Dual-Wavelength Recording, a Simple Algorithm to Eliminate Interferences Due to UV-Absorbing Substances in Capillary Electrophoresis

Academiejaar 2015 – 2016

Dual-wavelength recording, a simple algorithm to eliminate interferences due to UV-absorbing substances in capillary electrophoresis.

Liesbeth SEAUX

Promotor: Prof. dr. J. Delanghe

Masterproef voorgedragen in de master in de specialistische geneeskunde …………………………………

Academiejaar 2015 – 2016

Dual-wavelength recording, a simple algorithm to eliminate interferences due to UV-absorbing substances in capillary electrophoresis.

Liesbeth SEAUX

Promotor: Prof. dr. J. Delanghe

Masterproef voorgedragen in de master in de specialistische geneeskunde …………………………………

VOORWOORD

Ik wens via deze weg Prof. Dr. J. Delanghe te bedanken voor mij de kans die hij mij bood om dit artikel tot stand te brengen. Naast het geven van deze gelegenheid en de steun bij het uitvoeren van de testen en het schrijven, heeft het enthousiasme van Prof. Dr. J. Delanghe samen met de MLT’s van labo speciale scheikunde mijn interesse in deze deeldiscipline nog vergroot. Ik hoop dan ook in de toekomst mij als klinisch bioloog hier nog verder in te kunnen specialiseren.

Liesbeth Seaux 2248 Electrophoresis 2014, 35, 2248–2252

Liesbeth Seaux Research Article Soﬁe Van Houcke Els Dumoulin Tom Fiers Dual-wavelength recording, a simple Elke Lecocq Joris R. Delanghe algorithm to eliminate interferences due

Department of Clinical to UV-absorbing substances in capillary Chemistry, Microbiology and Immunology, Ghent University electrophoresis Hospital, Ghent, Belgium Analytical interferences have been described due to the presence of various exogenous UV-absorbing substances in serum. Iodine-based X-ray contrast agents and various an- Received May 26, 2014 tibiotics have been reported to interfere with interpretation of serum protein pherograms, Accepted June 16, 2014 resulting in false diagnosis of paraproteinemia. In the present study, we have explored the possibility of measuring UV absorbance at two distinct wavelengths (210 and 246 nm) to distinguish between true and false paraproteins on a Helena V8 clinical electrophoresis instrument. This study demonstrates that most substances potentially interfering with serum protein electrophoresis show UV-absorption spectra that are distinct from those of serum proteins. Scanning at 246 nm allows detection of all described interfering agents. Comparing pherograms recorded at both wavelengths (210 and 246 nm) enables to distinguish paraproteins from UV-absorbing substances. In case of a true paraprotein, the peak with an electrophoretic mobility in the gamma-region decreases, whereas the X-ray contrast media and antibiotics show an increased absorption when compared to the basic setting (210 nm). The ﬁnding of iodine-containing contrast media interfering with serum protein electrophoresis is not uncommon. In a clinical series, interference induced by contrast media was reported in 54 cases (of 13 237 analyses), corresponding with a prevalence of 0.4%. In the same series, 1631 true paraproteins (12.3%) were detected. Implemen- tation of the proposed algorithm may signiﬁcantly improve the interpretation of routine electrophoresis results. However, attention should still be paid to possible interference due to presence of atypical proteins fractions (e.g., tumor markers, C3).

Keywords: Antibiotics / Capillary electrophoresis / Iodine-containing contrast media / Paraproteins DOI10.1002/elps.201400259

1Introduction ioxitalamic acid, meglumine iotroxate, iopromide, iobitridol, iodixanol) and antibiotics [4,5] (e.g., piperacillin–tazobactam, Current guidelines for diagnosis and monitoring of mono- sulfamethoxazole–trimetoprim, and ampicilline–sulbactam) clonal gammopathy of undetermined significance (MGUS) have been reported to interfere with interpretation of serum and smoldering (asymptomatic) multiple myeloma require protein pherograms, resulting in false diagnosis of parapro- quantification of paraproteins in serum [1]. CE of serum pro- teinemia (Table 1). teins has become a standard analysis for the detection and The Helena V8 is a clinical CE instrument (Helena quantification of paraproteins (M-proteins). Most commonly Biosciences Europe, Newcastle, UK), which is equipped the quantification of electrophoretic protein fractions is based with a monochromator and therefore allows to measure on absorbance of the peptide bonds (200–210 nm). Occa- absorbance at any wavelength in the range between 200 sionally, analytical interferences have been described due to and 600 nm [6, 7]. As the interfering substances show UV the presence of various exogenous UV-absorbing substances absorbance spectra that are distinct from those of proteins in serum. Iodine-based contrast agents [2, 3] (e.g., iohexol, (Table 1), analyzing serum protein pherograms at different wavelengths could allow detection of interfering agents and facilitate pherogram interpretation. Correspondence:ProfessorJorisR.Delanghe,Departmentof Clinical Chemistry, Ghent University Hospital, De Pintelaan 185, In the present study we want to explore the possibility B-9000 Gent, Belgium of measuring absorbance at two distinct wavelengths (210 E-mail: [email protected] and 246 nm) in order to distinguish between true and false

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com ⃝ Electrophoresis 2014, 35,2248–2252 CE and CEC 2249

Ta b l e 1 . List of interfering substances lowing electrophoretic separation, polyclonal antisera against ␥, ␣, ␮ heavy Ig chains, and ␬ and ␭ light chains were used to Compound ␭ (nm) Expected max fix the specific Ig chains. The precipitated immune complexes concentration range in serum were then visualized with acid-violet stain [8]. Ig concentration (IgA, IgM, IgG) in serum was assayed Ioxitalamic acid 240 1.0–62.3 g/L immunonephelometrically on a BN II analyzer (Siemens, Iobitridol 242 2.6–84 g/L Marburg, Germany) [9] using Siemens polyclonal antibodies Iohexol 242 1.1–22 g/L (Siemens, Marburg, Germany). Iopromide 242 7.7–31 g/L Iodixanol 245 0.9–33 g/L Iomeprol 244 0.6–24 g/L 2.3 Subjects Sulfamethoxazole– 257 (sulfamethoxazole) 3.4 0.3 and ± trimetoprim 289 (trimethoprim) 46.3 2.7 mg/L ± A laboratory information-system query for all serum protein (after 1 h) electrophoresis requests between January 1, 2013, and March 31, 2014, was done. The comments of the 13 237 reports were paraproteins and to validate dual-wavelength reading of checked for presence of paraproteins, and remarks dealing serum protein electrophoresis in the clinical laboratory. with interferences due to contrast media and fibrinogen. Concomitantly, a series of serum samples containing paraproteins (n 31), IgG (n 15), IgA (n 9), IgM 2Materialsandmethods = = = (n 7) with M-protein concentrations ranging from 4.0 to = 32.7 g/L was analyzed in a similar way. 2.1 Electrophoresis

In the Helena V8 instrument, 2 ␮Lofeachserumsample 2.4 UV spectra is 50-fold diluted and is automatically injected. Separation of proteins is obtained by applying a voltage of 9 kV in eight UV-absorption spectra (wavelength range: 200–300 nm) of fused-silica capillaries controlled by Peltier effect. A ninth sera containing paraproteins, sera spiked with contrast media capillary is used as reference (buffer only). In the standard (concentration: 52 mg/L), and aqueous solutions of contrast operating mode, direct detection of proteins is performed media (concentration range: 52 to 10 mg/L) were recorded in by measuring UV absorbance of the separated fractions at a Shimadzu UV-1800 spectrophotometer (Shimadzu, Kyoto, 210 nm. Throughput is 90 samples/h. Helena V8 CE was used Japan). Due to high absorption of sera, sera samples were 40- in a standard and modified setting for serum electrophoresis. to 100-fold diluted in 0.9% NaCl. In the modified setting, the reading wavelength was changed from 210 to 246 nm, whereas all other settings of the instrument were kept identical. 3Results Pooled serum from a group of normal subjects (n = Figure 1 compares the UV-absorption spectra of a 1:100 20) was used as a standard matrix for studying ana- dilution in saline of a human serum pool (total protein lytical interferences. Pooled serum samples were spiked with iohexol (Omnipaque, final concentration in serum: 51.8 g/L), iomeprol (Iomeron, 71.44 g/L), iopromide (Ultra- vist, 76.9 g/L), ioxitalamic acid (Telebrix, 74.8 g/L), iodixanol (Visipaque, 65.2 g/L), iobitidrol (Xenetix, 76.8 g/L), and sulfamethoxazole-trimetoprime (Eusaprim, 40 g/200 g/L), respectively, and analyzed at the two wavelengths. In order to test interference caused by fibrinogen, two plasma samples containing 0.36 and 0.47 g/L fibrinogen were subjected to electrophoresis and recorded at both wavelengths. In order to investigate the detection range of the interfering substances, a dilution series of iohexol in normal pooled serum was prepared. Final iohexol concentration ranged from 52 to 52 g/L.

2.2 Immunoﬁxation—immunonephelometry

Immunoﬁxation was used for the identiﬁcation of parapro- Figure 1. UV-absorption spectrum of human serum (total protein teins. Hereby, a commercial agarose gel based electrophoresis content: 73 g/L, (A) and an aqueous solution (51.8 mg/L) of iohexol system (Sebia Hydrasys, Sebia, Evry, France) was used. Fol- (B).

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com ⃝ 2250 L. Seaux et al. Electrophoresis 2014, 35,2248–2252

Ta b l e 2 . UV absorption of plasma proteins and contrast media

Compound a (210 nm, a (246 nm, Ratio 1 1 1 1 Lg− cm− ) Lg− cm− ) A210/A246

Plasma proteins 21 0.6 35 Iohexol 26.0 34.8 0.75 a, specific absorption coefficient; A210,absorbanceat210nm; A246,absorbanceat246nm. content: 73 g/L) and an aqueous solution of iohexol (0.01%, 51.8 mg/L). Absorption maxima for plasma proteins (210 nm) strongly differ from the one for the investigated substances. Table 1 summarizes the absorbance maxima of the investigated substances, as well as their expected concentration range in serum. With the exception of sulfamethoxazole/trimethoprim (257/289 nm), all substances show UV absorption maxima between 240 and 245 nm [10–12]. Table 2 compares the specific absorption coefficients for plasma proteins and iohexol. Figure 2 compares the pherograms recorded at 210 and 246 nm of a serum of a patient who underwent contrast radiography 2 h before serum sampling. Figure 2A shows the

Figure 3. Effect of additional reading (246 nm) on the pherogram of a serum specimen containing a paraprotein. (A) Pherogram recorded at 210 nm. (B) Pherogram of the same sample, recorded at 246 nm. The monoclonal peak (arrow) is not enhanced, as compared to samples containing contrast media. The scale in the Y-axis is arbitrarily set at 100 for the highest fraction (albumin, A 35). 210/246 =

spectrum at 210 nm, Fig. 2B shows the spectrum obtained at 246 nm. When comparing results of pooled serum samples containing various amounts of iohexol, the method gave reliable results in the presence of iohexol concentrations range from 0.05 to 52 g/L. Analytical sensitivity for detecting interfering substances can be further improved by changing the sample dilution in the instrument, which enables to improve sensitivity by a factor of 4. Figure 3 compares the pherograms recorded at 210 and 246 nm of a serum specimen containing a paraprotein. All investigated paraproteins (n 31) could be detected because of = the reduction in absorbance at 246 nm. In all cases, albumin and paraprotein fractions showed a similar relative decrease Figure 2. Effect of additional reading (246 nm) on the pherogram in absorbance. The plasma samples containing ﬁbrinogen of a serum specimen containing contrast medium (Iomeron, con- showed a similar reduction in absorbance (Fig. 4). taining iomeprol) following a recent radiographic investigation. In a consecutive clinical series (January 2013–March (A) Pherogram recorded at 210 nm, with two suspected peaks in 2014) of 13 237 serum protein electrophoresis, investigated the ␤1-region (marked by the arrows). (B) Pherogram of the same in our routine laboratory, interference induced by contrast sample, recorded at 246 nm: the area under curve of the contrast media (peaks 1 and 2) is now increased compared to the strongly media was reported in 54 cases, corresponding with a preva- reduced albumin peak. The scale in the Y-axis is arbitrarily set at lence of 0.41%. No interference due to presence of antibiotics 100 for the highest fraction for each pherogram. or other drugs could be detected in this series. In the same

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com ⃝ Electrophoresis 2014, 35,2248–2252 CE and CEC 2251

centrations), a similar reduction of A210/A246 (absorbance at 210 nm/absorbance at 246 nm) will take place for albumin and the paraprotein. When an interfering substance is present, the area under the curve of the interfering substance will relatively increase as compared to albumin. The migra- tion time of the peak allows identification of the interfering compound. As in a modern hospital environment, a broad variety of technical investigations occur in a short time interval. In young and healthy subjects, the biological half-life of contrast containing media is approximately 2 h following intravenous administering [13, 14], full clearance is obtained circa 8 h after administration in patients with a normal renal function. In case of an impaired renal clearance, this time window increases. On average, patients showing the analytical interference due to contrast media were 60 10 years old. As ± paraproteinemia mainly occurs in the older age groups [15], where renal elimination rate of contrast media is often lower due to renal insufficiency [16, 17], the problem is most commonly encountered in the elderly. The finding of iodine-containing contrast media interfering with serum protein electrophoresis is not uncommon following medical imaging requiring contrast media. Our data demonstrate that for approximately every 30 serum speci- Figure 4. Effect of additional reading (246 nm) on the phero- mens presenting with a true paraprotein, there is one sample gram of a plasma specimen containing fibrinogen. (A) Pherogram showing an analytical interference due to the presence of recorded at 210 nm. (B) Pherogram of the same sample, recorded at 246 nm. All peaks are recognizable at 246 nm. The scale in the aUV-absorbingsubstance.Thepresentmethodallowsde- Y-axis is arbitrarily set at 100 for the highest fraction (albumin, tection of the interference even beyond the expected clinical A 35). 210/246 = range. Addition of activated charcoal (0.2 g to 1 mL of serum) to serum, followed by agitation of the mixture in a vortex- mixer for 20 s, and subsequent centrifugation (2000 g, series, a total number of 1631 true paraproteins cases (12.3%) × were detected. In the same series, interference due to pres- 5–10 min, 25°C) has been proposed to remove interference ence of fibrinogen was detected in 22 (0.16%) cases. caused by contrast media [10]. However, this method requires a larger sample volume and takes additional steps, which means an additional source of error. Moreover, in some cases, 4Discussion a second or a third centrifugation stage was necessary to achieve complete clarity of serum, which makes this earlier This study demonstrates that most substances potentially approach less practical than a dual-wavelength reading. The interfering with serum protein electrophoresis show UV- presented algorithm is simple and does not require any sam- absorption spectra that are quite distinct from those of serum ple pretreatment and allows an efficient detection of analytical proteins. As most interfering compounds show absorption interferences. maxima between 240 and 260 nm, the choice of 246 nm as a However, attention should still be paid to possible in- second wavelength allows detection of all described interfer- terference due to fibrinogen or presence of atypical proteins ing agents. Comparing pherograms recorded at both wave- fractions: for example, complement C3 fractions [18] or pres- lengths (210 and 246 nm) enables to distinguish paraproteins ence of huge amounts of tumor-marker glycoproteins [19] from additional peaks due to presence of other UV-absorbing may occasionally induce interpretation problems. However, substances. In case of a true paraprotein, the peak with the latter conditions are extremely rare in comparison to in- an electrophoretic mobility in the gamma region virtually terferences due to contrast media. disappears, whereas the tested contrast media and antibiotics Implementation of the algorithm based on dual- show an increased absorption when compared to the basic wavelength absorption may significantly improve the in- setting (210 nm). terpretation of routine electrophoresis results. As only se- To facilitate comparison of both pherograms, albumin lected cases with doubtful peaks in the alpha-2, beta, (the most abundant plasma protein fraction) can be used and gamma region should be verified, the proposed algo- as an internal standard. In case of incomplete disappear- rithm has only a marginal effect on the laboratory’s daily ance of the paraprotein (in case of huge paraprotein con- workload.

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com ⃝ 2252 L. Seaux et al. Electrophoresis 2014, 35,2248–2252

The authors wish to thank Mr. Robbert Ramak (Sysmex) [8] Bossuyt, X., Bogaerts, A., Schiettekatte, G., Blanckaert, for helpful discussions and Helena for providing the necessary N., Clin. Chem. 1998, 44,944–949. reagents. [9] Fink, P., Roemer, M., Haeckel, R., Fateh Mogadam, A., Delanghe, J., Gressner, A., Dubs, R., J. Clin. Chem. Clin. The authors have declared no conﬂict of interest. Biochem. 1989, 27,261–276. [10] Arranz-Pena,˜ M. L., Gonzalez-Sagrado,´ M., Olmos- Linares, A. M., Fernandez-Garc´ ıa,´ N., Martın-Gil,´ F. J., 5 References Clin. Chem. 2000, 46,736–737. [11] Grose, W. E., Bodey, G. P.,Loo, T. L., Antimicrob. Agents [1] Kyle, R. A., Durie, B. G., Rajkumar, S. V., Landgren, O., Chemother. 1979, 15,447–451. Blade, J., Merlini, G., Kroger,¨ N., Einsele, H., Vesole, D. H., Dimopoulos, M., San Miguel, J., Avet-Loiseau, H., Ha- [12] Lopez-Mart´ ınez,´ L., Lopez-de-Alba,´ P. L., de-Leon-´ jek, R., Chen, W. M., Anderson, K. C., Ludwig, H., Sonn- Rodrıguez,´ L. M., Yepez-Murrieta M. L., J. Pharm. eveld, P., Pavlovsky, S., Palumbo, A., Richardson, P. G., Biomed. Anal. 2002, 30,77–85. Barlogie, B., Greipp, P.,Vescio, R., Turesson, I., Westin, J., [13] Olsson, B., Aulie, A., Sveen, K., Andrew, E., Invest. Ra- Boccadoro, M., International Myeloma Working Group, diol. 1983, 18,177–182. Leukemia 2010, 24,1121–1127. [14] Bourin, M., Jolliet, P., Ballereau, F., Clin. Pharmacokinet. [2] Bossuyt, X., Electrophoresis 2004, 25,1485–1487. 1997, 32,180–193. [3] Vermeersch, P., Marien,¨ G., Bossuyt, X., Clin. Chem. [15] Ong, F., Hermans, J., Noordijk, E. M., Kieviet, 2006, 52,2312–2313. W. D., Wijermans, P. W., Seelen, P. J., Kluin- [4] Brouwers, A., Marien,¨ G., Bossuyt, X., Clin. Chem. Lab. Nelemans, J. C., J. Clin. Epidemiol. 1997, 50, Med. 2006, 44,910–911. 909–915. [5] Siede, D., Moller,¨ H., Siede, W. H., Regeniter, A., Clin. [16] Lorusso, V., Taroni, P., Alvino, S., Spinazzi, A., Invest. Chem. Lab. Med. 2008, 46,1468–1469. Radiol. 2001, 36,309–316. [6] Chartier, C., Boularan, A. M., Dupuy, A. M., Badiou, S., [17] Elseviers, M. M., Verpooten, G. A., De Broe, M. E., De Bargnoux, A. S., Cognot, C., Eliaou, J. F., Cristol, J. P., Backer, G. G., Lancet 1987, 21,457. Clin. Biochem. 2011, 44,1473–1479. [18] Delanghe, J. R., Speeckaert, R., Speeckaert, M. M., [7] Poisson, J., Fedoriw, Y., Henderson, M. P., Hainsworth, Pathology 2014, 46,1–10. S., Tucker, K., Uddin, Z., McCudden, C. R., Clin. Biochem. [19] Delanghe, J. R., De Buyzere, M. L., Casneuf, V., Peeters, 2012, 45,697–699. M., Clin. Chem. 2008, 54,1572–1574.

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com ⃝ NEDERLANDSE SAMENVATTING

Inleiding/achtergrond:

Bij capillaire eiwitelektroforese is interferentie door contrastmiddelen een veel voorkomend probleem (1). Hierdoor kan men verkeerdelijk denken dat patiënten een extra eiwitfractie vertonen. Dit zorgt voor bijkomende onderzoeken zoals beenmergpuncties die voor de patiënt ook pijnlijk kunnen zijn. Daarnaast zorgt deze interferentie ook voor een grotere arbeidstijd bij de klinisch bioloog die moet na gaan of de patiënt effectief beeldvorming met toediening van contrast ondergaan heeft. De Helena V8 is een capillair eiwitelektroforesetoestel dat een monochromator bevat. Hierdoor kan het de absorbantie meten op verschillende golflengtes tussen 200 en 600 nm (2).

Materialen en methode

Gepoolde serumstalen werden gespiked met verschillende middelen: iohexol, iomeprol, meglumine- ioxitalamaat, iopromide, iodixanol, allen joodhouden contrastmiddelen en met cotrimoxazole. Hierna werden deze geanalyseerd op 2 verschillende golflengtes. Met iohexol werden er verdunningen gemaakt om de gevoeligheid te bepalen. Ook werden 2 plasmastalen getest voor de interferentie met fibrinogeen en patiëntenstalen met een gekend paraproteïne.

Hiernaast werd ook in de periode januari 2013 tot eind maart 2014 in het laboratoriuminformaticasysteem gezocht naar het aantal aanvragen voor eiwitelektroforeses. Er werd naar de commentaren van deze eiwitelektroforeses gekeken voor mogelijkse interferentie door contrastmiddel of fibrinogeen.

UVabsorptiespectra van sera met paraproteïnes, gespiked met contrastmiddelen en waterige oplossingen van contrastmiddelen werden bepaald op een Shimadzu UV-1800 spectrofotometer.

Resultaten

Absorptiemaxima van serumproteïnes (210 nm) verschillen sterk van de onderzochte stoffen. Met uitzondering van cotrimoxazole (257/289 nm) hadden alle joodhoudende contrastmiddelen een absorptiemaxum rond de 240-245 nm.

Bij meting op 246 nm werd er bij aanwezigheid van joodhoudend contrastmiddel een grote piek van het contrastmiddel gezien. Bij stalen zonder joodhoudende contrastmiddelen werd op 246 nm een ferogram geobserveerd dat vergelijkbaar is met dat op 210 nm. Bij fibrinogeen werd een lichte daling van de piek geobserveerd. Fibrinogeen is haast niet te onderscheiden van een paraproteïne.

Van januari 2013 tot maart 2014 werden in het UZ Gent 13237 eiwitelektroforeses uitgevoerd. Interferentie door contrastmiddel werd in 54 (0.41%) gevallen gerapporteerd. Interferentie door andere geneesmiddelen zoals antibiotica werden niet gerapporteerd. 1631 (12.3%) monoklonale pieken werden gerapporteerd en 22 (0.16%) gevallen van interferentie door fibrinogeen.

Discussie

Deze studie toont aan dat bepalingen op verschillende golflengtes toelaat om interfererende middelen te onderscheiden van paraproteïnes. Wanneer het paraproteïne niet volledig verdwijnt kan albumine als interne standaard worden gebruikt. En de piekdaling moet dan in albumine even groot zijn als in het paraproteïne.

Joodhoudende contrastmiddelen kunnen door meting op 246 nm worden opgespoord. Hierbij moet wel nog steeds aandacht worden gevestigd op andere mogelijke interferenten zoals fibrinogeen en complement C3 fracties (3) of grote hoeveelheden glycoproteïnes van tumormarkers (4).

Aanwezigheid van interfererende substanties komt in ons ziekenhuis voor bij 1 staal op 30 stalen positief voor een paraproteïne.

Deze methode is veel minder arbeidsintensief dan andere beschreven methodes (5) om contrastmiddelen uit te zuiveren/detecteren. Gezien enkel verdachte stalen op 246 nm moeten herbepaald worden heeft deze methode weinig effect op de dagelijkse routine.

Referenties

1. Bossuyt, Xavier. "Interferences in clinical capillary zone electrophoresis of serum proteins." Electrophoresis 25.10‐11 (2004): 1485-1487.

2. Chartier, Céline, et al. "Evaluation of two automated capillary electrophoresis systems for human serum protein analysis." Clinical biochemistry 44.17 (2011): 1473-1479.

3. Delanghe, Joris R., Reinhart Speeckaert, and Marijn M. Speeckaert. "Complement C3 and its polymorphism: biological and clinical consequences." Pathology-Journal of the RCPA 46.1 (2014): 1- 10. 4. Delanghe, Joris R., et al. "Unusual serum electrophoresis pattern in a woman with pancreatic carcinoma." Clinical chemistry 54.9 (2008): 1572-1574.

5. Arranz-Peña, María Luisa, et al. "Interference of iodinated contrast media in serum capillary zone electrophoresis." Clinical chemistry 46.5 (2000): 736-737.