Establishment and validation of a liquid-phase blocking ELISA for the detection of antibodies elicited by the foot-and-mouth disease virus A/ASIA/Sea-97 lineage

Figures & data

Table 1. List of pig serum samples used in this study. All categories of sera were used for the establishment of the LPBE and evaluation of its application in the field. Panels 1 and 2 included sequentially collected serum samples from pigs vaccinated with a monovalent vaccine and subsequently challenged with the homologous vaccine strain of FMDV serotype A (collected at 0, 21 and 28 dpv and at 1–7 dpc). Panel 3 included FMDV-negative sera obtained from naïve pigs that had been neither vaccinated nor infected. Panel 4 included field sera obtained from a conventional pig farm with historical FMD vaccination records.

Category	Sample size (n)	Vaccine antigen	Vaccination	Challenge virus	Bleeding (weeks old)
Panel 1	142	A/Yeoncheon/SKR/2017	1 dose	A/Yeoncheon/SKR/2017	Sequentially
Panel 2	94	A/Pocheon/SKR/2010	1 dose	A/Pocheon/SKR/2010	Sequentially
Panel 3	101	–	–	–	7–8
Panel 4	33	A22/IRQ/24/64	2 doses (homologous)	-	22
	34	A24/Cruzeiro + A/Argentina/2001	2 doses (homologous)	–
	33	A/Zabaikalsky/2013	2 doses (homologous)	–
	30	A22/IRQ/64 and A/Zabaikalsky/2013	2 doses (heterologous)	–
	18	A24/Cruzeiro + A/Argentina/2001 and A/Zabaikalsky/2013	2 doses (heterologous)	–

Figure 1. Frequency distributions of the PI values (%) measured for known positive and negative sera, indicating the presence of antibodies against FMDV serotype A detected by the LPBE. Whether test sera were positive or negative was previously determined based on the level of NAb titers against A/YC (A) and A/PC (B). A reciprocal log₁₀ NAb titer of 1.2 (16-fold) or more was considered positive.

Table 2. Diagnostic performance of the LPBE for the detection of antibodies raised against FMDV serotype A. The diagnostic values of the LPBE for serum panels 1 and 2 were computed by comparison with the results of the VNT using contingency table analysis.

Source of sera (n)	Positive		Negative		Fisher** P value	Diagnostic value^†
	No. of true	No. of false	No. of true	No. of false		Con (%)	Sen % (95% CI)	Spe % (95% CI)	PPV % (95% CI)	NPV % (95% CI)	K value (95% CI)
Panel 1 * (n = 142)	88	9	40	5	< 0.0001	90.1	90.7 (83.1-95.7)	88.9 (75.9-96.3)	94.6 (87.9-98.2)	81.6 (68-91.2)	0.78 (0.67-0.88)
Panel 2 * (n = 94)	62	8	23	1	< 0.0001	90.4	88.6 (78.6-94.9)	95.8 (78.8-99.8)	98.4 (91.5-99.9)	74.2 (55.4-88.1)	0.77 (0.63-0.91)

*Sequentially collected serum samples from pigs vaccinated with a monovalent vaccine and subsequently challenged with the homologous vaccine strain of FMDV, A/Yeoncheon/SKR/2017 or A/Pocheon/SKR/2017.

**Statistical significance of the interaction of the LPBE with the VNT.

^† Con, concordance; Sen, sensitivity; Spe, specificity; PPV, positive predictive value; NPV, negative predictive value; k, Cohen’s kappa.

Figure 2. Correlation of endpoint antibody titers measured by the LPBE and VNT in serum panels 1 (n = 142) and 2 (n = 94). To assess agreement between the two assays, scatter plots with linear regression curves and Bland–Altman scatter plots were generated from quantitative data for serum panel 1 (A and C) and serum panel 2 (B and D) obtained with each assay.

Table 3. Diagnostic performance of the LPBE in comparison to the commercial SP-A ELISA. The diagnostic values of the two different ELISAs for serum panel 4 (n = 148) were computed based on the assumption that the VNT was the standard reference test (gold standard) using contingency table analysis.

ELISA method	Positive		Negative		Fisher* P value	Diagnostic value**
	No. of true	No. of false	No. of true	No. of false		Con (%)	Sen % (95% CI)	Spe % (95% CI)	PPV % (95% CI)	NPV % (95% CI)	K value (95% CI)
LPBE	138	1	7	2	< 0.0001	97.3	99.3 (96.1-99.9)	77.8 (39.9-97.2)	98.6 (94.9-99.8)	87.5 (47.6-99.7)	0.74 (0.49-0.98)
PrioCHECK^TM FMDV type A ELISA	48	91	7	2	0.2724	37.2	34.5 (26.7-43.1)	77.8 (39.9-97.2)	96 (86.3-99.5)	0.071 (0.03-14.1)	0.022 (0-0.074)

*Statistical significance of the interaction of the LPBE with the VNT.

**Con, concordance; Sen, sensitivity; Spe, specificity; PPV, positive predictive value; NPV, negative predictive value; k, Cohen’s kappa.

Table 4. ROC curve analysis of the LPBE versus the VNT (gold standard) on different categories of serum samples. The optimal cutoff value, diagnostic sensitivity, specificity, accuracy and AUC of the LPBE were computed with a 95% confidence interval. The computed AUC value was near 1, which indicates a higher discriminatory potential to distinguish between positive and negative samples with maximized specificity and sensitivity.

Source of sera (n)	No. of positive	No. of negative	Cutoff value** of the VNT (log10)	ROC curve analysis results^†
				Cutoff value of the LPBE (log10)	Sen (%)	Spe (%)	ACC (%)	AUC (95% CI)	p value
Panel 1 (n = 142)	97	45	1.2 (16-fold)	1.69	89.7	95.6	91.5	0.938 (0.898-0.978)	< 0.0001
Panel 2 (n = 94)	70	24		0.890	88.6	95.8	90.4	0.938 (0.890-0.986)	< 0.0001
Panel 4 (n = 148)	139	9		2.14	76.3	100	77.7	0.915 (0.856-0.974)	< 0.0001
Combined* (n =  384)	306	78		1.69	93.1	88.5	92.2	0.945 (0.923-0.968)	< 0.0001

*A total of 384 serum samples, including those from serum panel 1 (n =  142), panel 2 (n = 94) and panel 4 (n = 148).

**Criteria for classification between positive and negative sera. According to the OIE Manual of Diagnostic Test and Vaccines for Terrestrial Animals, a reciprocal log₁₀ neutralizing antibody (NAb) titer of 1.2 or more was considered positive.

^† Sen, sensitivity; Spe, specificity; ACC, accuracy; AUC, area under the ROC curve.