

olives less than 0.2 mg HT per olive, which is in line with earlier

ﬁndings [47] This work may contribute to facilitating more mea-

surements in this applied setting thereby providing bigger data and

thereby more understanding of variables involved in the conver-

sion of HT to HVA, e.g. the way of administration and the time of

day olives are consumed

Besides nutrition as a source for HVA, urinary HVA may be

produced by neuroblastomas and phaeochromocytomas [49-51] A

further practical application of urinary HVA may be its use as a

biomarker that can be measured frequently at home, e.g during

tumor treatment We propose measurements after food interven-

tions early in the morning after overnight fasting as a good timing

with relatively low inﬂuence of other food sources

For home procedures packing of fused silica capillaries are

preferred over making monolithic columns (simpler and work-

ing with less toxic materials) Moreover, most of the polymer-

based monolithic columns seem to be unable to eﬃciently sepa-

rate small molecules [52] Chromatography of hydrophilic oxidiz-

able urinary components using reversed phase column material

has been shown here without organic modiﬁer added to the elu-

ent When analyzing more lipophilic compounds with the present

system the use of organic modiﬁers in the eluent may negatively

affect the wicking speed of the eluent alongside the working elec-

trode When the transport becomes too slow, a make-up ﬂow can

be added from the upper reservoir, analogous to the principle pub-

lished before [17] Alternatively the construction of the electro-

chemical cell may be adapted to make the ﬂow over the electrode

independent of the wicking [53]

4 Conclusions

The work presented shows the feasibility of constructing one’s

own low-cost HPLC system for urine analysis after food interven-

tions at home Key components are a hand pump, requiring no bat-

teries and capillary LC, in combination with a simple electrochemi-

cal detection system with a ﬂow generated by wicking and gravity

along a BDD electrode The present system has a low weight ( <

5 kg, excl laptop) and is potentially a portable system for mea-

surements in the ﬁeld, but its robustness will have to be improved

for that purpose, especially regarding the electrochemical detec-

tor Figures of merit also comprise its relatively easy construction,

when compared to other low-weight systems [11], its low price,

and its sensitive detection In addition, a stable eluent ﬂow re-

sulting from the presented pump can be regarded as favorable

ported systems should be considered The isocratic mode of elution and the direct injection of urine contributed to a long elution time before the next sample could be injected However, for a limited number of samples to be analyzed at home this should not con- stitute a problem Other detectors could open a new window for detecting compounds, but their acquisition could considerably increase the price of the setup

As an application we analyzed fresh urine and focused on one peak that was easily detectable and that showed a large variation during the day This peak was at the retention time of HVA and its concentration in urine was dependent on the intake of HT However, other compounds in the urine may be measured as well Other detectors and superﬁcially porous silica particles or monolithic columns may be used to broaden the scope and speed of analysis of easily measurable components without sample pretreatment Hopefully, this low-cost approach will stimulate more research at home into the physiology of oneself or of dear ones and

in the near future it may contribute to managing health

Declaration of Competing Interest

The authors declare that JL and PvdS are board members of a non-proﬁt Foundation for the promotion of the use of chromatography at home

CRediT authorship contribution statement Jan Lankelma: Investigation, Conceptualization, Methodology, Writing original draft Dirck J van Iperen: Investigation, Re- sources Paul J van der Sluis: Investigation, Writing review & editing

Acknowledgements

We are grateful to Jaap Tijmes and Arthur van de Tetrix for their help with the construction of the PEEK modules Errol Dekkinga of D-Tech Staalbouw has helped with a prototype of the pump module in stainless steel Tom Tijmes is thanked for IT support and colleagues of the VU instrumental development group are gratefully acknowledged for skillful technical advice Martijn van Faassen (Department of Laboratory Medicine, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands) is gratefully acknowledged for analysis by HPLC-MS-MS Henk Dekker

is acknowledged for advice on data management We thank colleagues in the nutrition research ﬁeld for valuable discussions and colleagues in the lab for experimental support Hans Westerhoff is gratefully acknowledged for improvements in the style and struc- ture of the manuscript

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi: 10.1016/j.chroma.2021.461925

Trang 10

References

[1] S.S Gollamudi, E.J Topol, N.E Wineinger, A framework for smartphone-

enabled, patient-generated health data analysis, PeerJ 4 (2016) e2284, doi: 10

7717/peerj.2284

[2] E.L Rossini, M.I Milani, E Carrilho, L Pezza, H.R Pezza, Simultaneous deter-

mination of renal function biomarkers in urine using a validated paper-based

microﬂuidic analytical device, Anal Chim Acta 997 (2018) 16–23, doi: 10.1016/j

aca.2017.10.018

[3] C.M Silveira, T Monteiro, M.G Almeida, Biosensing with Paper-Based Minia-

turized Printed Electrodes-A Modern Trend, Biosensors-Basel 6 (2016), doi: 10

3390/bios6040051

[4] E Carrilho, A.W Martinez, G.M Whitesides, Understanding wax printing: a

simple micropatterning process for paper-based microﬂuidics, Anal Chem 81

(2009) 7091–7095, doi: 10.1021/ac901071p

[5] A.H Emwas, C Luchinat, P Turano, L Tenori, R Roy, R.M Salek, D Ryan,

J.S Merzaban, R Kaddurah-Daouk, A.C Zeri, G.A.N Gowda, D Raftery,

Y.L Wang, L Brennan, D.S Wishart, Standardizing the experimental condi-

tions for using urine in NMR-based metabolomic studies with a particu-

lar focus on diagnostic studies: a review, Metabolomics 11 (2015) 872–894,

doi: 10.1007/s11306- 014- 0746- 7

[6] S Sharma, A Plistil, R.S Simpson, K Liu, P.B Farnsworth, S.D Stearns, M.L Lee,

Instrumentation for hand-portable liquid chromatography, J Chromatogr A

1327 (2014) 80–89, doi: 10.1016/j.chroma.2013.12.059

[7] Y Li, K Pace, P.N Nesterenko, B Paull, R Stanley, M Macka, Miniaturised

electrically actuated high pressure injection valve for portable capillary liquid

chromatography, Talanta 180 (2018) 32–35, doi: 10.1016/j.talanta.2017.11.061

[8] S.C Lam, L.J Coates, M Hemida, V Gupta, P.R Haddad, M Macka, B Paull,

Miniature and fully portable gradient capillary liquid chromatograph, Anal

Chim Acta 1101 (2020) 199–210, doi: 10.1016/j.aca.2019.12.014

[9] Y Li, M Dvorak, P.N Nesterenko, R Stanley, N Nuchtavorn, L.K Krcmova, J Au-

fartova, M Macka, Miniaturised medium pressure capillary liquid chromatog-

raphy system with ﬂexible open platform design using off-the-shelf microﬂu-

idic components, Anal Chim Acta 896 (2015) 166–176, doi: 10.1016/j.aca.2015

09.015

[10] T.S Kephart, P.K Dasgupta, J.N Alexander, An affordable high-performance

pumping system for gradient capillary liquid chromatography, J

Microcolumn Sep 11 (1999) 299–304, doi: 10.1002/(Sici)1520-667x(1999)11:

4 299::Aid- Mcs7 3.0.Co;2- 1

[11] F Rahimi, S Chatzimichail, A Saifuddin, A.J Surman, S.D Taylor-Robinson, A

Salehi-Reyhani, A Review of Portable High-Performance Liquid Chromatogra-

phy: the Future of the Field?, Chromatographia, 83 (2020) 1165-1195 DOI:

10.1007/s10337-020-03944-6

[12] S Chatzimichail, D Casey, A Salehi-Reyhani, Zero electrical power pump for

portable high-performance liquid chromatography, Analyst 144 (2019) 6207–

6213, doi: 10.1039/c9an01302d

[13] C.Y He, Z.F Zhu, C.Y Gu, J Lu, S.R Liu, Stacking open-capillary electroosmotic

pumps in series to boost the pumping pressure to drive high-performance liq-

uid chromatographic separations, Journal of Chromatography A 1227 (2012)

253–258, doi: 10.1016/j.chroma.2011.12.105

[14] K.B Lynch, A.P Chen, Y Yang, J.J Lu, S.R Liu, High-performance liquid chro-

matographic cartridge with gradient elution capability coupled with UV ab-

sorbance detector and mass spectrometer for peptide and protein analy-

sis, Journal of Separation Science 40 (2017) 2752–2758, doi: 10.1002/jssc

201700185

[15] L.J Coates, S.C Lam, A A Gooley, P.R Haddad, B Paull, H.J Wirth, Modular,

cost-effective, and portable capillary gradient liquid chromatography system

for on-site analysis, Journal of Chromatography A 1626 (2020), doi: 10.1016/j

chroma.2020.461374

[16] Y Saito, K Jinno, T Greibrokk, Capillary columns in liquid chromatography:

between conventional columns and microchips, J Sep Sci 27 (2004) 1379–1390,

doi: 10.10 02/jssc.20 0401902

[17] J Lankelma, Z Nie, E Carrilho, G.M Whitesides, Paper-based analytical device

for electrochemical ﬂow-injection analysis of glucose in urine, Anal Chem 84

(2012) 4147–4152, doi: 10.1021/ac3003648

[18] K Peckova, J Musilova, J Barek, Boron-Doped Diamond Film Electrodes-New

Tool for Voltammetric Determination of Organic Substances, Crit Rev Anal

Chem 39 (2009) 148–172, doi: 10.1080/10408340903011812

[19] A Kraft, Doped diamond: A compact review on a new, versatile electrode ma-

terial, Int J Electrochem Sc 2 (2007) 355–385 http://www.electrochemsci.org/

papers/vol2/2050355.pdf

[20] H.D Meiring, E van der Heeft, G.J ten Hove, A.P.J.M de Jong, Nanoscale

LC-MS(n): technical design and applications to peptide and protein anal-

ysis, Journal of Separation Science 25 (2002) 557–568, doi: 10.1002/

1615-9314(20020601)25:9 557::Aid-Jssc557 3.0.Co;2-F

[21] T.H Walter, P Iraneta, M Capparella, Mechanism of retention loss when C8

and C18 HPLC columns are used with highly aqueous mobile phases, J Chro-

matogr A 1075 (2005) 177–183, doi: 10.1016/j.chroma.2005.04.039

[22] Timothy P Bradshaw, Introduction to Peptide and Protein HPLC, Vol

1., Phenomenex, p 37 https://phenomenex.blob.core.windows.net/documents/

d425ff1c- 61a8- 4bfe- 9c0a- 6abe585f1d24.pdf

[23] N.M Devitt, R.E Moran, J.M Godinho, B.M Wagner, M.R Schure, Measur- ing porosities of chromatographic columns utilizing a mass-based total pore- blocking method: Superﬁcially porous particles and pore-blocking critical pressure mechanism, J Chromatogr A 1595 (2019) 117–126, doi: 10.1016/j.chroma 2019.02.045

[24] S.C.B Oliveira, A.M Oliveira-Brett, Boron doped diamond electrode pre- treatments effect on the electrochemical oxidation of dsDNA, DNA bases, nucleotides, homopolynucleotides and biomarker 8-oxoguanine, J Electroanal Chem 648 (2010) 60–66, doi: 10.1016/j.jelechem.2010.06.020

[25] J Lankelma, H Poppe, Design and Characterization of a Coulometric Detector with a Glassy Carbon Electrode for High-Performance Liquid-Chromatography,

J Chromatogr 125 (1976) 375–388, doi: 10.1016/S0 021-9673(0 0)83369-5 [26] H Hotta, H Sakamoto, S Nagano, T Osakai, Y Tsujino, Unusually large num- bers of electrons for the oxidation of polyphenolic antioxidants, Bba-Gen Sub- jects 1526 (2001) 159–167, doi: 10.1016/S0304-4165(01)00123-4

[27] J Dutrieu, Y.A Delmotte, Simultaneous Determination of Vanilmandelic Acid (Vma), Homovanillic-Acid (Hva) and 5-Hydroxy-3-Indoleacetic Acid (5-Hiaa) in Urine by High-Performance Liquid-Chromatography with Coulometric Detec- tion, Fresen Z Anal Chem 317 (1984) 124–128, doi: 10.10 07/Bf0 0594062 [28] W.P.H de Boer, J Lankelma, Two-dimensional semi-parametric alignment of chromatograms, J Chromatogr A 1345 (2014) 193–199, doi: 10.1016/j.chroma 2014.04.034

[29] M.D.M Dryden, A.R Wheeler, DStat: A Versatile, Open-Source Potentiostat for Electroanalysis and Integration, Plos One 10 (2015), doi: 10.1371/journal.pone

0140349 [30] A Prüss, C Kempter, J Gysler, T Jira, Extracolumn band broadening in capillary liquid chromatography, Journal of Chromatography A 1016 (2003) 129–141, doi: 10.1016/S0021-9673(03)01290-1

[31] A Prüss, C Kempter, J Gysler, T Jira, Evaluation of packed capillary liquid chromatography columns and comparison with conventional-size columns, Journal

of Chromatography A 1030 (2004) 167–176, doi: 10.1016/j.chroma.2003.11.017 [32] K Vanderlinden, G Desmet, K Broeckhoven, Measurement of the Band Broad- ening of UV Detectors used in Ultra-high Performance Liquid Chromatography using an On-tubing Fluorescence Detector, Chromatographia 82 (2019) 489–

498, doi: 10.1007/s10337- 018- 3622- 1 [33] M Robles-Almazan, M Pulido-Moran, J Moreno-Fernandez, C Ramirez- Tortosa, C Rodriguez-Garcia, J.L Quiles, M Ramirez-Tortosa, Hydroxytyrosol: Bioavailability, toxicity, and clinical applications, Food Res Int 105 (2018) 654–

667, doi: 10.1016/j.foodres.2017.11.053 [34] A.E Stroe, F Amin, A Hashmi, D Densmore, T Kahn, P.J Knott, Diurnal variation in plasma homovanillic acid: not a renal phenomenon, Biol Psychiatry 41 (1997) 621–623, doi: 10.1016/s0 0 06-3223(96)0 0526-4

[35] F Amin, M Davidson, K.L Davis, Homovanillic acid measurement in clinical research: a review of methodology, Schizophr Bull 18 (1992) 123–148, doi: 10 1093/schbul/18.1.123

[36] I.B Jaganath, W Mullen, C.A Edwards, A Crozier, The relative contribution of the small and large intestine to the absorption and metabolism of rutin in man, Free Radic Res 40 (2006) 1035–1046, doi: 10.1080/1071576060 077140 0 [37] S Cohrs, Z Guan, K Pohlmann, W Jordan, J Pilz, E Ruther, A Rodenbeck, Nocturnal urinary dopamine excretion is reduced in otherwise healthy sub- jects with periodic leg movements in sleep, Neurosci Lett 360 (2004) 161–164, doi: 10.1016/j.neulet.2004.02.056

[38] M Kanehisa, M Araki, S Goto, M Hattori, M Hirakawa, M Itoh, T Katayama,

S Kawashima, S Okuda, T Tokimatsu, Y Yamanishi, KEGG for linking genomes

to life and the environment, Nucleic Acids Res 36 (2008) D4 80–D4 84, doi: 10 1093/nar/gkm882

[39] G.W Lambert, G Eisenhofer, G.L Jennings, M.D Esler, Regional homovanillic acid production in humans, Life Sci 53 (1993) 63–75, doi: 10.1016/ 0024- 3205(93)90612- 7

[40] R Katzenschlager, A Evans, A Manson, P.N Patsalos, N Ratnaraj, H Watt,

L Timmermann, R Van der Giessen, A.J Lees, Mucuna pruriens in Parkinson’s disease: a double blind clinical and pharmacological study, J Neurol Neurosurg Psychiatry 75 (2004) 1672–1677, doi: 10.1136/jnnp.2003.028761

[41] Carlo Agostoni, Jean-Louis Bresson, Susan Fairweather-Tait, Albert Flynn, Ines Golly, Hannu Korhonen, Pagona Lagiou, Martinus Løvik, Rosan- gela Marchelli, Ambroise Martin, Bevan Moseley, Monika Neuhäuser-Berthold, Hildegard Przyrembel, Seppo Salminen, Yolanda Sanz, Sean (J.J.) Strain, Stephan Strobel, Inge Tetens, Daniel Tomé, Hendrik van Loveren, Hans Ver- hagen, Scientiﬁc Opinion on the substantiation of health claims related to polyphenols in olive and protection of LDL particles from oxidative damage (ID

1333, 1638, 1639, 1696, 2865), maintenance of normal blood HDL-cholesterol concentrations (ID 1639), maintenance of normal blood pressure (ID 3781),

“anti-inﬂammatory properties” (ID 1882), “contributes to the upper respira- tory tract health” (ID 3468), “can help to maintain a normal function of gas- trointestinal tract” (3779), and “contributes to body defences against external agents” (ID 3467) pursuant to Article 13(1) of Regulation (EC) No 1924/2006 1 , Efsa J 9 (2011) 2033 https://efsa- onlinelibrary- wiley- com.vu- nl.idm.oclc.org/ doi/abs/10.2903/j.efsa.2011.2033

[42] S.J Rietjens, A Bast, G.R Haenen, New insights into controversies on the antioxidant potential of the olive oil antioxidant hydroxytyrosol, J Agric Food Chem 55 (2007) 7609–7614, doi: 10.1021/jf0706934

Tiêu đề	Towards Using High-Performance Liquid Chromatography at Home
Tác giả	Jan Lankelma, Dirck J. van Iperen, Paul J. van der Sluis
Trường học	VU University Amsterdam
Chuyên ngành	Analytical Chemistry
Thể loại	journal article
Năm xuất bản	2021
Thành phố	Amsterdam

Định dạng
Số trang	11
Dung lượng	2 MB