RESUMEN
La determinación del nivel de anticuerpos por medio de ensayos ELISA exige construir una curva de calibración según el modelo logístico de 4 parámetros (4PL); no obstante, es común que el bioquímico realice estimaciones y ajuste los datos de calibración con modelos alternativos. Se buscó determinar para un ensayo ELISA semicuantitativo el modelo que mejor ajusta los datos de calibración y cuál de los modelos alternativos explorados genera menor error relativo porcentual (ERP) al predecir las concentraciones de los calibradores. Para esto se empleó un ELISA indirecto y se ajustaron las densidades ópticas (DO) de los calibradores según los modelos i) exponencial ii) Boltzmann iii) Boltzmann semi log iv) Deming v) regresión lineal vi) 4PL vii) cuadrático. Se encontró que el mejor ajuste lo proveen los modelos v) (R2=0,9914), i) (R2=0,9652) y iii) (R2=0,9650). Sin embargo, los modelos i) y iii) tienen mejor desempeño en el procedimiento de ajuste inverso: el ERP se mantuvo ≤20% en el rango cubierto por los 6 calibradores (0-100 UI/mL). Los modelos lineales iv) y v) presentaron ERP elevados en el rango testeado. En resumen, el ajuste exponencial i) y el ajuste de Boltzmann iii), combinan valores de R2 y ERP comparables al modelo 4PL, por lo cual es inapropiado cualquier tipo de ajuste lineal.
Determination of the levels of specific antibodies by semiquantitative ELISAs requires the construction of a 4 parameter logistic regression (4PL) calibration curve. However, alternative models are often employed to adjust and estimate calibration data. The aims of this work were to determine the model that best adjusts calibration data, and which alternative model generates a lower relative percentage error (RPE) in the prediction of calibrator concentrations. An IgA anti-gliadin ELISA was used. The optical density values (OD) of calibrators were adjusted with the following mathematical models: i) Exponential, ii) Boltzmann, iii) Boltzmann semilog, iv) Deming, v) Linear regression, vi) 4PL, and vii) Quadratic. Results indicated that the best adjustment is given by models v) (R2=0.9914), i) (R2=0.9652) and iii) (R2=0.9650). However, models i) and iii) performed better in the reciprocal adjustment procedure: RPE values were ≤20% for all the calibrator levels analyzed (0-100 IU/mL). The linear models iv) and v) had high RPE values. To sum up, the exponential (i) and Boltzmann (iii) adjustments present R2 and RPE values similar to those of the 4PL, the use of any linear adjustment being inappropriate.
A determinação do nível de anticorpos por meio de ELISA requer a construção de uma curva de calibração de acordo com o modelo logístico de 4 parâmetros (4PL); no entanto, é comum o bioquímico fazer estimativas e ajustar os dados de calibração com modelos alternativos. Procurou-se determinar para um ensaio ELISA de modelo semiquantitativo o modelo que melhor se ajusta os dados de calibração e qual dos modelos alternativos explorados gera menor erro relativo percentual (ERP) ao prever as concentrações dos calibradores. Para isso, um ELISA indireto foi utilizado e se ajustaram as densidades ópticas (DO) dos calibradores segundo os modelos i) exponencial ii) Boltzmann iii) Boltzmann semi log iv) Deming v) regressão lineal iv) 4PL vii) quadrático. Verificou-se que o melhor ajuste é fornecido pelos modelos v) (R2= 0,9914), vi) (R2= 0,9652) e iii) (R2= 0,9650). No entanto, os modelos i) e iii) têm melhor desempenho no procedimento de ajuste inverso: o ERP permaneceu ≤20% na faixa coberta pelos 6 calibradores (0-100 UI/mL). Os modelos lineares iv) e v) apresentaram alto ERP na faixa testada. Em resumo, o ajuste exponencial i) e o ajuste de Boltzmann iii) combinam valores de R2 e ERP comparáveis ao modelo de 4PL, sendo inadequado qualquer tipo de ajuste linear.
Asunto(s)
Calibración , Ensayo de Inmunoadsorción Enzimática , Inmunoensayo , Métodos , Inmunoglobulina A , Modelos Lineales , Modelos Logísticos , Eficiencia , Alergia e Inmunología , Estándares de Referencia , Gliadina , AnticuerposRESUMEN
Objective To study the chemical kinetics characteristics in a new revised method with low usage amount of arsenic trioxide for determining urinary iodine by arsenite-ceric catalytic spectrophotometry using ammonium persulfate digestion,and to study the impact of operating bias in arsenite-ceric reaction temperature and reaction time on final results in this method.Methods The absorbances (A) of arsenite-ceric reaction of iodine standard series were measured at different reaction temperature and time,and the results were analyzed according to the chemical kinetics equation.The change values and half-life of A values of the new revised method and the current standard method were calculated.The chemical kinetics model of reaction system for this new revised method was deduced from experimental results.The calculation formula of result relative error for urinary iodine determination was deduced based on constants reaction temperature and reaction time and reaction rate constant factor.The result relative errors caused by operation deviation of reaction temperature or reaction time in the determination of urinary iodine were calculated.Results The usage amount of arsenious acid solution in the new revised method was only a quarter of usage amount of the current standard method(WS/T 107-2006).A values of each concentration of standard curve series at different reaction time t were obtained,the lnA to t mapping was a straight line,the linear correlation coefficients were-0.9995--0.9999.These results were in accord with the characteristic of chemical first-order reaction.Relationships between the reaction rate constant K and the reaction temperature T in the temperature range of 20-35 ℃ were well accord with Arrhenius equation.The A values and iodine concentrations (C) at various experimental temperatures showed good C =a + blnA linear relation,the absolute value of the linear correlation coefficient(| r |) > 0.9990.After calculation and comparison of changes in the half-life of A values in the new revised method and in the original standard method at 20,25,30,35 ℃ reaction temperature,half-life of A values of 0-300 μg/L iodine standard series in the new revised method and in the original standard method were 191.0-11.4 min and 66.8-10.2 min at 25℃,respectively.Under the same conditions of 25 ℃ for 40 min,the gradient of A values of 0-300 μg/L iodine standard curve in the new revised method was similar to that of the original standard method(slope-133.7,-139.2,respectively).But differences between A values of standard curve and the reaction initial absorbance(A0) in the original standard method were 1.4 to 3.7 times those of the new revised method.A chemical kinetics model of reacting system for this method was established.A series of urinary iodine results relative error data were obtained when reaction temperature deviation was ± 1,± 0.5,± 0.3 ℃ or reaction time deviation was ± 1 min for sample test tubes.Data showed that relative errors of urinary iodine results caused by reaction temperature deviation or reaction time deviation in the new revised method were less than those of the original standard method.Conclusions The iodine-catalyzed arsenite-ceric reaction in the new revised method is a first-order reaction,when measuring 0-300 μg/L urinary iodine at 20-35 ℃,and 300-1200 μg/L urinary iodine at 20-30 ℃,the calibration relation of C =a + blnA is established when arsenite-ceric catalytic spectrophotometry is kept at a certain stable temperature and in certain stable reaction time.Compared with the original standard method,using the revised method with low usage amount of arsenic trioxide for measuring urinary iodine,the arsenite-ceric reaction rate is slow down.As a result,this method is easier to operate and has better precision and accuracy.
RESUMEN
Weighing collected samPles to determine the IPS' content of indoor air and adopting the length of collecting time as 4 , 6and8 hours, the experimental accuracy and analytic errors were seParatdy analysed in different length of collecting time, as a result, when the IPS' content is less than 0.300mg.m-3, the smallest co-fficient of variation came from the statistic result of which its length of collecting time was 8 hours and in anaylsis its Permissible relative errors were not more than 10%, so it could be determined that the apporiate length of ccollecting time was 8 hours ( CIP