Rapid detection of erythropoiesis-stimulating agents in urine and serum

Analytical Biochemistry 420 (2012) 101–114

Contents lists available at SciVerse ScienceDirect

Analytical Biochemistry journal homepage: www.elsevier.com/locate/yabio

Rapid detection of erythropoiesis-stimulating agents in urine and serum Maria Lönnberg a,⇑, Maria Andrén b, Gunnar Birgegård c, Malin Drevin b, Mats Garle d, Jan Carlsson a,b a

Department of Physical and Analytical Chemistry, Uppsala University, SE-751 24 Uppsala, Sweden MAIIA Diagnostics, SE751 38 Uppsala, Sweden c Department of Hematology, Uppsala University, SE-751 24 Uppsala, Sweden d Doping Control Laboratory, Karolinska University Hospital, SE-141 86 Stockholm, Sweden b

a r t i c l e

i n f o

Article history: Received 16 February 2011 Received in revised form 24 August 2011 Accepted 17 September 2011 Available online 29 September 2011 Keywords: EPO Erythropoietin Affinity chromatography Doping control Image scanner Lateral flow immunoassay Microcolumn Protein isoforms WGA

a b s t r a c t A rapid and easy-to-use test kit, EPO WGA MAIIA, which can be used for distinguishing various endogenous human erythropoietins (hEPOs) and several recombinant hEPO and EPO analogues, has been evaluated. The test is based on chromatographic separation of the glycosylated isoforms of EPO using wheat germ agglutinin (WGA) and a sensitive immunoassay using anti-EPO carbon black nanostrings and image scanning for quantification. All of the reactions take place along the porous layer of a lateral flow microcolumn containing WGA and anti-EPO zones. The presence of molecules resembling hEPOs, such as Mircera, was detected by the aberrant affinity interaction with the antibody zone on the strip. It was possible to distinguish nine recombinant hEPOs expressed in hamster and human cell lines, as well as Aranesp and Mircera, from endogenous urine hEPO. The required amount of EPO in the samples, a few picograms, is very low compared with other methods for EPO isoform identification. This EPO isoform determination method opens the possibility to monitor recombinant EPO therapy for clinical research and seems to be a valuable candidate to the arsenal of EPO doping control tests. Ó 2011 Elsevier Inc. All rights reserved.

Erythropoietin (EPO)1 is a glycoprotein hormone responsible for the homeostatic regulation of red cell production, being up-regulated by hypoxia through a recently described oxygen-sensing mechanism [1,2]. Assays for measurement of endogenous erythropoietin were developed both as tools for studies of pathophysiology of anemia and polycythemia and for diagnostic purposes in the same conditions. Biological assays, such as the polycythemic mouse assay, were specific in the sense that they measured biologically active EPO but had low sensitivity and were expensive and time-consuming [3]. Radioimmunoassays with better sensitivity and lower cost were developed and honed into measuring only bioactive EPO [4,5]. In addition to the medically motivated need for detection and measurement of endogenous EPO, a growing need for corresponding methods

⇑ Corresponding author. E-mail address: [email protected] (M. Lönnberg). Abbreviations used: EPO, erythropoietin; ESA, erythropoiesis-stimulating agent; hEPO, human EPO; rhEPO, recombinant hEPO; pI, isoelectric point; SDS–PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis; WGA, wheat germ agglutinin; s.c., subcutaneous; WADA, World Anti-Doping Agency; GlcNAc, N-acetylglucosamine; CBNS, carbon black nano-strings; dbpp, delta blackness per pixel; PMI, percentage of migrated isoforms; RAM, relative analyte migration; AbQ, antibody line quotient; IEF, isoelectric focusing; CHO, Chinese hamster ovary; BHK, baby hamster kidney; SD, standard deviation; CV, coefficient of variation; SAR, sarcosyl; PEG, polyethylene glycol. 1

0003-2697/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2011.09.021

for exogenous erythropoiesis-stimulating agents (ESAs) has emerged as result of the use of these agents as performance-enhancing agents in endurance sports [6,7] and the health risks incurred by such use [8–10]. In this regard, the detection of ESA doping is dependent on the possibility to distinguish between endogenous EPO and exogenous ESAs in test samples from athletes. Furthermore, there is a need to distinguish between EPO produced in liver and kidneys for research in EPO physiology. In fetal life EPO is produced solely in the liver, and shortly after birth it is shifted to be produced in the kidney [11,12]. For adults, more then 90% of EPO comes from the kidney, whereas the liver produces up to 10% during normal conditions. During extreme hypoxia, the production from the liver may account for up to 50% [13]. The different endogenously produced forms of human EPO (hEPO) show microheterogeneities in their carbohydrate structures. EPO circulating in the blood during the evening contains less negatively charged isoforms than during the morning [14]. Liver-produced EPO in blood from newborn infants contains less negatively charged isoforms than EPO produced in the kidney [14]. Kidney tumors also seem to produce hEPO with the same charge as liver EPO and recombinant hEPO (rhEPO). Doping in sports using hormones and the development of EPO doping control procedures have been reviewed recently [15–17]. The isoforms of the glycoprotein hormone hEPO from various

102

Detection of erythropoiesis-stimulating agents / M. Lönnberg et al. / Anal. Biochem. 420 (2012) 101–114

production sites (e.g., human cells, Chinese hamster cells) show differences in charge [14,18] or isoelectric point (pI) [19] related to the presence of various numbers of sialic acids and sulfonate structures in the carbohydrate side chains. Minor differences in apparent molecular mass of rhEPO compared with endogenous EPO have been revealed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) [20,21]. Affinity chromatography separation, using the lectin wheat germ agglutinin (WGA), has been used to distinguish hEPO forms [22] by their varying content of polylactosamine structures [23]. The use of mass spectrometry for measuring size differences of intact or fragmented EPO, or glycan composition, seems not (so far) to have sufficient sensitivity to measure the presence of low concentrations of hEPO in biological samples [24–26]. All of these different methods require skilled technicians, several hands-on steps, and expensive equipment and also take several days to perform. There is a large number of ESA variants on the market [8,27,28] that differ in biological activity and structure (e.g., glycosylation) due to cell line expression and details in the manufacturing process such as culture conditions and purification methods. New ESAs are continuously entering the market, and this will make doping control even more difficult than it is today [29]. The recently described MAIIA technology [30] seems to be a suitable technology for improved isoform identification of glycoproteins, such as hEPO, because it can distinguish minor differences between subpopulations. The technology is flexible, and several types of ligands can be applied to distinguish subpopulations through affinity chromatography interaction [31–33]. The detection sensitivity and specificity have been shown to be high [34,35]. It seems possible to adapt the technology to high-throughput testing. This article evaluates a rapid and easy-to-use EPO isoform determination method based on the novel MAIIA technology combining WGA affinity separation and ultra-sensitive EPO immunoassay in a thin lateral flow strip. The antibody interaction profile, formed between EPO in the analyzed sample and anti-EPO antibody present on the detection zone of the strip, can also be used to reveal the presence of EPO analogues.

Materials and methods Biological samples Urine specimens from healthy volunteer donors (men and women between 20 and 60 years of age) and from patients receiving EPO therapy were collected after approval by the local ethics committee. The 14 tested patients from the hematology department each received a subcutaneous (s.c.) dose of 10,000 to 30,000 IU of epoetin or 150 lg of Aranesp at each injection, for some patients every day but for most of them once a week, as shown together with the results. Urine samples from patients with kidney failure receiving Aranesp (30 lg/every third week to 50 lg/week, n = 18) and NeoRecormon (n = 8) were also obtained, but without information about injection day and sample collection day. From participants in a major sports competition (athletics), urine samples (n = 101) were collected and an EPO doping test was performed by a World Anti-Doping Agency (WADA)-accredited anti-doping laboratory in accordance with WADA technical document TD2004EPO. Blood samples (n = 150) were also collected. For the imprecision study, 405 affinity-purified urine and plasma samples from horses and humans injected with epoetin b was used, with approximately 50% of the samples containing rhEPO in excess. These studies had approval from the local ethics committee. The serum and plasma samples used were leftovers from a health control. The specimens were stored in aliquots at 20 °C.

Recombinant EPO Eprex, recombinant epoetin a, was purchased from Janssen–Cilag (Sollentuna, Sweden). NeoRecormon, recombinant epoetin b, and Mircera, methoxy polyethylene glycol-epoetin b, were obtained from Roche (Mannheim, Germany). Aranesp, the recombinant EPO analogue darbepoetin a, was purchased from Amgen (Thousands Oak, CA, USA). For Eprex and NeoRecormon, 1 IU of epoetin corresponds to 8.4 and 8.3 ng of recombinant EPO, respectively. Retacrit, recombinant epoetin f, was purchased from Hospira Enterprises (Hoofddorp, Netherlands), and Dynepo, recombinant epoetin d, was obtained from Shire Pharmaceuticals (Basingstoke, UK). Epomax, recombinant epoetin x (Lek Pharmaceutical and Chemical Company, Ljubljana, Slovenia), was obtained from Slovenia, and EPIAO (3SBio, Shenyang Sunshine Pharmaceutical, Shenyang, China) was obtained from Jia Lin Hao(Shandong E-Hua, Shandong, China), Ji Mai Xin (NCPC GencTech Biotechnology, Shijiazhuang, China), and Ning Hong Xin (Huaxin Pharm, Leshan, China). Affinity purification of EPO from biological fluids An EPO Purification Kit (cat. no. 0250) was obtained from MAIIA Diagnostics (Uppsala, Sweden, www.maiiadiagnostics.com), and EPO from urine and serum samples was purified according to the instructions from the producer using the recommended addition of bovine serum albumin (BSA) and detergent to the reagents. The urine samples (5–33 ml) collected from athletes during a major sports competition were purified with a prototype to this kit using the same procedure but with a slightly wider anti-EPO monolith, £13 mm. The recovery for urine specimens from the competition study, exceeding an EPO concentration of 1.5 ng/L in urine, was 49%, calculated after measuring the EPO concentration in urine and in eluate. Basic laboratory procedure for EPO WGA MAIIA An EPO WGA MAIIA prototype kit was obtained from MAIIA Diagnostics and used in accordance with the instructions from the supplier. Fig. 1 describes the test principle. 25 ll sample of diluted affinity purified EPO was dispensed into four microtiter wells for the duplicate determination. Other wells were prepared by dispensing 25 ll each of elution buffer containing low (optimized) and high (100 mM) concentrations of N-acetylglucosamine (GlcNAc), anti-EPO-CBNS (carbon black nano-strings), and washing buffer. After mounting and drying the strips, the intensity in the anti-EPO zone of the strip was detected by an Epson Expression 1680 scanner (Epson, Sollentuna, Sweden) used in reflection mode with an optical resolution of 600 ppi and 16-bit sample depth in accordance with the scanner detection instructions (MAIIA Diagnostics). MAIIAcalc software calculated the intensity signal in the anti-EPO zone as delta blackness per pixel (dbpp). The principle for detection of grayscale intensity for the carbon black line has been described previously [34]. Standardization and concentration determination A standard curve of epoetin b (NeoRecormon) was obtained by dilution from the provided stock solution of 10 lg/L to 3 to 1000 ng/L using the sample dilution buffer. The standard series was measured by the EPO WGA MAIIA method using elution buffer high, which releases all EPO for migration to the anti-EPO zone. The EPO concentration of unknown samples was obtained from their elution buffer high values. The concentration values for unknown samples using high and low elution buffer in EPO WGA MAIIA was calculated by using a four-parameter logistic curve fit program (WorkOut 2, PerkinElmer, Turku, Finland) for the signal intensity

103

Detection of erythropoiesis-stimulating agents / M. Lönnberg et al. / Anal. Biochem. 420 (2012) 101–114

from the strips used for the standard dilution series. During some conditions (e.g., using the same scanner and reagent lot and

standardized temperature and humidity during assay), the use of stored signal values for the standard curve have been found to work well for calculations of the EPO concentration and, thus, have been used to obtain several of the results presented in this study. The EPO immunoassay concentration estimation is illustrated in Fig. 2, showing an rhEPO b standard curve dilution series with the signal (dbpp) obtained after test. The dbpp values, when using total desorption mode for the EPO WGA MAIIA kit, were in good agreement with the values obtained when replacing the EPO WGA MAIIA strips with anti-EPO strips produced for the EPO Quantification Kit (without the WGA zone), indicating that total desorption of EPO from the WGA zone was obtained. For EPO WGA MAIIA, using NeoRecormon as standard, the unit ng hEPO/L was established from a conversion factor of 1 U = 8.3 ng, as noted by the supplier. Optimizing concentration of GlcNAc in elution buffer low

1

2

3

5

Desorption rhEPO

6

Desorption endogenous EPO

By using a dilution series of GlcNAc from the elution buffer high (100 mM GlcNAc) (e.g., 3, 5, 7, or 10 mM) together with 100 mM in elution buffer low, the optimal resolution between values for different EPO subpopulations can be found. The selected GlcNAc concentration in elution buffer low is then used for the actual test series. Epoetin b, NeoRecormon, should always be used as a reference.

60

Calculation of EPO WGA MAIIA results The EPO WGA MAIIA value, with the unit percentage of migrated isoforms (PMI), was established by calculating the percentage ratio of the apparent concentration values for the strips used with elution buffer low and high. Only concentration values between 3 and 500 ng/L (inside the measuring range) were used. The EPO concentration in eluates was calculated from the strip used with elution buffer high. The EPO concentration in the biological sample was estimated by calculating with the EPO recovery of the affinity purification and the applied sample volume.

50

mm

Flow

70

Sample application

4

30

40

Absorbent sink

20

Capture zone with antiEPO

Cutting line

100000 10

MAIIA strip

Separation zone Start

R

T

R

T

Fig.1. The EPO WGA MAIIA procedure is shown. The top panel illustrates the complete procedure with the preceding affinity purification using disposable antiEPO columns (1), and the eluate is dispensed in the first well. The strip is moved, by hand, from well to well reacting with affinity-purified sample (2), elution buffer low or high giving retarded and total desorption (3), anti-EPO CBNS (5), and washing buffer (6). A cutting step (4) for removing the WGA zone prior to the immunoassay detection step is required. When using 5 min of incubation for each step, 56 single strips can be handled with a total testing time of 20 min. Scanner detection of the grayscale intensity in the combined capture and detection zone is finally performed, and after concentration determination of EPO, the percentage of EPO that has passed the WGA zone and reached the anti-EPO zone during retarded conditions is calculated and used for identification of the EPO subpopulation. The bottom panel illustrates the positions at the MAIIA microcolumn strip of the 8-mm-long separation zone with immobilized WGA lectin and the 1-mm capturing zone with anti-EPO and the interactions during steps 2 and 3 for rhEPO and endogenous EPO. In the sample application step, the majority of hEPO isoforms are retained in the initial part of the WGA zone. During the desorption or elution phase, the selected low concentration of the WGA binding competing sugar GlcNAc in the well (3) gives a retarded (R) migration rate for EPO, different for rhEPO and endogenous EPO, illustrated here with 25% and 75%, respectively, passing the cutting line. For every sample, a second strip is used, for which the well (3) contains a high GlcNAc concentration so that all EPO pass the lectin zone without interaction, and the total (T) amount of EPO reaches the capturing zone. The percentage of migrated isoforms (PMI) is calculated by comparing the obtained amount of EPO that has reached the anti-EPO capture zone for the strips performing retarded and total desorption of EPO.

Delta blackness per pixel ± 1SD

0

Lateral flow IA strip

10000

1000

100 1

10

100

1000

EPO (ng/L) Fig.2. In the EPO WGA MAIIA test, NeoRecormon has been used as standard for calculating the hEPO concentration in samples as well as the amount of EPO migrating through the WGA zone during retarded conditions. Concentration determination is possible when total desorption of EPO is obtained from the WGA zone by using elution buffer high, allowing all forms of hEPO to pass the WGA zone without retardation. The total release of hEPO was confirmed by showing that the same signal levels were obtained by replacing the WGA MAIIA strip with the corresponding anti-EPO lateral flow immunoassay (IA) strip, which has no WGA zone. The recommended measuring range for quantitative determination, using 25 ll of sample, is between 3 and 500 ng/L hEPO.

104

Detection of erythropoiesis-stimulating agents / M. Lönnberg et al. / Anal. Biochem. 420 (2012) 101–114

Calculation of EPO AbQ MAIIA results

Statistics

The dbpp values obtained over the anti-EPO zone from a strip after the EPO WGA MAIIA assay can be shown by the software MAIIAcalc for positions between upstream (0.21 to 0.04 mm) and downstream (0.04 to 0.51 mm), with the maximum intensity (peak) at position 0 mm. The concentration for samples was estimated against the NeoRecormon standard dilution series using the peak values, whose intensity is also used for EPO WGA MAIIA calculations. The same procedure was performed with the dbpp values obtained at a second position downstream from the peak. The unit relative analyte migration (RAM) for a sample was obtained by calculating the ratio between concentration at the second position (e.g., 0.38 mm) and concentration at the peak position (0 mm). The second position is indicated in the unit (e.g., RAM0.38). EPO-related molecules, such as Mircera, have less interaction with the antibody, resulting in it is not being as rapidly captured as rhEPO and continuing to migrate along the antibody zone. The concentration, relative to rhEPO, in the second position is higher for Mircera (for example), which thereby gets a higher RAM value than rhEPO. The principle for AbQ (antibody line quotient) calculation, using differences in antibody interaction in the anti-EPO capturing and detection zone on the EPO WGA MAIIA strip, is also illustrated together with the results. It is essential to use concentration determination at each position because the dbpp intensity profile is different at low and high signal levels. Calculation of RAM values was performed on the MAIIA strips used with elution buffer high but can also be used with elution buffer low.

Differences between groups were examined by Student’s t test, and statistical significance was accepted at P < 0.05.

Measurement of EPO concentration in urine using a lateral flow immunoassay The thawed urines were gently turned end-over-end to distribute precipitates evenly, and an aliquot was transferred to another tube together with UPD buffer (MAIIA Diagnostics) using 9 parts of urine and 1 part of UPD buffer. The urine precipitates were instantly dissolved, and 0.5 ml of the obtained solution was desalted on a Nap 5 column (GE Healthcare, Uppsala, Sweden) by elution with 1 ml of 20 mM Tris (pH 7.5) buffer supplemented with 75 mM NaCl, 0.1% Tween 20, and 0.02% NaN3. The recovery in the desalting step was calculated by also measuring the urine specimen with a known amount of NeoRecormon added. The EPO concentration in the desalted urine mixtures was estimated by using 200 ll of sample in the EPO Quantification Kit (cat. no. 0100, MAIIA Diagnostics) and used in accordance with the instructions from the supplier. For the samples from the major sports competition series with a concentration value below 1.5 ng/L in urine, the EPO concentration was estimated by measuring the EPO amount of the affinity-purified eluate using the recovery figure (49%) obtained for the samples above 1.5 ng/L EPO in urine. Isoelectric focusing A WADA-accredited anti-doping laboratory performed isoelectric focusing (IEF) in accordance with the TD2004EPO rules. For detection of EPO blotted to the membrane, a monoclonal mouse antibody (clone AE7A5, R&D Systems, Oxford, UK) was used. This antibody was blotted to a second membrane and detected by the use of a goat anti-mouse immunoglobulin G (IgG)-biotin antibody (Sigma–Aldrich, St. Louis, MO, USA) combined with a streptavidin/ horseradish peroxidase complex (BioSpa, Milano, Italy) and a chemiluminescent substrate (CPS 160, Sigma–Aldrich). An image of the signal intensity was obtained by a charge-coupled device (CCD) camera (Fuji LAS-1000).

Results Improving resolution by regulating WGA interaction The resolution, the ability to differentiate between normal endogenous human urinary EPO and exogenous EPO, relies on establishment of optimal conditions to obtain different migration rates for EPO forms along the WGA zone on the strip. The WGA ligand density, length of the zone, flow rate, and flow time affect the final position of EPO on the strip. The selected desorption conditions (e.g., concentration of competing sugar derivative GlcNAc) is the most convenient way to regulate the resolution. Fig. 3 shows that varying the concentration of GlcNAc in the elution buffer can regulate the resolution between two subpopulations. The best resolution in this system between endogenous urinary EPO and rhEPO b was found when using 7 to 10 mM GlcNAc, whereas endogenous urinary EPO and Mircera was best resolved using 3 mM GlcNAc. When total desorption with 100 mM GlcNAc was carried out, no difference in migration time was found between the EPO subpopulations. The WGA interaction was strongest for Aranesp, strong with NeoRecormon, weaker with endogenous urinary EPO, and weakest with Mircera. To generate a figure for the interaction strength of the different EPOs with the WGA zone, which is independent of elution conditions and ligand density, the PMI value for a standardized EPO population such as epoetin b (NeoRecormon) can be used in conjunction with the PMI value PMIbxx. With this approach, the EPO WGA MAIIA results shown in Fig. 3, using 7 mM GlcNAc, will be presented as 11, 24, 65, and 81 PMIb24 for Aranesp, NeoRecormon, normal endogenous EPO, and Mircera, respectively. When using 3 mM GlcNAc, the values will be 6, 13, 19, and 71 PMIb13 for the same respective EPO subpopulations. If one wants to differentiate between endogenous urinary-produced EPO and rhEPO, PMIb20–40 seems to be optimal. Recombinant epoetins and EPO analogues Nine epoetins and darbepoetin a were tested with EPO WGA MAIIA. The recombinant proteins were affinity purified after addition to buffer and to urine or were applied without affinity purification. No significant difference was obtained as a result of the affinity purification, and the mean PMI value (n = 3) was calculated for each epoetin and compared with the values obtained for endogenous EPO from urine specimens. Epoetins produced in Chinese hamster ovary (CHO), baby hamster kidney (BHK, epoetin x), and human cell lines (epoetin d, Dynepo) all were highly significantly (P < 0.001) distinguished from the 31 tested urine specimens from a healthy non-athlete population having mean ± standard deviation (SD) values of 73.3 ± 7.6 PMIb30, as shown in Table 1. All epoetins and darbepoetin a were estimated as aberrant from endogenous EPO when using a cutoff at 97.8% confidence limit below the mean endogenous value, and all epoetins except one were estimated as aberrant from endogenous EPO when using a cutoff at 99.9% confidence limit. The EPO-like preparation darbepoetin a (Aranesp) was even better distinguished from the endogenous population than epoetins. Mircera, the EPO-like molecule with a PEG substituent, was not identified in the EPO WGA MAIIA method using PMIb30. But by using PMIb13, as shown in Fig. 3 (or even better by its aberrant interaction profile with the antibodies in the detection

105

Detection of erythropoiesis-stimulating agents / M. Lönnberg et al. / Anal. Biochem. 420 (2012) 101–114

B

A

PMIβ38

PMIβ24

PMIβ24 PMIβ18 PMIβ13

NeoRecormon (epoetin β) (PMI) Fig.3. Panel A shows how the resolution between the migration distances for the EPO subpopulations can be optimized by using different concentrations of the competing sugar derivative GlcNAc in the WGA elution buffer. It is obvious that the best resolution between affinity-purified normal endogenous urinary EPO and rhEPO is obtained with 7 to 10 mM GlcNAc for the EPO WGA MAIIA strip used. If Mircera and endogenous urinary EPO are to be distinguished, 3 mM GlcNAc will be the optimal concentration. To have a measure of the interaction strength of the WGA zone, the PMI value for epoetin b, NeoRecormon (seen on the x axis in panel B), is used in conjunction with the PMI value for unknown samples. The values for endogenous urinary EPO can then be presented as both 65 PMIb24 and 19 PMIb13.

Table 1 EPO WGA MAIIA values for nine different epoetins and the EPO analogue darbepoetin a (Aranesp). Reference

EPO WGA MAIIA

Classification

PMIb30

SD

Urine hEPO (n = 31)

Endogenous EPO from healthy individuals

73.3

7.6

ESA (n = 3) Epoetin Epoetin Epoetin Epoetin b Epoetin x Epoetin d Epoetin Epoetin a Epoetin f Darbepoetin a CERA

EPIAO Ning Hong Xin Ji Mai Xin NeoRecormon Epomax Dynepo Jia Lin Hao Eprex Retacrit Aranesp Mircera

51.7 46.9 36.0 30.0 29.8 29.6 26.3 25.1 20.4 7.7 76.7

8.0 3.9 2.3 6.3 6.1 2.2 6.8 1.8 3.9 2.8 1.3

Aberrant Aberrant Aberrant Aberrant Aberrant Aberrant Aberrant Aberrant Aberrant Aberrant Normal

Note: The values were clearly distinguishable (P < 0.001) from affinity-purified endogenous urinary EPO found in 31 urine samples from healthy individuals. Mircera was not identified using EPO WGA MAIIA with a WGA interaction strength of PMIb30. Instead, the interaction with the antibody zone on the same strip, obtained from the AbQ calculation, can be used for identification, as shown in Figs. 7 and 8.

zone, as obtained with the AbQ calculation), the presence of this molecule can be identified (see also the section about Mircera below). Different forms of endogenous EPO It seems that a distinct population of EPO forms, besides normal endogenous and exogenous rhEPO, can be found in human urine samples. Using a GlcNAc concentration that is lower than the one used for distinguishing rhEPO revealed the presence of this EPO

population. Fig. 4 shows the PMI results for affinity-purified EPO from healthy individuals in urine (n = 7) and in serum (n = 5) compared with the results obtained for rhEPO b (n = 3). It was found that EPO in urine from several ESA-treated patients showed weak interaction with WGA, as illustrated in the figure with the results from 1 patient receiving rhEPO b, with a total EPO concentration in urine of 8.5 ng/L. For the patients who received ESAs due to kidney dysfunction, 16 urine samples were tested with EPO WGA MAIIA with a lower GlcNAc concentration and 6 of them (38%) showed an aberrantly weak WGA interaction in the range of 59.2

106

Detection of erythropoiesis-stimulating agents / M. Lönnberg et al. / Anal. Biochem. 420 (2012) 101–114

to 82.7 PMIb18. The 25 urine samples from healthy individuals showed a mean value of 39.7 ± 8.7 PMIb18. Samples from ESA-treated patients The EPO isoform distribution was analyzed with the EPO WGA MAIIA test kit after affinity purification of 5-ml urine samples from 25 healthy volunteer donors and 14 anemic hematology department patients treated with s.c. injections of different types of ESA. The results are found in Table 2, showing both EPO concentration and PMI value. The mean concentration of EPO in urine from the healthy individuals was 22.7 ± 27.2 ng/L, whereas the concentrations varied between 5 and 798 ng/L in the urine samples from 13 of the patients and was as high as 18.000 ng/L for 1 patient. The urine specimens from the healthy donors showed a mean EPO WGA MAIIA value of 73.7 ± 6.9 PMIb26. The prepared controls, NeoRecormon in buffer, and a mixture of 60% endogenous EPO and 40% NeoRecormon were also tested and showed values of 26.2 ± 2.4 and 58.2 ± 4.3 PMIb26, respectively, at four different test occasions. The presence of epoetin a and b was clearly distinguished (P < 0.001) even up to 7 days after the last injection in urine from the patients receiving s.c. injections. For 11 of the patients, the values were between 30.5 and 43.8 PMIb26. These values are highly significantly below the mean value for EPO from healthy individuals and show that the urine samples contain recombinant EPO. One patient received the last dose of NeoRecormon 9 days prior to the sample collection day, and the sample (U178) was found to have normal endogenous EPO. The patient sample (U128) with an extraordinarily high EPO concentration in the urine, 18,000 ng/L, had a concentration too high to be explained by the injection with a normal dose of NeoRecormon given 3 days before the sample collection. Sample U175 with a MAIIA value of 91.3 PMIb26 contained EPO with only weak binding to WGA and, thus, had an EPO population aberrant from both endogenous

Fig.4. Several distinct EPO subpopulations seem to be found in human samples, as indicated with EPO WGA MAIIA. Instead of using an interaction strength of PMIb35, optimal for distinguishing rhEPO from normal endogenous EPO, conditions with stronger interaction are required to distinguish the endogenous subpopulations. The interaction PMIb15 resolves three different endogenous subpopulations: normal urine hEPO, normal serum hEPO, and aberrant endogenous EPO in urine. The aberrant EPO subpopulation was found in several urine samples from ESA-treated patients.

Table 2 EPO WGA MAIIA values for EPO in urine from 25 healthy individuals and from patients receiving ESAs. Reference

urinary EPO ng/L

EPO WGA MAIIA PMIb26

22.7 27.2

73.7 6.9

Urine Endogenous urinary EPO from healthy individuals, n = 25

Mean SD

Controls –

NeoRecormon in buffer, n = 4

–

Mixture of 60 % endogenous urinary EPO and 40% NeoRecormon, n = 4

Mean SD Mean SD

26.2 2.4 58.2 4.3

Patient sample

Drug

s.c. dose

Days after last injection

urinary EPO ng/L

EPO WGA MAIIA PMIb26

U092 U126 U133 U134 U179 U128 U132 U159 U055 U158 U094 U157 U178 U175

NeoRecormon NeoRecormon Aranesp NeoRecormon Aranesp NeoRecormon NeoRecormon NeoRecormon NeoRecormon NeoRecormon Eprex NeoRecormon NeoRecormon NeoRecormon

20,000 lU/day 30,000 lU/week 150 ug/week 30,000 lU/week 150 ug/week 30,000 lU/week 30,000 lU/week 30,000 lU/week 30,000 lU/week 30,000 lU/week 10,000 lU/week 30,000 lU/week 30,000 lU/week 10,000 lU/2nd week

0.25 1 2 2 2 3 4 4 5 5 7 7 9 17

208 310 127 23 196 18,000 24 50 63 264 798 10 33 5

35.3 37.2 30.5 36.1 35.1 67.6 40.0 33.9 35.7 43.8 29.3 35.0 67.2 91.3

Note: The mean ± SD for the healthy individuals was 73.7 ± 6.9 PMIb26. A total of 11 patients receiving therapy with the last injection up to 7 days before sample collection could clearly be distinguished (P < 0.0001) from the healthy population. Two urine samples, U128 and U178, were found to have EPO isoform distribution in accordance with endogenous EPO. Sample U128 had an extremely high total EPO concentration, 18,000 ng/L, indicating very high endogenous EPO production. Sample U178 was collected 9 days after the last injection, and the concentration of rhEPO is probably much lower than the concentration of endogenous EPO due to the rapid EPO clearance of rhEPO. Sample U175 shows a PMI value higher than normal, indicating that this sample from the hematology department also contains the aberrant EPO found in 38% of the tested urine samples from patients treated with ESAs due to kidney failure.

107

Detection of erythropoiesis-stimulating agents / M. Lönnberg et al. / Anal. Biochem. 420 (2012) 101–114

urine samples. EPO in this sample, with a total concentration of 5 ng EPO/L in urine, shows the same type of WGA interaction as EPO from the 38% of urine samples tested from anemic patients receiving ESAs due to kidney dysfunction.

80

EPO WGA MAIIA (PMI)

70

Samples containing mixtures of EPO forms

60

Affinity-purified endogenous urinary EPO from 1 healthy individual and rhEPO were mixed in different proportions and analyzed at 15 different occasions with EPO WGA MAIIA. Endogenous urine EPO, a mixture of 40% rhEPO and 60% endogenous EPO, and a mixture of 88% rhEPO and 12% endogenous EPO showed means of 69.8 ± 4.2, 53.6 ± 3.3, and 30.9 ± 2.7 PMI, respectively. Fig. 5 shows that a sample containing only 40% rhEPO can significantly be distinguished from endogenous EPO. Samples with such mixtures might be useful for standardization of the PMI values.

50

40

30

Identification of Mircera using WGA and antibody (AbQ) interaction

20 -20 -10

0

10

20

30

40

50

60

70

80

90

100

% Epoetin β in mixture with endogenous urinary EPO Fig.5. EPO WGA MAIIA could significantly (P < 0.0001) distinguish rhEPO b in samples containing down to 40% rhEPO and 60% endogenous EPO from samples containing only endogenous EPO. The PMI value determination can, if required, be standardized by including series of samples with different proportions of rhEPO and endogenous EPO in the test. Such standardization can equalize migration time differences due to the influence of temperature and reduced WGA activity during storage of the MAIIA strip. Error bars show 1 SD.

EPO and rhEPO. The aberrantly weak binding to WGA was confirmed by using EPO WGA MAIIA, with the resolution PMIb15 showing a PMI value highly significantly above the values for normal

The EPO analogue Mircera is composed of epoetin b linked to a single methoxy polyethylene glycol polymer of 30 kDa, resulting in a molecule with 60 kDa molecular mass. The binding affinity of the Mircera epoetin moiety to WGA and to anti-EPO antibody is much weaker than that of free epoetin b. Both interactions can be used for identification of Mircera. Fig. 6A shows the dbpp values for Mircera, Aranesp, and NeoRecormon at different positions along the anti-EPO capturing and detection zone on the strip. Fig. 6B shows the RAM results for different groups of affinity-purified samples when calculated on several second positions. The mean RAM0.38 values shown were 1.36 ± 0.12 RAM0.38 for EPO (n = 13), 2.25 ± 0.24 RAM0.38 for Aranesp (n = 2), and 2.84 ± 0.35 RAM0.38 for Mircera (n = 12). When using a cutoff at 99.99% confidence

Mircera in urine and serum Mircera

Aranesp in serum

Aranesp

NeoRecormon in serum

NeoRecormon

Endogenous EPO from urine and serum

B

5.5

10000

5.0

9000

4.5

8000

4.0

EPO AbQ MAIIA (RAM)

Immunoassay signal (dbpp)

A 11000

7000 6000 5000

4000

3.5 3.0 2.5 2.0

3000

1.5

2000

1.0

1000

0.5

0 0

0.1 0.2 0.3 0.4 0.5 Measurement position downstream peak (mm)

0.6

0.0 0.15

0.2 0.25 0.3 0.35 0.4 0.45 0.5 Measurement position downstream peak (mm)

0.55

Fig.6. The EPO AbQ MAIIA calculation uses the differences found between EPO and the analogues Mircera and Aranesp in their interaction with the anti-EPO antibody zone when using the EPO WGA MAIIA test. Panel A shows part of the immunoassay signal for recombinant EPO and analogues applied to buffer along the 1-mm anti-EPO detection zone, where 0 is the peak position. Panel B shows the RAM values using the AbQ calculation on the ratio between concentrations obtained at the second downstream positions and at the peak position. The samples were affinity-purified endogenous EPO from urine and serum (n = 6 + 4), NeoRecormon applied to serum (n = 3), Aranesp applied to serum (n = 2), and Mircera applied to urine and serum (n = 7 + 5). The 99.9% confidence limit for the endogenous EPO population, illustrated in the figure on each second measuring point, shows that Aranesp and Mircera can be distinguished from EPO using every second measurement position. The optimal second measuring point at position 0.38 mm was selected to achieve good resolution but without reducing the immunoassay signal too much.

108

Detection of erythropoiesis-stimulating agents / M. Lönnberg et al. / Anal. Biochem. 420 (2012) 101–114

Competition samples

rhEPO added to urine Urine from healthy individuals Urine from patients with kidney dysfunction Mircera applied to buffer Mircera applied to urine Mircera applied to serum 6.0

EPO AbQ MAIIA (RAM)

5.0

4.0

3.0 mean + 2SD 2.0

1.0

0.0 0

20

40 60 EPO WGA MAIIA (PMIβ14)

80

Fig.7. The results from two methods to differentiate Mircera from endogenous EPO, both using the EPO WGA MAIIA strip, are shown. The presence of a large polyethylene glycol (PEG) substituent conjugated to epoetin b in the EPO analogue Mircera seems to disturb the binding of the epoetin carbohydrate structures to WGA as well as the interaction with the anti-EPO antibody. EPO from urine and Mircera added to buffer and urine were affinity purified and applied to EPO WGA MAIIA. The x axis shows the results for EPO WGA MAIIA PMIb14, during which conditions Mircera shows a significantly weaker interaction with WGA than endogenous EPO, whereas the resolution between rhEPO and endogenous EPO is not optimal. The y axis shows the AbQ calculation for EPO WGA MAIIA PMIb100, the total desorption mode, using the 0.38-mm downstream peak as the second measurement position. In both test systems, all Mircera samples show values outside the 99.99% confidence limit. The results from urine samples from some patients with kidney dysfunction show a WGA interaction comparable to Mircera, but the AbQ calculation reveals that this is not Mircera; these samples contain an aberrant endogenous form of urine EPO.

limit, all samples containing Aranesp and Mircera were distinguished from EPO samples. Fig. 7 shows the results for the two optional methods available for distinguishing Mircera from endogenous EPO. With EPO WGA MAIIA, the samples showed mean values of 30.2 ± 6.1 and 59.6 ± 2.6 PMIb14 for affinity-purified endogenous urine EPO (n = 9) and Mircera added to buffer and urine (n = 7), respectively. The same samples were 1.22 ± 0.26 and 3.99 ± 0.57 RAM0.38, respectively, using the AbQ calculation from the run with total desorption mode. The test concentration ranges for the diluted affinity-purified samples were 12 to 122 ng/L for the samples above 40 PMI and 41 to 109 ng/L for the normal endogenous samples between 20 and 40 PMI, showing that the PMI value is independent of the EPO concentration. The RAM value was also independent of concentration; for Mircera samples the test concentration range was 86 to 122 ng/L, whereas for EPO samples the range was 21 to 227 ng/L. In both test systems, all Mircera samples showed values outside the 99.99% confidence limit. The aberrant EPO population, different from normal endogenous EPO, rhEPO, and Aranesp found in some urine from ESA-treated patients, showed the same weak interaction to WGA as Mircera but was distinguished by the EPO AbQ MAIIA values.

Samples from a major sports competition were analyzed with EPO WGA MAIIA after the ordinary IEF-based doping analysis, and the results can be found in Fig. 8, both the PMI value and the estimated EPO concentration in urine. For MAIIA one sample (1%) was undetectable, whereas for IEF 28% of the 101 urine samples did not show bands with sufficient intensity for further densitometric evaluation. No positive sample was found with the IEF method using the TD2004EPO regulations. The 22 urine specimens from healthy non-athlete individuals showed mean values of 84.7 ± 7.2 PMIb32. Of the 100 detectable samples, there were 8 MAIIA values below the 99% limit and 6 samples below the 99.9% confidence limit with values in the rhEPO range. Of these samples, 5 were from female athletes and their blood samples showed hemoglobin values near or above 16 g/dl, even up to 18.6 g/dl. All of the urine samples with MAIIA values in the rhEPO range had an EPO concentration below 3 ng/L. Correct measurement of PMI values for very low EPO concentrations was confirmed by diluting the samples from normal individuals and measurement in an additional run. The EPO concentrations in the urine specimens showed that samples from athletes had a different distribution of EPO concentrations compared with samples from non-athlete healthy individuals. Of the 101 athlete specimens, 31% contained EPO below 3 ng/ L, whereas only 7.5% of 67 specimens from non-athletes were in that range. In addition, 85% of urine specimens from 13 athletes with a hemoglobin concentration above 16 g/L had an EPO concentration below 3 ng/L, indicating that this is an interesting concentration range for doping control. The 8 samples with rhEPO corresponding EPO WGA MAIIA values had also aberrant AbQ values. Some of the athlete urine specimens showed high PMIb32 values, and 25 samples were selected for further testing with EPO WGA MAIIA using a resolution optimal for detecting weak interacting EPO forms. The 25 urine samples from healthy non-athlete individuals showed a mean value of 39.7 ± 8.7 PMIb18. The results for 2 of the athlete samples were above the 99.9% confidence limit with EPO WGA MAIIA. With the IEF method, these samples showed the highest intensity in the basic bands 4 and 5. The results for 2 more samples were above the 99.7% confidence limit, with 1 sample showing the highest intensity for the basic bands 3 and 4 in the IEF method and the other sample not being detectable. Imprecision The immunoassay measurement of affinity-purified urine and plasma using low and high desorption mode showed mean coefficients of variation (CVs) of 6.0% (n = 405, mean = 46 ng/L) and 5.4% (n = 404, mean = 95 ng/L), respectively, between the duplicates. The figure for immunoassay imprecision was in the same range as for Immulite 2000 EPO, using an automated random access analyzer, showing a mean within-run CV of 4.6% [36]. For the PMI ratio calculation, the intraassay CV was 6.7% as an average when tested at 13 different test occasions for 157 controls between 30 and 80 PMI. The mean interassay CV was 7.0% for the controls at 70, 54, and 31 PMI when measured in duplicate at 15 different test occasions. EPO WGA MAIIA: Three different results from two strips Table 3 shows the three different types of results obtained from a typical EPO WGA MAIIA test. The MAIIA technology offers two affinity interactions on the same strip, WGA MAIIA on the WGA zone and the AbQ MAIIA on the anti-EPO zone, which together can reveal differences between EPO and EPO-like subpopulations.

109

Detection of erythropoiesis-stimulating agents / M. Lönnberg et al. / Anal. Biochem. 420 (2012) 101–114

Athlete samples, men or women

Athlete samples, men

Athlete samples, women

99%CL

99.9%CL 110

100

16.6 g/dl 90 17.1 g/dl 17.0 g/dl 16.8 g/dl 80

16.9 g/dl

17.5 g/dl 70

15.9 g/dl 17.2 g/dl

60

16.0 g/dl 17.3 g/dl 17.0 g/dl 18.6 g/dl

17.4 g/dl 50 0

1

10 EPO concentration in urine (ng/L)

100

1000

Fig.8. Samples from a major sports competition were analyzed with EPO WGA MAIIA after the ordinary IEF-based doping analysis. No positive sample was found with the IEF method. Total urine EPO concentration compared with the PMI value for EPO WGA MAIIA is shown. The concentration of haemoglobin was measured for some of the athletes (, N) and is noted to the right. As shown, 31% of the 101 athlete specimens contained EPO below 3 ng/L compared with 7.5% of 67 tested urine specimens from non-athletes. There were 8 MAIIA values (of 100 detectable samples) below the 99% confidence limit and 6 values below the 99.9% limit. There were 5 female athletes with aberrantly low MAIIA values showing hemoglobin above the limit of 16 g/dl for participation in, for example, cross-country skiing. Low EPO concentration, low MAIIA value, and high hemoglobin value seem to indicate doping.

Moreover, it is also possible to calculate the total EPO concentration in the eluate, and together with a reliable affinity purification using disposable anti-EPO monoliths, it is possible to estimate the concentration of the analyte in the original biological sample. When comparing EPO concentrations estimated in urine using EPO lateral flow immunoassay with the EPO WGA MAIIA, the correlation was good (R2 = 0.97, y = 1.02x + 2.8).

Discussion Methods for identification of EPO isoform distribution EPO varieties, in the low picomolar concentration range found in urine and plasma, can be distinguished by methods based on electrophoretic or chromatographic separation of the EPO isoforms in combination with a specific and sensitive antibody-based detection system. These methods identify exogenous EPO by its aberrant distribution of isoforms due to differences in charge/pI, apparent molecular mass, or WGA interaction compared with the endogenous EPO isoforms. The methods need a minimum amount of EPO to be able to measure EPO isoforms (EPO sensitivity) and, for the separation step, a certain degree of resolution for identification of isoforms aberrant from the endogenous forms (EPO isoform sen-

sitivity). However, there are many types of exogenous EPO and the methods distinguish different structures on them, with the result that one method can show high sensitivity for epoetin x and no sensitivity at all for epoetin y, whereas for another method it can be vice versa, as shown in Table 4. Interference in the methods can be due to low specificity in the detection system such as detection of proteins other than EPO [37,38], presence of endogenous EPO from cells other than human kidney cells [39], transport of plasma forms into urine after exercise [40,41], and presence of degradation products of endogenous EPO [21]. High variation in EPO isoform distribution in the samples from the healthy reference population (biological variation) and high variation in measured isoform distribution in the test (test variation) are other factors that reduce the EPO isoform sensitivity. The EPO isoform sensitivity is not trivial to establish and needs to be measured for each type of rhEPO. One method might be to measure samples where rhEPO has been mixed with endogenous EPO in different proportions, and the figure for the sensitivity limit will be calculated as percentage of rhEPO. For the IEF-based EPO doping control method, a sample containing 71% of rhEPO a and b mixture has been shown to be very close to the cutoff level [42]. Thus, the ability for the EPO WGA MAIIA to measure down to 40% epoetin b, as reported in this study, is a clear advantage.

110

Detection of erythropoiesis-stimulating agents / M. Lönnberg et al. / Anal. Biochem. 420 (2012) 101–114

Table 3 Typical results from EPO WGA MAIIA test showing that three different methods can be applied when using two MAIIA strips. A Retarded desorption

B Total desorption

Low ng/L

CV %

High ng/L

CV %

PMIb26

Urine specimens from healthy individuals U320 70.4 U251 68.1 U252 81.6 U239 65.7 U331 74.7 U336 263.1 U338 70.0 U339 80.7 U340 82.3 U342 55.9 U344 80.9 U345 89.3 Control rhEPO b 300 ng/L 82.9

6.0 10.8 2.0 4.3 4.7 2.9 6.7 1.5 6.4 4.2 0.2 2.9 8.9

117.0 101.8 109.7 95.0 114.0 385.5 111.6 123.2 107.8 95.6 115.7 133.6 324.0

2.5 0.9 1.9 5.8 2.0 1.9 0.7 1.5 16.7 0.1 2.9 3.3 4.6

60.2 66.9 74.4 69.1 65.5 68.2 62.8 65.5 76.3 58.5 70.0 66.8

Control rhEPO b 300 ng/L

82.3

32.2

311.8

6.1

26.4

0.94

1.05

Control 50% rhEPO

45.1

1.3

105.9

2.7

42.6

1.00

1.26 1.36

EPO WGA MAIIA results

C WGA MAIIA

25.6

D AbQ MAIIA Low RAM

E AbQ MAIIA High RAM

F

G Eluate

H Sample

I Urine

Dilution times

EPO ng/L

Volume mL

EPO ng/L

1.25 1.51 1.24 1.15 1.28 1.47 1.23 1.23 1.15 0.92 1.31 1.14 0.96

1.49 1.50 1.46 1.52 1.44 1.50 1.38 1.21 1.38 1.20 1.26 1.54 1.18

30 20 20 20 80 100 15 25 20 15 25 100

3510 2036 2194 1899 9118 38547 1674 3080 2157 1434 2891 13360

20 20 20 20 20 20 20 20 20 20 20 20

14.9 8.6 9.3 8.0 38.6 163.1 7.1 13.0 9.1 6.1 12.2 56.5

Control 50% rhEPO

50.8

0.1

110.5

4.1

46.0

1.21

Control 40% rhEPO

57.6

12.0

122.6

1.2

47.0

1.08

1.58

Control 40% rhEPO

64.5

2.9

123.3

3.9

52.3

1.26

1.38

Control Aranesp 300 ng/L

9.9

2.5

259.7

1.1

3.8

1.47

3.24

Control Aranesp 300 ng/L

11.7

0.3

275.4

1.3

4.3

1.91

3.13

Control Mircera 300 ng/L

73.4

1.3

86.9

0.9

84.4

4.33

4.69

Control Mircera 300 ng/L

73.5

0.9

84.1

4.8

87.4

4.32

4.26

Note: Columns A and B show the apparent concentration (in ng/L) obtained from the two strips for retarded (using elution buffer low) and total (using elution buffer high) desorption from the WGA zone, and in column C the PMI value calculated from these figures shows the interaction with WGA. Underlined values are outside the 99% reference limits. Columns D and E show the results from AbQ calculation with the RAM values for the interaction with the antibody zone. Aranesp and Mircera, the EPO-like molecules, show aberrant interaction, and values above the 99% reference limit are underlined. The EPO concentration in the eluate obtained from the affinity purification (column G) can be calculated by using the dilution factor (column F) for the eluate applied to the strip and the total desorption concentration (column B). The concentration in urine (column I) can be calculated from the eluate concentration (column G), the affinity purification application volume (column H), and the EPO recovery (here 65%).

Table 4 Interaction table for endogenous and exogenous EPO analyzed with different types of separation methods.

EPO form Source Producing cell Separation A WGA interaction B Charge C Apparent size

1 Urine EPO adult

2 Serum EPO adult

3 Degradation of carbohydrate Endogenous

4 Epoetin b NeoRecormon Recombinant CHO cells

5 Epoetin x Epomax Recombinant BHK cells

6 Darbepoetin a Aranesp Recombinant CHO cells

Endogenous Kidney cells (main)

Endogenous Kidney cells (main)

o



)?

+

+

++

o o

+ o

)+ )

++ +

+++ 

 ++

Note: The results for WGA interaction show that the endogenous forms are well distinguished from the recombinant EPOs. Degradation of the carbohydrate structure is not expected to result in increased binding to WGA, and should thus not result in a false identification of rhEPO. For separation based on charge or pI it has been shown that urine EPO is advantageous to separate from rhEPO, while serum EPO has a charge closer to rhEPO [18,41]. Degradation of sialic acids on the carbohydrate structure might result in a charge in the range of rhEPO [21]. Separation based on apparent size (SDS–PAGE) might have difficulties to identify epoetin x [21] as a degradation of e.g. sialic acid will show reduced apparent molecular mass in the same way as epoetin x. Same interaction as EPO normal endogenous urine (o); stronger WGA interaction, more positively charged forms, or higher mass (+); weaker WGA interaction, more negatively charged forms, or lower mass (-).

As reported previously [21], it was difficult to identify epoetins produced in BHK cells by the SDS–PAGE method. Moreover, epoetin from a human cell line (Dynepo) showed gel positions closer to endogenous EPO than other epoetins. This was found for both IEF and SDS–PAGE [21], but Dynepo was still well detected with SDS–PAGE [20]. The characteristic property of Dynepo, with a narrow molecular mass distribution, is probably due to the purification regime. With the EPO WGA MAIIA method, EPO produced in BHK and Dynepo showed the same interaction with WGA as epoetin b and, therefore, could efficiently be distinguished from endogenous EPO.

Prior to the analytical test, some of the methods use an EPO affinity purification step. This purification concentrates EPO from a large sample volume and increases the detection specificity, which is still required for the methods that use immunoblotting from the electrophoresis gel (IEF and SDS–PAGE). This purification is also needed to avoid reduction of ligand (WGA) capacity by other glycoproteins in the miniaturized MAIIA chromatographic separation step. It is essential to secure the quality of the affinity purification process by measuring EPO recovery, confirming retained isoform profile, and to make sure that no cross-contamination can occur between samples due to repeated use of columns.

Detection of erythropoiesis-stimulating agents / M. Lönnberg et al. / Anal. Biochem. 420 (2012) 101–114

The advent of a purification kit with disposable columns simplifies this procedure [43,44]. Standardization of EPO and EPO isoform distribution Isoform distribution methods might require two types of standardization, one for EPO quantification and one for establishing the chromatographic or electrophoretic migration position of EPO isoforms. The MAIIA test uses standardization (rhEPO standard in different concentrations) for quantification to make a correct ratio calculation. Moreover, a reliable estimation of the total EPO concentration in the sample can be done. Standardization of isoform distribution in the separation systems is a difficult problem and can be solved in various ways. In the IEF method, a mixture of epoetin a and b and Aranesp is always applied, often on both sides, to every gel. Bands from the epoetins are included in the basic area, and bands from Aranesp are included in the acidic area. In short, the presence of aberrant EPO forms is identified if bands in the basic/acidic range show higher intensity than bands in the endogenous range. The results have been expressed in percentage of basic/acidic forms or in a measurement of ratio between bands in the aberrant and endogenous range. For SDS–PAGE, urine EPO preparation is applied on the gel as a standard, and the centroid or boundaries of the width of urine EPO distribution are compared with the results from the EPO in the samples. The results have been expressed in relative mobility by applying EPO variants with different molecular weights [20]. In accordance with these methods, it will be necessary to include some type of chromatographic standardization in the EPO WGA MAIIA test. A plasma or urine EPO-based standard can be mixed with rhEPO b, as shown in Fig. 5, with a proportion of endogenous and rhEPO that corresponds to the PMI cutoff limit established with samples from the reference population. Such a cutoff system makes it possible to easily assess whether a sample is aberrant or not in the same way as for IEF and SDS–PAGE. However, for the isoform separation systems, it would be advantageous to give the results in figures, which enables statistical calculations and facilitates comparison of results from different laboratories and different analytical runs. The results from the different studies reported here showed mean endogenous urine EPO values for healthy individuals of 73.7 ± 6.9 PMIb26, 73.3 ± 7.6 PMIb30, and 84.7 ± 7.2 PMIb32. The WGA interaction with epoetin b was in the recommended range of 20 to 40 PMI for distinguishing rhEPO and endogenous EPO. When lower epoetin b interaction was selected, for detection of different endogenous EPO forms, the endogenous urine EPO reference mean value was lowered to 39.7 ± 8.7 PMIb18 and 30.2 ± 6.1 PMIb14. The procedure for optimization of the GlcNAc concentration has been used for each produced batch of strips, and different concentrations have been used. The interaction strength between EPO and WGA, and thus the PMI value, is regulated by both GlcNAc and the density of the immobilized WGA ligands. Because the WGA interaction strength seemed to show variation for different production batches and seemed to be reduced after some months of storage, it was not possible to use a recommend GlcNAc concentration. Thus, the use of the optimization procedure of the GlcNAc concentration is recommended until the producer has stabilized the WGA ligand function. The most important quality, low differences in WGA interaction within a strip batch, was good, as shown by the imprecision figures. There is a need for further studies of the MAIIA system, testing several production batches and test conditions, to propose a suitable chromatographic standardization system. It might be advantageous to have two or three standards with different PMI values included in each run and to calculate the sample PMI from these

111

values. For the detection of rhEPO in urine with MAIIA, it seems possible to use a selected cutoff control in each run and a yes/no answer, in accordance with the current EPO doping test where an ‘‘adverse analytical finding’’ is reported if rhEPO is available. EPO sensitivity of the detection system The normal EPO concentration in serum is approximately 30 to 153 ng/L [45] using a conversion factor of 1 IU = 8.3 ng. For the lowest normal concentration, 15 pg of EPO will be available in 0.5 ml of serum. The concentration in urine was for normal noncompetition individuals down to 1 ng/L (5 pg/5 ml urine). Urine samples collected during a competition showed EPO down to 0.1 ng/L (0.5 pg/5 ml). The optimal application amount for EPO WGA MAIIA is in the range of 5 to 15 pg of EPO per strip. The detection limit for the EPO lateral flow test has been measured previously [35] and found to be 0.2 ng/L EPO when using 50 ll of sample and 0.01 pg of EPO per strip. The quantification limit [46] is estimated to be 10 times higher, approximately 0.1 pg per strip. To measure MAIIA values of 80, 30, and 10 PMI using two strips, at least 0.25, 0.7, and 2 pg will be required of endogenous EPO, rhEPO b, and Aranesp, respectively. For the MAIIA test, it is recommended to start by purifying 0.5 ml of serum or 5 ml of urine, which will be sufficient for most of the samples. The preceding affinity purification has been tested with up to 2 ml of serum and 20 ml of urine. For the sarcosyl (SAR)–PAGE [47] method, 0.7 pg of rhEPO (Dynepo, which gives a very narrow band) directly applied to the gel can be faintly seen on the image. For the IEF method, approximately 3.4 ng/L in 20 ml of urine sample is required, assuming that 50% recovery of EPO is achieved in the preceding ultrafiltration step, corresponding to a required application amount of approximately 30 pg per lane [19]. The minimal application amount required for detection of rhEPO with the MAIIA test, 0.7 pg, is very competitive with the 0.7 and 30 pg of EPO required for SAR–PAGE and IEF, respectively. The sensitivity for the detection on gels has been enhanced (C. Reichel, Doping Control Laboratory, AIT Seibersdorf Laboratories, Seibersdorf, Austria) but has not, to our knowledge, been published so far. Low EPO concentrations in athlete urine were also found in a study [48] using the results from 3050 samples analyzed with the IEF EPO doping test, where 22% of the in-competition samples and 14% of the out-of-competition samples from men were below the detection limit for this method. The reason for the low EPO values was suggested to be due to the negative feedback inhibition of EPO production after EPO doping, highly diluted urine, and/or manipulation of urine samples by the addition of proteases. The number of samples with undetectable EPO is in accordance with our results from the reported competition, where 28% of the samples had no detectable EPO when using the IEF method but only 1% when using the MAIIA method. Our results indicate that a low EPO concentration, a MAIIA value in the rhEPO range, and a high hemoglobin concentration are related to reduced EPO production due to the increased hemoglobin concentration after ceased injection of rhEPO. However, when only a low concentration of rhEPO is left in the urine, a very sensitive EPO detection method, preceded by high recovery EPO purification, is required to be able to show the isoform distribution. Detection of EPO analogue Mircera Mircera is a conjugate of epoetin b and a polyethylene glycol (PEG) structure of equal size. This PEG conjugation reduces the interaction with other molecules such as the EPO receptor,

112

Detection of erythropoiesis-stimulating agents / M. Lönnberg et al. / Anal. Biochem. 420 (2012) 101–114

anti-EPO, and WGA. In the MAIIA lateral flow sandwich immunoassay, 450 ng/L Mircera (means of 400 and 500 ng/L when two batches were tested), corresponding to 225 ng/L epoetin, was required to achieve the same signal level as for 100 ng/L epoetin b. The difference between batches indicates that the linked PEG molecule, in one or more of the three possible positions at EPO, is preventing interaction of the EPO conjugate with the anti-EPO antibodies and that the distribution of the PEG substituent between the mentioned positions might not be the same for different batches. The affinity purification step prior to the EPO WGA MAIIA test gives recoveries of Mircera of 57%, 40%, and 30% when added to buffer, serum, and urine, respectively [44]. These recovery figures are approximately 50% lower than those for NeoRecormon. In summary, to achieve the same signal in the immunoassay after affinity purification, a nine times higher amount of Mircera is required compared with hEPO in the biological sample. The required amount of Mircera in serum for the AbQ approach when using two strips will be approximately 5 pg, assuming a yield of 40% in the affinity purification resulting in 2 pg of purified Mircera. This can be compared with the accredited IEF method [49], which requires 40 pg of Mircera applied to the gel. An improved version of SDS– PAGE for detection of Mircera, SAR–PAGE, showed a sensitivity limit of 3 pg of Mircera applied to the gel [47]. A sandwich immunoassay specific for Mircera, with anti-EPO and anti-PEG antibodies, was found to have a detection limit of 30 pg/ml Mircera and a cutoff limit at 100 pg/ml for serum samples [50]. The minimum application amount required for detection of Mircera with the MAIIA test, 2 pg, is very competitive with the 3, 40, and 10 to 20 pg of Mircera required for SAR–PAGE, IEF, and the Mircera sandwich immunoassay, respectively.

Possibility to use serum and urine specimens Serum or plasma samples are relatively homogeneous biological materials, mostly with predictable storage stabilities for their peptides and proteins. This is not the case for urine, where the composition can vary considerably, causing matrix effects and varying EPO concentrations. A urine specimen is preferred for doping analysis of low molecular substances, such as steroids, whereas serum or plasma seems to be better for analysis of peptides and proteins, such as human growth hormone (hGH) [51] and hEPO, which appear at much higher concentrations in serum than in urine. With the advent of larger ESA molecules, such as Mircera, serum or plasma is recommended because the large molecular mass significantly reduces the excretion into urine [49]. The collection of blood is a much simpler procedure [52] than the less secure collection of urine, which occasionally, in spite of the rigorous procedure, has been shown to allow manipulation of the sample [48]. Blood-based tests are dominating for clinical chemistry analysis and also seem to be of great benefit for direct identification of illegal EPO doping. A problem with this approach is that the currently accredited IEF-based method is not suitable for analysis of serum or plasma samples because the resolution between rhEPO and endogenous EPO affinity purified from serum is much less than that from the more acidic urine EPO [41]. However, it is possible to detect Mircera in serum with the IEF method [49]. Methods based on molecular mass determination seem to show the same result for rhEPO in serum and urine EPO. With the EPO WGA MAIIA method, it seems as though serum EPO can be equally, or even better, distinguished from rhEPO compared with urine EPO. By using an EPO isoform determination technique such as plasma MAIIA, the sample from the blood passport [53] analysis can also be used for EPO concentration and EPO isoform determination in addition to hemoglobin and reticulocyte percentage analysis.

EPO doping testing In summary, 8% of the athlete samples tested here showed aberrantly low MAIIA values in combination with aberrant AbQ values. This indicates that the samples contain an exogenous EPO-resembling molecule (not EPO because AbQ is aberrant) having WGAreacting structures such as a recombinant variety produced in CHO or BHK cells. The low EPO concentration found in urine and the very high hemoglobin concentration, as tested for 5 of the athletes (females), is a strong indication of doping. None of the tested urine specimens was positive with the accredited IEF-based doping analysis using the TD2004EPO regulation. Moreover, of the 25 athlete samples selected for the MAIIA PMIb18 test for detecting values higher than normal, 4 samples (16%) showed aberrantly high values, indicating the presence of only a weak interacting WGA form of EPO. The AbQ values were normal, showing that there were EPO molecules in the samples. If it had been Mircera, which also showed high MAIIA values, aberrant AbQ values also should have been found. In addition, 3 of the samples were detectable in IEF, but with high intensity in the basic bands 3 to 5. This EPO interaction pattern resembles the one for patients treated with ESAs due to kidney dysfunction. In spite of the dangers and performance-enhancing efficiency of ESA injections in sports, the general feeling is that a surprisingly limited number of EPO doping tests are performed. This might be due to the laborious procedure for the WADA-accredited EPO IEF method [54] and the alternative SDS–PAGE method. Increasing the testing frequency would most certainly have a strong preventive effect on EPO doping. However, to obtain such an increase, the cost per sample needs to be significantly reduced for the final customers, the sport federations. By introducing a screening test, based on high-throughput testing for more than 100,000 tests annually, the large number of negative samples could efficiently be cleared to a moderate cost. The MAIIA test seems to have the sensitivity required for a screening test and has the potential to be further developed for highthroughput testing. The advent of a rapid and affordable screening method would enable further testing and confirmation of the limited group of suspected samples, which accounts for less then 5% of the samples, with more elaborate methods. All three EPO isoform methods, based on pI, molecular mass, and WGA interaction, complement each other, and together they will have the specificity required for a confirmation procedure. Different endogenous EPO forms It seems that several distinct populations of EPO forms can be found in human samples. In the healthy adult, EPO from the kidney is dominating during normal conditions. The aberrant population of EPO forms found in 38% of urine samples from ESA-treated patients with kidney dysfunction may be a dominating endogenous production of EPO in the liver given that patients with renal anemia are only partially corrected with ESAs for their anemia [55]. Thus, the MAIIA method seems to be a valuable tool for clinical research to reveal how many of these patients in fact produce liver EPO. However, it was also found that 16% of the athlete urine samples collected during a competition contained EPO forms with the same weak WGA interaction as found in the patient samples. In this case, besides the production of liver EPO due to hypoxia [13], the presence of novel recombinant human EPO forms not so far tested with WGA interaction and EPO with degraded carbohydrate structures might be possible explanations for these results, and further characterization of these samples is required. The difference in PMIb15 values between hEPO from serum and urine specimens indicates lower numbers or fewer exposed polylactosamine structures for serum EPO. It has been reported that

Detection of erythropoiesis-stimulating agents / M. Lönnberg et al. / Anal. Biochem. 420 (2012) 101–114

serum EPO is less negatively charged [14] or has less acidic pI [41] compared with urine EPO. This may depend on different tubular handling, particularly tubular reabsorption, of isoforms of EPO [14]. Conclusion By a single EPO WGA MAIIA test with a total test time of approximately 1 h (20 min for affinity purification of a 5-ml urine sample, 30 min for MAIIA, and 10 min for detection and calculation), it is possible to distinguish several EPO forms from both plasma and urine requiring only a few picograms of EPO. The EPO WGA MAIIA method, in combination with the MAIIA EPO purification kit, can distinguish endogenous EPO from EPO produced in CHO cell cultures (epoetin a, b, n, and four Chinese variants), BHK cell cultures (EPO x), human cell cultures (EPO d), and the EPO analogues darbepoetin a and Mircera. The detection sensitivity is superior to the current accredited IEF method, and isoform identification is possible down to 0.7 pg of rhEPO, 2 pg of Aranesp, and 5 pg of Mircera in the biological sample. It is possible to distinguish rhEPO b in urine samples even when endogenous EPO is dominating in the EPO population, down to 40% rhEPO and 60% endogenous EPO. Thus, the MAIIA technology allows rapid and sensitive detection of doping with erythropoietin-stimulating agents. The EPO form with weak WGA interaction found in urine from patients treated with ESAs, which does not seem to be exogenous or a normally produced endogenous form, is interesting for further clinical research studies. Acknowledgments The authors thank Lina Lönnberg, Mikael Lönnberg, Niclas Rollborn, and Trikien Quach for technical assistance. Thanks also go to Peter Hemmersbach and Yvette Dehnes at the Doping Control Laboratory in Oslo, Norway, for providing reagents. The authors are grateful for the financial support and encouragement from Phadia (Uppsala, Sweden) and VINNOVA (Swedish governmental agency for innovation systems) in project P25917-1 and from the World Anti-Doping Agency (WADA) in project 05A1JC. References [1] G.L. Semenza, HIF-1 and mechanisms of hypoxia sensing, Curr. Opin. Cell Biol. 13 (2001) 167–171. [2] D. Lando, D.J. Peet, J.J. Gorman, D.A. Whelan, M.L. Whitelaw, R.K. Bruick, FIH-1 is an asparaginyl hydroxylase enzyme that regulates the transcriptional activity of hypoxia-inducible factor, Genes Dev. 16 (2002) 1466–1471. [3] A.J. Erslev, J. Caro, E. Kansu, O. Miller, E. Cobbs, Plasma erythropoietin in polycythemia, Am. J. Med. 66 (1979) 243–247. [4] J.F. Garcia, S.N. Ebbe, L. Hollander, H.O. Cutting, M.E. Miller, E.P. Cronkite, Radioimmunoassay of erythropoietin: circulating levels in normal and polycythemic human beings, J. Lab. Clin. Med. 99 (1982) 624–635. [5] L. Wide, C. Bengtsson, G. Birgegard, Circadian rhythm of erythropoietin in human serum, Br. J. Haematol. 72 (1989) 85–90. [6] B. Berglund, B. Ekblom, Effect of recombinant human erythropoietin treatment on blood pressure and some haematological parameters in healthy men, J. Intern. Med. 229 (1991) 125–130. [7] B. Ekblom, B. Berglund, Effect of erythropoietin administration on maximal aerobic power, Scand. J. Med. Sci. Sports 1 (1991) 88–93. [8] S.E. Franz, Erythropoiesis-stimulating agents: development, detection, and dangers, Drug Test. Anal. 1 (2009) 245–249. [9] N. Piloto, H.M. Teixeira, E. Teixeira-Lemos, B. Parada, P. Garrido, J. Sereno, R. Pinto, L. Carvalho, E. Costa, L. Belo, A. Santos-Silva, F. Teixeira, F. Reis, Erythropoietin promotes deleterious cardiovascular effects and mortality risk in a rat model of chronic sports doping, Cardiovasc. Toxicol. 9 (2009) 201–210. [10] E.R. Eicher, Better dead than second, J. Lab. Clin. Med. 120 (1992) 359–360. [11] K.U. Eckardt, P.J. Ratcliffe, C.C. Tan, C. Bauer, A. Kurtz, Age-dependent expression of the erythropoietin gene in rat liver and kidneys, J. Clin. Invest. 89 (1992) 753–760. [12] E.D. Zanjani, J.L. Ascensao, P.B. McGlave, M. Banisadre, R.C. Ash, Studies on the liver to kidney switch of erythropoietin production, J. Clin. Invest. 67 (1981) 1183–1188.

113

[13] C.C. Tan, K.U. Eckardt, J.D. Firth, P.J. Ratcliffe, Feedback modulation of renal and hepatic erythropoietin mRNA in response to graded anemia and hypoxia, Am. J. Physiol. Renal Physiol. 263 (1992) F474–F481. [14] L. Wide, C. Bengtsson, Molecular charge heterogeneity of human serum erythropoietin, Br. J. Haematol. 76 (1990) 121–127. [15] S. Lamon, N. Robinson, M. Saugy, Procedures for monitoring recombinant erythropoietin and analogs in doping, Endocrinol. Metab. Clin. North Am. 39 (2010) 141–154. [16] O. Barroso, I. Mazzoni, O. Rabin, Hormone abuse in sports: the antidoping perspective, Asian J. Androl. 10 (2008) 391–402. [17] J. Segura, J.A. Pascual, R. Gutierrez-Gallego, Procedures for monitoring recombinant erythropoietin and analogues in doping control, Anal. Bioanal. Chem. 388 (2007) 1521–1529. [18] L. Wide, C. Bengtsson, B. Berglund, B. Ekblom, Detection in blood and urine of recombinant erythropoietin administered to healthy men, Med. Sci. Sports Exerc. 27 (1995) 1569–1576. [19] F. Lasne, L. Martin, N. Crepin, J. de Ceaurriz, Detection of isoelectric profiles of erythropoietin in urine: differentiation of natural and administered recombinant hormones, Anal. Biochem. 311 (2002) 119–126. [20] M. Kohler, C. Ayotte, P. Desharnais, U. Flenker, S. Lüdke, M. Thevis, E. VölkerSchänzer, W. Schänzer, Discrimination of recombinant and endogenous urinary erythropoietin by calculating relative mobility values from SDS gels, Intl. J. Sports Med. 29 (2008) 1–6. [21] C. Reichel, R. Kulovics, V. Jordan, M. Watzinger, T. Geisendorfer, SDS–PAGE of recombinant and endogenous erythropoietins: benefits and limitations of the method for application in doping control, Drug Test. Anal. 1 (2009) 43– 50. [22] L. Franco Fraguas, J. Carlsson, M. Lönnberg, Lectin affinity chromatography as a tool to differentiate endogenous and recombinant erythropoietins, J. Chromatogr. 1212 (2008) 82–88. [23] M. Takeuchi, S. Takasaki, H. Miyazaki, T. Kato, S. Hoshi, N. Kochibe, A. Kobata, Comparative study of the asparagine-linked sugar chains of human erythropoietins purified from urine and the culture medium of recombinant Chinese hamster ovary cells, J. Biol. Chem. 263 (1988) 3657–3663. [24] C. Neususs, U. Demelbauer, M. Pelzing, Glycoform characterization of intact erythropoietin by capillary electrophoresis–electrospray–time of flight–mass spectrometry, Electrophoresis 26 (2005) 1442–1450. [25] G. Stubiger, M. Marchetti, M. Nagano, C. Reichel, G. Gmeiner, G. Allmaier, Characterisation of intact recombinant human erythropoietins applied in doping by means of planar gel electrophoretic techniques and matrix-assisted laser desorption/ionisation linear time-of-flight mass spectrometry, Rapid Commun. Mass Spectrom. 19 (2005) 728–742. [26] P.E. Groleau, P. Desharnais, L. Cote, C. Ayotte, Low LC–MS/MS detection of glycopeptides released from pmol levels of recombinant erythropoietin using nanoflow HPLC–chip electrospray ionization, J. Mass Spectrom. 43 (2008) 924– 935. [27] S.S. Park, J. Park, J. Ko, L. Chen, D. Meriage, J. Crouse-Zeineddini, W. Wong, B.A. Kerwin, Biochemical assessment of erythropoietin products from Asia versus US epoetin alfa manufactured by Amgen, J. Pharm. Sci. 98 (2009) 1688–1699. [28] I.C. Macdougall, M. Ashenden, Current and upcoming erythropoiesisstimulating agents, iron products, and other novel anemia medications, Adv. Chronic Kidney Dis. 16 (2009) 117–130. [29] W. Jelkmann, Novel erythropoietic agents: a threat to sportsmanship, Med. Sport. 11 (2007) 32–42. [30] M. Lönnberg, Membrane-Assisted Isoform Immunoasay: Separation and Determination of Protein Isoforms (Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology, No. 691), Acta Universitatis Upsaliensis, 2002. [31] M. Lonnberg, J. Carlsson, Membrane assisted isoform immunoassay: a rapid method for the separation and determination of protein isoforms in an integrated immunoassay, J. Immunol. Methods 246 (2000) 25–36. [32] M. Lonnberg, J. Carlsson, Chromatographic performance of a thin microporous bed of nitrocellulose, J. Chromatogr. B 763 (2001) 107–120. [33] M. Lonnberg, J. Carlsson, Lab-on-a-chip technology for determination of protein isoform profiles, J. Chromatogr. A 1127 (2006) 175–182. [34] M. Lönnberg, J. Carlsson, Quantitative detection in the attomole range for immunochromatographic tests by means of a flatbed scanner, Anal. Biochem. 293 (2001) 224–231. [35] M. Lönnberg, M. Drevin, J. Carlsson, Ultra-sensitive immunochromatographic assay for quantitative determination of erythropoietin, J. Immunol. Methods 339 (2008) 236–244. [36] W.E. Owen, W.L. Roberts, Performance characteristics of the Immulite 2000 erythropoietin assay, Clin. Chim. Acta 340 (2004) 213–217. [37] M. Beullens, J.R. Delanghe, M. Bollen, False-positive detection of recombinant human erythropoietin in urine following strenuous physical exercise, Blood 107 (2006) 4711–4713. [38] F. Lasne, No doubt about the validity of the urine test for detection of recombinant human erythropoietin, Blood 108 (2006) 1778–1779. author reply: 1779–1780. [39] Y. Dehnes, M. Lönnberg, M. Borgen, P. Hemmersbach, Shifted isoelectric profiles of endogenous EPO possibly due to different physiological origin, Recent Adv. Doping Anal. 16 (2008) 359–362. [40] S. Lamon, L. Martin, N. Robinson, M. Saugy, J. Ceaurriz, F. Lasne, Effects of exercise on the isoelectric patterns of erythropoietin, Clin. J. Sport Med. 19 (2009) 311–315.

114

Detection of erythropoiesis-stimulating agents / M. Lönnberg et al. / Anal. Biochem. 420 (2012) 101–114

[41] F. Lasne, L. Martin, J.A. Martin, J. de Ceaurriz, Isoelectric profiles of human erythropoietin are different in serum and urine, Intl. J. Biol. Macromol. 41 (2007) 354–357. [42] J. Mallorqui, R. Gutierrez-Gallego, J. Segura, C. de Bolos, J.A. Pascual, New screening protocol for recombinant human erythropoietins based on differential elution after immunoaffinity purification, J. Pharm. Biomed. Anal. 51 (2010) 255–259. [43] Y. Dehnes, S. Lamon, M. Lonnberg, Erythropoietin (EPO) immunoaffinity columns: a powerful tool for purifying EPO and its recombinant analogues, J. Pharm. Biomed. Anal. 53 (2010) 1028–1032. [44] M. Lonnberg, Y. Dehnes, M. Drevin, M. Garle, S. Lamon, N. Leuenberger, T. Quach, J. Carlsson, Rapid affinity purification of erythropoietin from biological samples using disposable monoliths, J. Chromatogr. A 1217 (2010) 7031–7037. [45] P. Mossuz, F. Girodon, S. Hermouet, I. Dobo, E. Lippert, M. Donnard, V. LatgerCannard, N. Boiret, V. Praloran, J.C. Lecron, Serum erythropoietin measured by chemiluminescent immunometric assay: an accurate diagnostic test for absolute erythrocytosis, Clin. Chem. 51 (2005) 1018–1021. [46] D.A. Armbruster, T. Pry, Limit of blank, limit of detection, and limit of quantitation, Clin. Biochem. Rev. 29 (Suppl. 1) (2008) S49–S52. [47] C. Reichel, F. Abzieher, T. Geisendorfer, SARCOSYL–PAGE: a new method for the detection of Mircera- and EPO-doping in blood, Drug Test. Anal. 1 (2009) 494–504. [48] S. Lamon, N. Robinson, P-E. Sottas, H. Henry, M. Kamber, P. Mangin, M. Saugy, Possible origins of undetectable EPO in urine samples, Clin. Chim. Acta 385 (2007) 61–66.

[49] F. Lasne, L. Martin, J.A. Martin, J. de Ceaurriz, Detection of continuous erythropoietin receptor activator in blood and urine in anti-doping control, Haematologica 94 (2009) 888–890. [50] S. Lamon, S. Giraud, L. Egli, J. Smolander, M. Jarsch, K.G. Stubenrauch, A. Hellwig, M. Saugy, N. Robinson, A high-throughput test to detect C.E.R.A. doping in blood, J. Pharm. Biomed. Anal. 50 (2009) 954–958. [51] M. Bidlingmaier, J. Suhr, A. Ernst, Z. Wu, A. Keller, C.J. Strasburger, A. Bergmann, High-sensitivity chemiluminescence immunoassays for detection of growth hormone doping in sports, Clin. Chem. 55 (2009) 445–453. [52] M. Saugy, N. Robinson, C. Saudan, The fight against doping: back on track with blood, Drug Test. Anal. 1 (2009) 474–478. [53] K. Sharpe, M.J. Ashenden, Y.O. Schumacher, A third generation approach to detect erythropoietin abuse in athletes, Haematologica 91 (2006) 356–363. [54] C. Ayotte, G. Gmeiner, F. Lasne, P. A. Pascual, C. Reichel, M. Saugy, Harmonization of the method for the identification of recombinant erythropoietins (i.e. epoetins) and analogues (i.e. darbepoetin and pegserpoetin), World Anti-Doping Agency technical document TD2009EPO, 2009. [55] H. Furuland, T. Linde, J. Ahlmen, A. Christensson, U. Strombom, B.G. Danielson, A randomized controlled trial of haemoglobin normalization with epoetin alfa in pre-dialysis and dialysis patients, Nephrol. Dial. Transplant. 18 (2003) 353– 361.