Evaluation of antigen-detection ELISAs in Mali

O. Diall, Y. Sanogo and K. Tounkara

Laboratoire Central Veterinaire, B.P. 2295, Bamako, Mali

Summary

The efficacy of a tsetse-control programme can be measured through tsetse-density monitoring on the one hand and through infection rate and packed cell volume (PCV) monitoring in cattle on the other hand. The objective of this study was to determine the suitability of the antigen-ELISA (Ag-ELISA) as an additional method for assessing the efficacy of a tsetse-control programme based on the use of traps and screens impregnated with deltamethrine. The buffy coat test was used as a reference to evaluate the available Ag-ELISA (using two kits: one from the International Livestock Research Institute (ILRI) and another from the International Atomic Energy Agency (IAEA)). Herd monitoring was carried out in 8 localities: 4 in the test area and 4 in an untreated area, and the results were compared over 12 months. The apparent tsetse density in the treated area was reduced by nearly 95% after one month of control, and few flies were caught thereafter. A similar trend was observed for rates of infection in cattle which were reduced from 6 to 2% in the treated area but increased from 3 to 10% in the untreated control area. At the beginning of the experiment, mean PCV was the same (29.4%) for cattle in both treated and untreated areas. After 12 months of operation, the mean PCV in the treated area was higher than that in the untreated area. Positive antigen prevalence did not correlate with trypanosomal infection rate in cattle, nor with apparent tsetse density. The two Ag-ELISAs suffered from low sensitivity as measured on a parasite positive population for both Trypanosoma congolense and T. vivax and the Ag-ELISA should not be recommended for trypanosomosis monitoring on its own.

Introduction

There is a move in Mali towards tsetse and trypanosomosis control using insecticide-impregnated traps and screens. There is a need for an improved trypanosomosis diagnosis for monitoring such control operations, as the currently available tests have limited sensitivity. The antigen-ELISA is a candidate test. In 1988 ILRAD proposed an Ag-ELISA for trypanosomosis diagnosis based on monoclonal antibodies specific for the three main species of cattle trypanosomosis (T. vivax, T. congolense and T. brucei) (Nantulya 1990; Nantulya and Lindqvist 1989; Masake and Nantulya 1991). The test was validated in ten African laboratories. The different results (Diall et al 1993) called for refinements of the test by IAEA.

The aim of the present study was twofold:

1. Evaluation of the Ag-ELISA as a tool for assessing the impact of tsetse control programmes on the incidence of trypanosomal infections in livestock.
2. Evaluation of the sensitivity and the specificity of the Ag-ELISA, and comparison of ILRI's and IAEA's kits.

Materials and methods

The sector of Niena in Mali was chosen for the pilot tsetse-control programme. It covers 500 km2 with 56,000 cattle. Two species of flies are prevalent: (Glossina palpalis gambiensis, G. tachinoides) alongside the River Bagoé and its affluents. The annual rain fall is 800–1000 mm.

Tsetse monitoring

Impregnated traps were deployed alongside the River Bagoé and its tributaries within the limits of the control area. Entomological surveys were carried out every 1–2 months using biconical traps to monitor the apparent density of tsetse flies at selected points on the river banks. For comparison purposes the same studies were conducted at other sites selected in a neighbouring untreated area.

Cattle infection monitoring and packed cell volume

Fifty animals were selected in each of eight villages. Four of the villages were located in the test area, while the four others were located in the non-treated (control) area. These animals were ear-tagged and bled every two months for parasite and antigen monitoring. The parasite monitoring was done using the buffy coat technique (BCT). Sera were collected and stored at -20°C for antigen-detection.

Mean packed red cell volume was compared at the beginning of the experiment and 12 months later. One hundred sera from trypanosome-negative (using BCT) animals with low packed cell volume (PCV) (PCV ≥ 25%) and 100 sera from trypanosome-negative (using BCT) animals with normal PCV (PCV ≥ 25%) were tested by Ag-ELISA in order to determine if there were significantly more antigen-positive cases in the low PCV group than in the normal PCV group for T. congolense and T. vivax.

Data analysis

To determine the effect of the tsetse control on trypanosomal prevalence in cattle, the results obtained in the treated area were first compared over different sampling periods and then with those in the control area over the same periods. Correlations were also made between tsetse fly density and trypanosomosis prevalence in both treated and control areas on one hand and with antigen positivity on the other hand.

Results

Entomolosgy

Before the beginning of tsetse control, the apparent tsetse densities were relatively high for both tsetse species and in both treated and untreated areas. One month after the beginning of the control, the reduction in apparent density was 93% for G. p. gambiensis and 98% for G. tachinoides (Figure 1). One year after the deployment of the targets hardly any flies were captured in the treated area. Apparent tsetse densities of G. p. gambiensis were 11.6 and 16.7 in untreated and treated areas, respectively, in March 1993, one month before the start of tsetse control, and 3.25 and 0, respectively, 13 months later in April 1994. Corresponding apparent densities for G. tachinoides were 2.4 and 3.7, before tsetse control, and 1.90 and 0, respectively, 12 months after tsetse control.

Cattle infections and PCV

At the beginning of the project the prevalence of trypanosomal infections in cattle in the selected herds was 6.2% in the treated area and 3.4% in the untreated area. It should be noted, however, that before the study animals were being regularly treated by the owners using either Berenil® or Trypamidium®, and this may account for the low trypanosomal infection rates in the cattle at the beginning of the experiment. Once the study started, however, animals were not treated independently by the owners and the detection of the parasite was the only indication for treatment. During the study, fluctuations were observed in infection rates in both areas.

Figure 1.  Variations of the densities in the treated and untreated areas.

One year after the beginning of the study, the prevalence of trypanosomal infections in the treated area was lower (2%) than in the untreated area (9%) (Table 1).

With regard to the PCV, the results of the antigen-detection ELISA tests using both kits are shown in Table 2 for samples from cattle with normal PCVs (≥ 25%) and low PCVs (<25%).

Table 1. Summary of parasitological and haematological results in eight villages.

^†Prev: trypanosomal prevalence

Table 2. Results of the antigen-detection ELISA test using IAEA and ILRI kits.

c² analysis was performed to determine if the observed differences in numbers detected positive between normal and low PCV samples were statistically significant (Table 3).

Table 3. c² values for comparing numbers detected positive between normal and low PCV samples.  

Table 4. Results of the antigen-ELISA test for BCT positive sera using AIEA and ILRI kits. 

^†Percentage in parentheses.

The results indicated that only for T. brucei was the difference statistically significant for both kits.

Correlation between parasitological and haematological results

Monthly parasitological and haematological parameters were recorded. The data (Figures 2A and 2B) demonstrate no correlation between antigen and parasitaemia results in either area. PCVs in animals treated and untreated areas were similar until month 11 but were lower in animals in untreated areas in month 12.

Evaluation of sensitivity and specificity

Of 1201 samples collected and tested using the BCT, 80 were positive to T. congolense and 50 to T. vivax. Using ILRI's and IAEA's reagents, these sera were further examined and the sensitivities and specificities, and the positive and negative predictive values (using the three monoclonals for each kit) were calculated and compared to BCT results. The c2 test was used to compare the performance of the ILRI and IAEA tests. The specificity of the Ag-ELISA was also determined on 596 sera collected from a trypanosome-free zone in Nara in Mali, but using only IAEA reagents.

The results of the evaluations are shown in Table 4 for IAEA and ILRI reagents, respectively.

Ag-ELISA test statistics on samples positive to T. congolense by BCT are shown in Table 5. The performance of both sets of reagents was poor when using monoclonal T. brucei and monoclonal T. vivax. The results given in Table 5 when monoclonal T. congolense was used show that ILRI's reagents detected more positive cases than did IAEA's reagents.

Figure 2A. Validations of the mean PCV, infection rates using the Ag-ELISA and BCT in treated areas.
Figure 2B.  Validations of the mean PCV, infection rates using the Ag-ELISA and BCT in untreated areas.

Table 5.  ELISA test results on samples positive to T. congolense by buffy coat technique.  

Table 6. Overall ELISA test performance compared to BCT (positive to T. congolense).  

Table 7. ELISA test results on samples (positive to T. vivax) by BCT.  

Table 6 gives the results of the overall comparison of the ELISA test to BCT (positive to T. congolense) assuming that sera positive to T. brucei and T. vivax monoclonals were false positives. The overall results showed that the ILRI reagents were more sensitive and had higher predictive values than the IAEA reagents.

ELISA test results on samples positive for T. vivax by BCT are shown in Table 7. The results showed very low sensitivities for both reagents for the T. brucei monoclonal. The results using T. congolense or T. vivax monoclonal again indicate low sensitivity of the test for both reagents.

Table 8 gives the results of the comparison of the ELISA test to BCT (positive to T. vivax) assuming that sera positive to T. brucei and T. congolense monoclonals are false positive. The overall results showed that ILRI's reagents are slightly more sensitive.

c² squared analysis was used to compare the performance of ILRI and IAEA kits, which detected 65/80 and 37/80 samples, respectively, positive to T. congolense of those detected positive to T. congolense by BCT, and 17/50 and 15/60 samples, respectively, positive to T. vivax of those detected positive to T. vivax by BCTs. The calculated value of 2 c²= 19.72 (P<0.001) demonstrated a large difference in performance of the two kits for BCT-detected T. congolense infections. The ILRI kit was more sensitive than the IAEA kit. However, there was no difference in performance of the two  kits in relation to T. vivax (c² = 0.68).

Table 8. ELISA test performance compared to BCT (positive to T. vivax).  

Table 9. Results of the evaluation of the specificity of Ag-ELISA from samples collected in a trypanosome-free zone.

Specificity of the antigen-detection ELISA

The specificity of the ELISA was evaluated in Seibersdorf (Austria) (D. Rebeski, personal communication) based on test results for 596 sera from Nara (Table 9).

High specificity was obtained with a cut-off point of 10% for all three species.

Discussion and conclusions

The estimation of the sensitivity and the specificity of a test requires two types of populations: one completely disease free, and one in which every animal is infected (Burges 1988). To fulfill these conditions in the present study the sera were collected from two different parts of the country: the first a trypanosome-free zone (Nara) and the second a trypanosome-infected area (Niena). One thousand, two hundred and one sera collected in Nara were used to estimate the sensitivity of the test and 596 sera from Niena were tested for the specificity of the assay.

Ease of application, cost and reliability are some of the important factors in the choice of a test. Other important factors are: which test is most sensitive, which test is most specific, are the tests independent? These criteria are important since tests that are high in sensitivity and/or specificity are ideally required (Burges 1988). The Ag-ELISA was evaluated using two sets of reagents: ILRI's and that of IAEA.

The following conclusions were made:

1. The specificity when measured on a negative population from a tsetse-free area was satisfactory for each trypanosomal species: 98–99% at a cut-off point of 10%.
2. Both kits (ILRI and IAEA ) suffered from low sensitivity when measured on parasite-positive populations for both T. congolense (respectively 81 and 46%) and T. vivax (respectively 34 and 25%).
3. There were significantly more antigen-positive cases (using the monoclonal T. brucei) in a low PCV population than in a normal PCV population for T. congolense.

In conclusion, the use of antigen-ELISA is not recommended for trypanosome monitoring due to its low sensitivity. Further, the prevalence of antigen-positive samples was found to be uncorrelated with the trypanosomal infection rate in cattle as determined by BCT.

References

Diall O., Diarra B. and Sanogo Y. 1993. Evaluation of monoclonal antibody-based antigen-detection immunoassays for the diagnosis of Trypanosoma congolense and Trypanosoma vivax infections in Mali. IAEA TecDoc 707:65–68.

Nantulya V.M. 1990. Trypanosomiasis in domestic animals: the problem of diagnosis. Revue Scientifique et Technique de I'Office International des Epizooties 9:357–367.

Nantulya V.M. and Lindqvist K.J. 1989. Antigen-detection enzyme immunoassays for diagnosis of Trypanosoma vivax, T. congolense and T. brucei infections in cattle. Tropical Medicine and Parasitology 40:267–272.

Masake R.A. and Nantulya V.M. 1991. Sensitivity of an antigen-detection enzyme immunoassay for diagnosis of Trypanosoma congolense infections in goats and cattle. Journal of Parasitology 77:231–236.

Burgess G.W. (ed). 1988. ELISA Technology in Diagnosis and Research. Graduate School of Tropical Veterinary Science, James Cook University of North Queensland, Townsville, Australia. 341 pp.