Limits...
Methodological aspects of ELISA analysis of thioredoxin 1 in human plasma and cerebrospinal fluid.

Lundberg M, Curbo S, Reiser K, Masterman T, Braesch-Andersen S, Areström I, Ahlborg N - PLoS ONE (2014)

Bottom Line: Increased plasma levels of Trx1 have been associated with human diseases suggesting that Trx1 is a marker for oxidative stress with putative clinical use.Importantly, analyses of human plasma and CSF without preventing HA interference may obscure the obtained data.Overall, the results of this study are crucial for the improvement of future studies on the association of Trx1 levels with various diseases.

View Article: PubMed Central - PubMed

Affiliation: Department of Clinical Science and Education, Karolinska Institutet, Stockholm, Sweden.

ABSTRACT
Thioredoxin-1 (Trx1) is a protein antioxidant involved in major cellular processes. Increased plasma levels of Trx1 have been associated with human diseases suggesting that Trx1 is a marker for oxidative stress with putative clinical use. However, the reported mean levels of Trx1 in the control cohorts vary a hundred-fold between studies (0.8-87 ng/ml), possibly due to methodological differences between the capture ELISA used in the different studies. The aim of this study was to investigate methodological aspects related to the ELISA measurement of Trx1. ELISAs utilizing different capture and detection combinations of antibodies to Trx1 and as well as recombinant human (rh) Trx1 standards from two sources were characterized. The different ELISAs were subsequently used to measure Trx1 in human plasma and cerebrospinal fluid samples (CSF) from healthy donors and from patients with various neurological diagnoses. The Trx1 standards differed in their content of monomeric and oligomeric Trx1, which affected the ELISAs composed of different antibody combinations. Thus, the levels of Trx1 determined in human plasma and CSF samples varied depending on the antibody used in the ELISAs and on the rhTrx1 standard. Furthermore, the relevance of preventing interference by heterophilic antibodies (HA) in human plasma and CSF was investigated. The addition of a HA blocking buffer to human samples drastically reduced the ELISA signals in many samples showing that HA are likely to cause false positive results unless they are blocked. In conclusion, the study shows that the design of a Trx1 ELISA in regards to antibodies and standards used has an impact on the measured Trx1 levels. Importantly, analyses of human plasma and CSF without preventing HA interference may obscure the obtained data. Overall, the results of this study are crucial for the improvement of future studies on the association of Trx1 levels with various diseases.

Show MeSH

Related in: MedlinePlus

Agreement of the level of Trx1 in CSF samples by pAb/pAb and mAb/pAb capture ELISA.Fifty-seven CSF samples from patients with diverse diagnosis were analysed for the level of Trx1 using the ELISA formats pAb/pAb and mAb/pAb in the presence of either incubation buffer or a HA blocking buffer. The agreement between the two ELISA formats used for analysis of samples diluted in incubation buffer (A) or HA blocking buffer (B) was assessed with Bland-Altman analysis. One of two experiments with similar results is shown. ––– Bias •••••••••••• CI Bias (95%) – – – – CI (95%).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4116216&req=5

pone-0103554-g006: Agreement of the level of Trx1 in CSF samples by pAb/pAb and mAb/pAb capture ELISA.Fifty-seven CSF samples from patients with diverse diagnosis were analysed for the level of Trx1 using the ELISA formats pAb/pAb and mAb/pAb in the presence of either incubation buffer or a HA blocking buffer. The agreement between the two ELISA formats used for analysis of samples diluted in incubation buffer (A) or HA blocking buffer (B) was assessed with Bland-Altman analysis. One of two experiments with similar results is shown. ––– Bias •••••••••••• CI Bias (95%) – – – – CI (95%).

Mentions: The analysis of CSF by the two ELISA formats described above were conducted using the HA blocking buffer to exclude a possible influence by HA. To study the agreement between the two ELISAs a Bland-Altman plot was established and showed to exhibit a bias of 0.87 (95% confidence interval, 0.17 to 1.57) (Fig. 6B). However, CSF has also been shown to contain HA with activity that interferes with the analysis of biological markers [22]. We thus decided to analyze the importance of preventing HA interference for the quantification of Trx1 in CSF. All CSF samples were diluted in incubation buffer (not blocking HA) and subsequently re-analyzed with the two ELISAs. The median level of Trx1 with the pAb/pAb ELISA was 6.3 ng/ml (interquartile range 3.6–13.8 ng/ml) and with the mAb/pAb ELISA 0.93 ng/ml (interquartile range 0.30–2.86 ng/ml). The Bland-Altman plot exhibited a higher bias of 5.98 (95% confidence interval, 0.93 to 11.03) (Fig. 6A). Studying the Bland-Altman plots it is evident that both plots exhibit a measurement error proportional to the mean. In these plots it means that the standard deviation increases with increasing mean values since the values of the mAb/pAb ELISA in some CSF samples were 100-folds lower than the values of the pAb/pAb ELISA and that some of the lower concentrations of Trx1 in the CSF samples were below detection limit in the mAb/pAb ELISA in the presence of HA blocking buffer. Comparison of the two Bland-Altman plots shows that the bias is markedly higher in incubation buffer when HA has not been blocked and it is also clear that the median range of the values is much higher in the absence of HA blocking buffer. Statistically the median levels are significantly higher for both ELISAs when HA have not been blocked (p<0.01). The correlations between the ELISAs determined with Spearman’s rank show that the correlation is better between the two ELISAs for CSF samples diluted in the HA blocking buffer (rs = 0.66) than in incubation buffer (rs = 0.37).


Methodological aspects of ELISA analysis of thioredoxin 1 in human plasma and cerebrospinal fluid.

Lundberg M, Curbo S, Reiser K, Masterman T, Braesch-Andersen S, Areström I, Ahlborg N - PLoS ONE (2014)

Agreement of the level of Trx1 in CSF samples by pAb/pAb and mAb/pAb capture ELISA.Fifty-seven CSF samples from patients with diverse diagnosis were analysed for the level of Trx1 using the ELISA formats pAb/pAb and mAb/pAb in the presence of either incubation buffer or a HA blocking buffer. The agreement between the two ELISA formats used for analysis of samples diluted in incubation buffer (A) or HA blocking buffer (B) was assessed with Bland-Altman analysis. One of two experiments with similar results is shown. ––– Bias •••••••••••• CI Bias (95%) – – – – CI (95%).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4116216&req=5

pone-0103554-g006: Agreement of the level of Trx1 in CSF samples by pAb/pAb and mAb/pAb capture ELISA.Fifty-seven CSF samples from patients with diverse diagnosis were analysed for the level of Trx1 using the ELISA formats pAb/pAb and mAb/pAb in the presence of either incubation buffer or a HA blocking buffer. The agreement between the two ELISA formats used for analysis of samples diluted in incubation buffer (A) or HA blocking buffer (B) was assessed with Bland-Altman analysis. One of two experiments with similar results is shown. ––– Bias •••••••••••• CI Bias (95%) – – – – CI (95%).
Mentions: The analysis of CSF by the two ELISA formats described above were conducted using the HA blocking buffer to exclude a possible influence by HA. To study the agreement between the two ELISAs a Bland-Altman plot was established and showed to exhibit a bias of 0.87 (95% confidence interval, 0.17 to 1.57) (Fig. 6B). However, CSF has also been shown to contain HA with activity that interferes with the analysis of biological markers [22]. We thus decided to analyze the importance of preventing HA interference for the quantification of Trx1 in CSF. All CSF samples were diluted in incubation buffer (not blocking HA) and subsequently re-analyzed with the two ELISAs. The median level of Trx1 with the pAb/pAb ELISA was 6.3 ng/ml (interquartile range 3.6–13.8 ng/ml) and with the mAb/pAb ELISA 0.93 ng/ml (interquartile range 0.30–2.86 ng/ml). The Bland-Altman plot exhibited a higher bias of 5.98 (95% confidence interval, 0.93 to 11.03) (Fig. 6A). Studying the Bland-Altman plots it is evident that both plots exhibit a measurement error proportional to the mean. In these plots it means that the standard deviation increases with increasing mean values since the values of the mAb/pAb ELISA in some CSF samples were 100-folds lower than the values of the pAb/pAb ELISA and that some of the lower concentrations of Trx1 in the CSF samples were below detection limit in the mAb/pAb ELISA in the presence of HA blocking buffer. Comparison of the two Bland-Altman plots shows that the bias is markedly higher in incubation buffer when HA has not been blocked and it is also clear that the median range of the values is much higher in the absence of HA blocking buffer. Statistically the median levels are significantly higher for both ELISAs when HA have not been blocked (p<0.01). The correlations between the ELISAs determined with Spearman’s rank show that the correlation is better between the two ELISAs for CSF samples diluted in the HA blocking buffer (rs = 0.66) than in incubation buffer (rs = 0.37).

Bottom Line: Increased plasma levels of Trx1 have been associated with human diseases suggesting that Trx1 is a marker for oxidative stress with putative clinical use.Importantly, analyses of human plasma and CSF without preventing HA interference may obscure the obtained data.Overall, the results of this study are crucial for the improvement of future studies on the association of Trx1 levels with various diseases.

View Article: PubMed Central - PubMed

Affiliation: Department of Clinical Science and Education, Karolinska Institutet, Stockholm, Sweden.

ABSTRACT
Thioredoxin-1 (Trx1) is a protein antioxidant involved in major cellular processes. Increased plasma levels of Trx1 have been associated with human diseases suggesting that Trx1 is a marker for oxidative stress with putative clinical use. However, the reported mean levels of Trx1 in the control cohorts vary a hundred-fold between studies (0.8-87 ng/ml), possibly due to methodological differences between the capture ELISA used in the different studies. The aim of this study was to investigate methodological aspects related to the ELISA measurement of Trx1. ELISAs utilizing different capture and detection combinations of antibodies to Trx1 and as well as recombinant human (rh) Trx1 standards from two sources were characterized. The different ELISAs were subsequently used to measure Trx1 in human plasma and cerebrospinal fluid samples (CSF) from healthy donors and from patients with various neurological diagnoses. The Trx1 standards differed in their content of monomeric and oligomeric Trx1, which affected the ELISAs composed of different antibody combinations. Thus, the levels of Trx1 determined in human plasma and CSF samples varied depending on the antibody used in the ELISAs and on the rhTrx1 standard. Furthermore, the relevance of preventing interference by heterophilic antibodies (HA) in human plasma and CSF was investigated. The addition of a HA blocking buffer to human samples drastically reduced the ELISA signals in many samples showing that HA are likely to cause false positive results unless they are blocked. In conclusion, the study shows that the design of a Trx1 ELISA in regards to antibodies and standards used has an impact on the measured Trx1 levels. Importantly, analyses of human plasma and CSF without preventing HA interference may obscure the obtained data. Overall, the results of this study are crucial for the improvement of future studies on the association of Trx1 levels with various diseases.

Show MeSH
Related in: MedlinePlus