- Email: [email protected]
Evaluation of an ELISA for Dictyocaulus viviparus-specific antibodies in cattle
Veterinary Parasitology, 47 (1993) 301-314 Elsevier Science Publishers B.V., Amsterdam
301
Evaluation of an ELISA for Dictyocaulus viviparus-specific antibodies in cattle A.M. Tenter, A. B e l l m e r a n d T. S c h n i e d e r InstitutEr Parasitologie, Tieriirztliche Hochschule Hannover, Hannover, Germany (Accepted 18 November 1992) ABSTRACT Tenter, A.M., Bellmer, A. and Schnieder, T., 1993. Evaluation of an ELISA for Dictyocaulus viviparusspecific antibodies in cattle. Vet. Parasitol., 47: 301-314. An ELISA using a protein extract from adult Dictyocaulus viviparus as antigen was evaluated with respect to its quality and suitability as a diagnostic method for epidemiological studies on dictyocaulosis. In experimentally infected animals, D. viviparus-specific antibodies were first detected between 30 and 44 days post-infection (dpi) and persisted until 107-128 dpi (end of examination), i.e. eight or more weeks longer than the patency period of the infections. Based on parasitological and serological results given by 16 first-year grazing cattle infected with D. viviparus and gastrointestinal nematodes and on the results given by 32 first-year grazing cattle infected only with gastrointestinal nematodes under field conditions, the D. viviparus ELISA showed sensitivities ranging between 69 and > 99% and specificities ranging between 91 and 97% from 8 weeks after turnout until time of housing (21 weeks after turnout ). For this part of the grazing season, positive predictive values estimated for 33, 50, 67 and 80% prevalences of infection ranged between 79 and 99% and negative predictive values between 42 and > 99%. High values (> 90%) for all four test characteristics were observed from 13 to 17 weeks after turnout. Positive predictive factors ranged between 0.90 and 0.97 and negative predictive factors between 0.83 and > 0.99 from 8 weeks after turnout until the end of the grazing season and were still as high as 0.88 and 0.74, respectively, at the time of housing. Hence the ELISA appeared to be a very suitable diagnostic method for epidemiological studies on dictyocaulosis.
INTRODUCTION I n f e c t i o n s w i t h t h e c a t t l e l u n g w o r m D i c t y o c a u l u s viviparus a r e c o m m o n i n cattle on pasture throughout the temperate regions of Europe. The introduction of new anthelmintics, a live vaccine, and control schemes have helped to decrease outbreaks of disease during the last few decades, but non-immune first-year grazing cattle may still be subject to severe disease in wet summers in which large numbers of nematode larvae survive on the herbage. Such outbreaks of dictyocaulosis can cause significant economic losses in the cattle industry. Epidemiological data are needed to develop models that can predict the
Correspondence to: Dr. A.M. Tenter, Institut •r Parasitologie der Tier~irztlichen Hochschule, Biinteweg 17, W-3000 Hannover 71, Germany.
© 1993 Elsevier Science Publishers B.V. All rights reserved 0304-4017/93/$06.00
302
A.M. TENTER ETAL
risk of dictyocaulosis outbreaks under specific environmental conditions and to develop more effective measures to prevent or control the outbreak of disease. However, only a few studies have been carried out with the aim of investigating the epidemiology of D. viviparus infections (Boon et al., 1982, 1984b, 1986; Bos and Beekman, 1985; Wassail, 1991 ). Some of these studies involve the enzyme-linked immunosorbent assay (ELISA) as a diagnostic method to discriminate between D. viviparus-infected and non-infected cattle herds or between areas where D. viviparus is endemic and areas where it is not endemic (Boon et al., 1984b, 1986; Bos and Beekman, 1985). The antigen preparations employed in these diagnostic ELISAs have been derived from D. viviparus larvae ( L 3 and L4) or from adult worms. All of these antigens proved to be suitable to detect D. viviparus-specific antibodies in infected cattle (Marius et al., 1979; Boon et al., 1982; Bos and Beekman, 1985; Wassall, 1991 ). However, it has been suggested previously that while antibody titers against larval antigen reflect the infection doses, i.e. the number of larvae taken up by an animal, antibody titers against adult worm antigen reflect the level of infection, i.e. the worm burden of an infected animal (Bos and Beekman, 1985 ). In addition, it is difficult to obtain the large amount of antigen required for seroepidemiological studies from larval stages of D. viviparus. For these reasons, we intend to employ an ELISA using adult worm antigen for an epidemiological survey on D. viviparus infections in northern Germany. To study the epidemiology of infectious diseases, it is essential to select highly sensitive and specific tests that can discriminate between infected and non-infected animals. For final analysis of the study, animals examined by such diagnostic tests are then usually placed into one of the following four categories: true-positives; true-negatives; false-positives; false-negatives. However, to interpret the test results accurately and to place the animals under study into the correct category it is necessary to know some of the operating characteristics of the test used (Griner et al., 1981; Krbberling, 1982). Hence, the aim of the study presented here was to obtain some information on the operating characteristics of the D. viviparus ELISA and then to use this information in a large-scale epidemiological study on dictyocaulosis in northern Germany. MATERIALS AND METHODS
Source of sera and experimental animals Normal bovine sera were obtained from 52 male and 48 female cattle that were born and raised helminth-free at the Institut f'dr Parasitologie, at outstations of the Tier~irztliche Hochschule Hannover, or at the cattle breeding association of Lower Saxony (Rinderproduktion Niedersachsen). Serum
AN ELISA FOR DICTYOCAULUS VIVIPARUS-SPECIFIC ANTIBODIES IN CATTLE
303
samples were collected from these animals at 4-15 months of age before turnout on pasture. Faecal samples also taken from these animals were checked for the absence of helminth eggs and larvae (Baermann, 1917; Boray and Pearson, 1960; Schmidt, 1971 ). To obtain D. viviparus-positive sera and to monitor the detection of antibodies against D. viviparus in experimentally infected animals, six calves were raised helminth-free. At the age of 6 months, the calves were inoculated perorally with 1000 infective larvae (L3) on each of three consecutive days. Serum samples were collected from these animals twice a week over experimental periods of 72, 86 or 128 days. Faecal samples were taken twice a week from four of these animals and examined for the presence of D. viviparus larvae and the absence of eggs or larvae from helminths other than D. viviparus (Baermann, 1917; Schmidt, 1971 ). To monitor the detection ofD. viviparus-specific antibodies in cattle under field conditions, 16 first-year grazing cattle were turned out onto a pasture that was contaminated with infective stages ofD. viviparus as well as gastrointestinal nematodes. Another 32 first-year grazing cattle were turned out onto a pasture that was only contaminated with infective stages of gastrointestinal nematodes in order to check the ELISA for any false-positive results caused by cross-reactive antibodies (Marius et al., 1979; Boon et al., 1982). The latter pasture was checked to be free of infective stages of D. viviparus by examination of herbage samples and by tracer calves which proved to be free of lungworms upon slaughtering. Pastures were contaminated with lungworms through D. viviparus larvae shed by seeder calves 2-4 weeks before the onset of the study and with gastrointestinal nematodes through overwintered larvae shed by cattle in the previous year. Experimental animals were turned out in mid-May and remained on pasture until early October or November. Serum and faecal samples were collected from all animals in intervals of 2-3 weeks during the whole grazing season (21-25 weeks) and once a month thereafter from animals infected with D. viviparus (up to 35 weeks).
Preparation of antigen Antigen was prepared from adult D. viviparus collected from a calf 35 days after inoculation with 3000 infective larvae. The worms were minced mechanically and then homogenized in 20 m M Tris buffer, pH 8.0. Proteins were extracted for 15 h at 4 ° C with Tris buffer containing 1% sodium deoxycholate and protease inhibitors (100 U ml-1 aprotinin, 2 m M phenylmethylsulphonyl fluoride). The extract was centrifuged at 30 000 g for 45 min to remove insoluble components, distributed into aliquots, and stored at - 20 ° C. The protein concentration of this stock antigen preparation was estimated to be 1.2 mg ml-~ by a modified Lowry assay (Protein Assay Kit No. P 5656; Sigma Chemical Company) using bovine serum albumin as standard. Before use in the ELISA, the D. viviparus antigen preparation was further solubilized
304
A.M. TENTER ET AL.
by adding an equal volume of 8 M urea and incubating the mixture for 15 min at 22 oC. Determination of optimum assay conditions The optimal concentrations of the ELISA reagents were determined by chequerboard assays using serial dilutions of the D. viviparus antigen preparation, of a positive and a negative reference serum obtained from the experimental animals and of a rabbit anti-bovine IgG antibody conjugated to horseradish peroxidase (Affini-Pure T M Whole Molecule; Jackson Immunoresearch Laboratories). The conjugate was specific for the Fc-fragment of bovine IgG and hence reacted only with antibodies of this class of immunoglobulins. To optimize the saturation of free binding sites on the ELISA plates after their coating with D. viviparus antigen 0.1, 0.5, 1 and 2% w/v gelatine, 1, 2 and 5% w/v ovalbumin, 5% v/v horse serum, 5% v/v foetal calf serum and 5% w/v bovine serum albumin (BSA, Fraction V powder; Sigma Chemical Company) were tested as blocking agents. The same blocking agents, as well as polyoxysorbitan-monolaurate (Tween 20; Sigma Chemical Company), were also tested at various concentrations as supplements of the reagent buffer (0.01 M phosphate-buffered 0.15 M saline, pH 7.4) to decrease further non-specific background caused by binding of primary or secondary antibodies to the ELISA plate. Examination of test sera ELISA plates (Immuno-Plate Maxisorp F96; N U N C ) were incubated ( 15 h, 4 ° C) with 100/tl per well of the D. viviparus antigen preparation diluted 1/40 with coating buffer (15 mM carbonate 35 mM bicarbonate, pH 9.6) and were then washed (three times, 5 min) with PBS. The sensitized wells were post-coated ( 1 h, 37 ° C) with 100/A of 5% BSA in PBS and then washed (three times, 5 min) with PBS containing 0.05% v/v Tween 20 (PBS-Tween). Test sera were diluted in 2-fold steps from 1/10 to 1/1280 with 1% BSA in PBS-Tween. The antigen-coated wells were incubated ( 1 h, 37 °C) with 100 #1 of each serum sample and then washed as before. Conjugate was diluted 1/ 2560 with 1% BSA in PBS-Tween, 100/zl was added to each well, and the plates were incubated and washed as before. The substrate solution used was 0.04% chromogen (o-phenylene-diamine dihydrochloride; Sigma Chemical Company) in 50 mM phosphate 25 mM citric acid (pH 5.0) containing 0.012% hydrogen peroxide. The plates were incubated ( 10 min, 22 ° C ) with 50/zl of the substrate solution in the dark and the reaction was stopped by the addition of an equal volume of 2.5 M sulphuric acid. Optical densities (O.D.) were measured at a wavelength of 490 nm against a control well that had not received serum or conjugate. To account for inter-
AN ELISA FOR DICTYOCA ULUS VIVIPARUS-SPECIFIC ANTIBODIES IN CATTLE
305
assay variation, a 1/ 10 dilution of a D. viviparus-specific reference serum was included in all assays performed and the measured O.D. values were converted into indices using the formula: Index = O.D. (test serum)/ O.D. (D. viviparusreferenceserum)
Evaluation of the ELISA All sera were examined twice in two independent assays. To calculate the correlation between the results obtained by ELISA and those obtained by faecal examination, we used regression analyses that were done with a scientific graph software program (Sigma-Plot; Jandel Scientific). This program uses the coefficient of determination R 2 for the assessment of correlation. Sensitivity, specificity and positive and negative predictive values were calculated as described by Griner et al. ( 1981 ) using the following formulae and definitions: Sensitivity equals true-positives/(true-positives+ false-negatives); the probability that the assay will be positive when the disease is present. Specificity equals true-negatives/(false-positives + true-negatives); the probability that the assay will be negative when the disease is not present. Positive predictive value equals true-positives/(true-positives+false-positives); the probability that the disease is present when the assay is positive. Negative predictive value equals true-negatives/(false-negatives+true-negatives ); the probability that the disease is not present when the assay is negative. Positive and negative predictive factors were calculated following K/Sbberling et al. (1984) using the formulae: Positive predictive factor equals sensitivity/(sensitivity plus 1 minus specificity). Negative predictive factor equals specificity/(specificity plus 1 minus sensitivity). RESULTS
Optimization of the ELISA The chequerboard assays were evaluated by calculating the differences between O.D. values measured for the positive and negative reference sera at each of the antigen, serum and conjugate dilutions tested. This evaluation determined the concentration of D. viviparus antigen to be optimal above 7.5 /tg protein per well (Fig. 1 ). To account for batch-to-batch variation of the D. viviparus antigen preparation, twice this amount of protein ( 15/tg per well, i.e. antigen dilution 1/40) was chosen as antigen concentration for the examination of test sera. Saturation of free binding sites on the ELISA plates was most effective when 5% BSA fraction V powder was used as blocking agent. Background reactions on the ELISA plates could be reduced further by supplementing the reagent buffer used for serum and conjugate incubations
306
A.M. TENTERET AL. 2.0~
~>~ cd3
1.5
0.0
•
(1)
1.0
//
L~
0
0.5
/7
q/ •
0.0 .... 0.1
.t,
"
z
/
£,
', ............... 1.0 10.0
tlg
"
100.0
Protein
Fig. 1. Determination of optimal concentration ofD. viviparus antigen. Data are shown for a 1/ 10 dilution of D. viviparus-specific reference serum ( O ) , a 1/ 10 dilution of D. viviparus-negalive reference serum ( A ) and the difference between them ( • ) . Conjugate was used at a dilution of 1/2560. The arrow indicates the antigen concentration chosen for the examination of test sera.
with 1% BSA fraction V powder and 0.05% Tween 20. The optimal dilution of conjugate was found to be 1/2560. The cut-off value between negative and positive ELISA results was calculated from the index values measured for the 100 normal bovine sera at a serum dilution of 1/ 10. This calculation rendered values of0.18 for the arithmetic mean and 0.11 for the standard deviation of reactions given by this group of sera. To minimize the amount of false-positive results, the arithmetic mean plus four standard deviations (0.62) of these reactions was arbitrarily chosen as cut-off value for the discrimination between positive and negative ELISA results. Serum titers were defined as being the last serum dilution showing an index value equal to or greater than this cut-off value ( > 0.62 ).
Detection of antibodies in experimentally infected animals Figure 2 demonstrates the detection of D. viviparus larvae and of IgG antibodies against D. viviparus in experimentally infected cattle. Larvae were first detected between 18 and 22 days post-infection (dpi) and remained detectable until 59-75 dpi. Seroconversion was observed between 30 and 44 dpi. Maximum titers of 1/40 to 1/ 160 occurred between 37 and 68 dpi. Antibody titers decreased to 1/20 and 1/ 10 thereafter and persisted until 107-128 dpi (i.e. end of examination). Linear regression analysis showed no significant correlation between faecal counts of larvae and ELISA results for this group of animals.
AN ELISAFORDICTYOCAULUS VIVIPARUS-SPECIFICANTIBODIESIN CATTLE lOO
°
307
I i
i
•
\oO
/
0.1+ ~'
0
~ ©
O.Oli,
o o
/
1
"
0
~
30
200[:
B
.+.::::° ~
//
•
104
~
+
~%',~~ o o vv•
•*¢•
20
,\AO •
*,*,,".
•
•
-
• "-~ -+
/
EL O) ~f2-"
12O
.
° +•el 0 0
90
60
/
2
,
1 J-
o
/ /
/
/ - - "~-
v',~
so Days
60 after
90
120
infection
Fig. 2. (A) Detection of D. viviparus larvae and (B) of IgG antibodies to D. viviparus in six cattle ( O , O , A, A, ~ , • ) infected experimentally•
Detection of antibodies in animals infected under fieM conditions The detection of D. viviparus larvae and of IgG antibodies against D. viviparus in eight cattle grazing pasture that was contaminated with infective stages ofD. viviparus and of gastrointestinal nematodes is shown in Figs. 3 (A) and 3 (B). Seven of these cattle acquired infections with lungworms and gastrointestinal nematodes (Ostertagia spp. and Cooperia spp. ) as demonstrated by faecal examination. One animal excreted eggs of gastrointestinal nematodes only. Eggs or larvae of helminths other than lungworms or gastrointestinal nematodes were not detected in any of the eight animals. Average egg counts varied between 30 and 410 eggs g-t faeces from 6 weeks after turnout onwards, with a maximum 11 weeks after turnout. D. viviparus larvae were detected between 3 and 11 weeks after turnout (June-July). By contrast, D. viviparus-specific antibodies were first detected 6-8 weeks after turnout (June-July) and reached maximum titers from 1/20 to 1/320 between 8 and 15 weeks after turnout (July-August). Antibody titers persisted until 17 weeks after turnout to the time of housing (September-November) and were still detectable in three of the eight animals 10 weeks after housing (January).
308
A.M. TENTER ET AL. lOO
o (D
lO :,
v
0 ½.
1
L
o~
o.1 o c] >
/
'<<
O.Ol 0
/
o.0ol ,~ 50
o
100
150
200
250
5OO
] I
.+-_
IO0 5O
(D 0 k_ O_
7o
•
•
z,
t
•
ml
7
•
I
I
t
10
•
•
•
el
ii
ip
•
•
•
t
•
•
•
II
ii
t.
•
5
(1) Ek:
............. 0
50
oo
,i
15o
2o0
5OO i
25o I
i
o o C~
10
•
m
•
5
I
] •11:
0
l:l:
"~ - }
i
50
Days
after
~}-L}
1O0
II ~ l Z - J
:
150
turnout
200
on
250
posture
Fig. 3. (A) Detection of D. viviparus larvae and (B) of D. viviparus-specific antibodies in eight cattle ( © , O , ~ , A , V , V , ~ , O ) infected with D. viviparus and gastrointestinal nematodes under field conditions. (C) ELISA results obtained for eight cattle infected only with gastrointestinal nematodes under field conditions: ( [ ] ) seven animals in which cross-reactive antibodies were absent; (11) one animal showing cross-reactive antibodies.
Again, there was no significant correlation between faecal counts of larvae and ELISA results as found by linear regression analysis. Figure 3 (C) shows the ELISA results obtained for eight cattle grazing pasture that was contaminated with infective stages of gastrointestinal nematodes, but not of D. viviparus. All of these animals acquired infections with gastrointestinal nematodes (Ostertagia spp. and Cooperia spp. ) as demon-
AN ELISA FOR
309
DICTYOCAULUS VIVIPARUS-SPECIFIC A N T I B O D I E S I N C A T T L E 0 o~ o0
c5
c:5
o0 o
~
~
~
~
~
c5
c~ 0
e4
o
•~
o
0 o
~
o
~
~
d ~ d
o o o o
c5
ox
Y.
0
c5
e~
8 "4"4
0
<
c~
88~ 0
~.o o~
o
¢~ o
'.o
o o
o d
I ~1
~1
r-- o r~ o dd~dd
I
'q. o
~o
v
v
aaN ~.~ "~a .~
< b"
m
m
~
Z
e~
310
A.M. TENTER ET AL.
strated by faecal examination. No eggs or larvae of other helminths were detected in these animals. Average egg counts varied between 50 and 450 eggs g-1 faeces from 3 weeks after turnout onwards, with a m a x i m u m 8 weeks after turnout. All faecal samples taken from this group of cattle were negative for D. viviparus larvae. However, one of the animals showed infrequent positive reactions with D. viviparus antigen in the ELISA with a m a x i m u m titer of 1/40 8 weeks after turnout (July). The sensitivities, specificities and positive and negative predictive factors calculated for the D. viviparus ELISA over the whole grazing season are shown in Table 1. For these calculations, all 16 cattle grazing pasture contaminated with D. viviparus as well as gastrointestinal nematodes were regarded as positive and all 32 cattle grazing pasture contaminated only with gastrointestinal nematodes were regarded as negative. Positive and negative predictive values were calculated for 33, 50, 67 and 80% prevalences of infection with D. viviparus using data of 32, 16, 8 and 4 control animals, respectively. In total, high values ( > 90%) for all test characteristics were observed from 13 to 17 weeks after turnout (August-September). DISCUSSION
Thus far, the only operating characteristics determined in another study evaluating the ELISA as diagnostic test for infections with D. viviparus have been sensitivity and specificity (Boon et al., 1982 ). These two characteristics are required to determine the accuracy of the test and to decide whether or not the ELISA is suitable for selection as a test procedure in diagnosis or screening (Griner et al., 1981 ). The knowledge of the sensitivity of the ELISA under study provides information on the likelihood that the test result will be positive when the animal tested is infected with D. viviparus, while the knowledge of the specificity of the ELISA provides information on the likelihood that the test result will be negative when the animal is not infected with D. viviparus. In previous studies as well as in the study presented here, the ELISA proved to be a very sensitive test for D. viviparus infections in cattle. Consistent with a previous study on animals continuously infected with various doses of infective D. viviparus larvae (Boon et al., 1984a), ELI SA results in the present study became positive about 2 weeks later than parasitological parameters. Seroconversion in experimentally infected cattle occurred 1-2 weeks after infections became patent, but D. viviparus-specific antibodies persisted up to 8 or more weeks longer than the patency period of the infections. Likewise, although we observed seroconversion up to 3 weeks later than the onset of faecal excretion of larvae in cattle infected under field conditions, D. viviparusspecific antibodies remained detectable for at least 6 weeks, in some cattle even for more than 24 weeks, after the end of patency.
AN ELISA FOR DICTYOCA ULUS VIVIPARUS-SPECIFIC ANTIBODIES IN CATTLE
31 1
Because most cattle in northern Germany are usually kept on pasture during the s u m m e r months it is very unlikely that infections with D. viviparus would be the only helminth infections acquired during the grazing season. At least it can be expected that infections with gastrointestinal nematodes would also be contracted. To reflect these natural conditions as much as possible, we based the calculations of the operating characteristics of the D. viviparus ELISA on data obtained from animals that had mixed infections with D. viviparus and gastrointestinal nematodes rather than data from animals that had monoinfections with D. viviparus. Likewise, the control group used in these calculations did not consist of helminth-free cattle, but of cattle infected with gastrointestinal nematodes. The level of sensitivity or specificity of an ELISA can be influenced by a cut-off value which is chosen arbitrarily. The choice of a high cut-off value usually results in higher specificity but lower sensitivity of the test, whereas the choice of a low cut-off value results in the reverse. Other authors showed that cross-reactions do occur between crude antigen preparations from adult lungworms and antibodies directed against gastrointestinal nematodes (Boon et al., 1982 ) and the infrequency of low antibody titers to D. viviparus antigen observed in three of the control animals examined here also suggests that these reactions are caused by cross-reactive antibodies against gastrointestinal nematodes rather than antibodies specific for lungworms. Infections with gastrointestinal nematodes are frequent infections in cattle on pasture and antigenic cross-reactions due to these parasites may, therefore, interfere significantly with the seroepidemiological determination of areas endemic or not endemic for lungworms. To minimize such interference, we decided to choose the arithmetic mean plus four standard deviations of the reactions given by the negative control group as the cut-off value between negative and positive ELISA results. Although these conditions may appear very stringent, they resulted in only moderately high false-negative rates ( _<6% between 13 and 17 weeks after turnout on pasture; 19-31% thereafter), but were able to keep false-positive rates as low as -< 9% over the whole period of examination (21 weeks ). As outlined above, sensitivity is calculated from the group of infected animals, while specificity is calculated from the group of non-infected control animals. However, when animals from the field are examined serologically, it is usually not known which of these animals are truly infected and which are truly non-infected. Therefore, when seroepidemiological studies are to be carried out, the questions asked are not the ones to which the answers are reflected in sensitivity and specificity, but instead: What is the likelihood that an infection with D. viviparus is present when the test result is positive and what is the likelihood that an infection with D. viviparus is absent when the test result is negative? To answer these questions accurately, it is important to know the positive and negative predictive values of the ELISA (Griner et
312
A.M. TENTER ET AL.
al., 1981; K~Sbberling, 1982 ) and we are not aware that such calculations have been carried out in any of the studies on D. viviparus infections published so far. While sensitivity and specificity are constant operating characteristics of the ELISA that remain unchanged as long as cut-off values and sampling conditions are not altered (Boon et al., 1982), predictive values are dependent on the pre-test probability of infection in the group of animals tested (Griner et al., 1981; KSbberling, 1982; K/Sbberling et al., 1984). Although respiratory diseases in cattle are often observed in northern Germany, it is not known in how many of these cases D. viviparus is present and may be the causative agent of the disease. However, studies in The Netherlands using the ELISA estimated 14-57% of non-grazing milk cows and 75-91% of cattle herds grazing pasture to be infected with lungworms (Boon et al., 1984b, 1986; Bos and Beekman, 1985). We therefore estimated the predictive values for prevalences of D. viviparus infection ranging from 33 to 80%. These calculations showed the positive predictive values of the ELISA to range between 79 and > 99% and the negative predictive values to range between 20 and > 99%. Between 13 and 17 weeks after turnout on pasture, both positive and negative predictive values were 80% or higher (84-99% for positive predictive values, 8 0 - > 99% for negative predictive values). Hence, this time would appear as the optimal sampling time for seroepidemiological surveys as 8 0 - > 99% of the positive and negative test results would reflect true-positives or true-negatives, respectively. Another option for collecting epidemiological data on D. viviparus infections would be to sample cattle herds at the time of housing, as a large number of other data may also be obtained easily at this time (Boon et al., 1986). For prevalences of infection of 50% or higher, the positive predictive values at the time of housing were still 88-97%. However, the relatively low sensitivity of the ELISA observed at the time of housing (69%) resulted in lower negative predictive values ranging between 42 and 74% for the same prevalences of infection. To account for these lower predictive values, an option would be to interpret seroepidemiological results obtained at the time of housing on the basis of the average test result of a herd, as has already been shown by other authors (Boon et al., 1982, 1984b; Bos and Beekman, 1985), and not on the basis of the results given by individual animals. For example, to account for a negative predictive value of 42% (pre-test prevalence 80%), if three animals of one herd are sampled and one of these three animals gives a positive ELISA result, the whole herd should be regarded positive. Because we have no information on the prevalence ofD. viviparus infection in northern Germany at present, we also calculated the positive and negative predictive factors of the ELISA. These test characteristics are independent of the prevalence and describe the diagnostic value of a test with respect to the profit of predictive accuracy gained by the test result (K~Sbbeding et al., 1984).
AN ELISA FOR DICTYOCA ULUS I~TV1PARUS-SPECIFIC ANTIBODIES IN CATTLE
313
The D. viviparus ELISA described here showed high positive and negative predictive factors of > 0.90 and > 0.83, respectively, from 8 weeks after turnout onwards and still had a positive predictive factor of 0.88 and a negative predictive factor of 0.74 at the time of housing. Hence, the ELISA appeared to be a very suitable diagnostic method both for detecting and for excluding infections with D. viviparus. When the predictive factor (c) is known, it is possible to calculate positive predictive values (PPV) for any prevalence (p) using the formula: P P V = p c / [ p c + ( 1 - p ) ( 1 - c ) ], and negative predictive values (NPV) using the formula: N P V = ( 1 - p ) c ' /[ ( 1 - p ) c ' +p( 1 - c ' ) ] (K6bberling et al., 1984). Hence, the knowledge of the predictive factors of the D. viviparus ELISA will now enable us to calculate its predictive values for any prevalence of D. viviparus infection in cattle populations, and will thereby provide valuable information on the correct interpretation of results obtained in seroepidemiological surveys. ACKNOWLEDGEMENTS
We wish to thank E. Brennecke and I. Billerbeck for technical assistance. The study was supported by a research grant from Lower Saxony.
REFERENCES Baermann, G., 1917. Eine einfache Methode zur Auffindung von Ankylostomum-(Nematoden)-Larven in Erdproben. Tijdschr. Diergeneeskd., 57:131-137. Boon, J.H., Kloosterman, A. and Van den Brink, R., 1982. The incidence of Dictyocaulus viviparus infections in cattle in The Netherlands. I. The Enzyme Linked Immunosorbent Assay as a diagnostic tool. Vet. Q., 4:155-160. Boon, J.H., Kloosterman, A. and Breukink, M., 1984a. Parasitological, serological and clinical effects of continuous graded levels of Dictyocaulus viviparus inoculations in calves. Vet. Parasitol., 16: 261-272. Boon, J.H., Kloosterman, A. and Van der Lende, T., 1984b. The incidence of Dictyocaulus viviparus infections in cattle in The Netherlands. II. Survey of sera collected in the field. Vet. Q., 6: 13-17. Boon, J.H., Ploeger, H.W. and Raaymakers, A.J., 1986. Seroepidemiological survey of Dictyocaulus viviparus infections in first-season grazing calves in The Netherlands. Vet. Rec., 119: 475-479. Boray, J.C. and Pearson, J.G., 1960. The anthelmintic efficiency oftetrachlorodifluorethane in sheep infested with Fasciola hepatica. Aust. Vet. J., 36:331-337. Bos, H.J. and Beckman, J., 1985. Serodiagnosis of lungworm infection in calves using ELISA. Proc. Joint IABS/WHO Symp. Diagnostics and Vaccines for Parasitic Diseases. S. Karger, Basel, Develop. Biol. Standard., 62: 45-52. Griner, P.F., Mayewski, R.J., Mushlin, A.I. and Greenland, P., 1981. Selection and interpretation of diagnostic tests and procedures. Ann. Int. Med., 94: 553-600. Kt~bberling, J., 1982. Der pr~idiktive Wert diagnostischer MaBnahmen. Dtsch. Med. Wochenschr., 107: 591-595. K/3bberling, J., Richter, K. and Tillil, H., 1984. The predictive f a c t o r - a method to simplify
314
A.M.TENTERET AL.
Bayes' formula and its application to diagnostic procedures. Klin. Wochenschr., 62: 586592. Marius, V., Bernard, S., Raynaud, J.P., Pery, P. and Luffau, G., 1979. Dictyocaulus viviparus in calves: Quantitation of antibody activities in sera and respiratory secretions by immunoenzymatic analysis. Ann. Rech. Vet., 10: 55-63. Schmidt, U., 197 i. Parasitologische Kotuntersuchung durch ein neues Verdfinnungsverfahren. Tier~irztl. Umsch., 26: 229-230. Wassail, D.A., 1991. Use of an ELISA for serodiagnosis of parasitic bronchitis in cattle. Vet. Rec., 129: 353-355.
301
Evaluation of an ELISA for Dictyocaulus viviparus-specific antibodies in cattle A.M. Tenter, A. B e l l m e r a n d T. S c h n i e d e r InstitutEr Parasitologie, Tieriirztliche Hochschule Hannover, Hannover, Germany (Accepted 18 November 1992) ABSTRACT Tenter, A.M., Bellmer, A. and Schnieder, T., 1993. Evaluation of an ELISA for Dictyocaulus viviparusspecific antibodies in cattle. Vet. Parasitol., 47: 301-314. An ELISA using a protein extract from adult Dictyocaulus viviparus as antigen was evaluated with respect to its quality and suitability as a diagnostic method for epidemiological studies on dictyocaulosis. In experimentally infected animals, D. viviparus-specific antibodies were first detected between 30 and 44 days post-infection (dpi) and persisted until 107-128 dpi (end of examination), i.e. eight or more weeks longer than the patency period of the infections. Based on parasitological and serological results given by 16 first-year grazing cattle infected with D. viviparus and gastrointestinal nematodes and on the results given by 32 first-year grazing cattle infected only with gastrointestinal nematodes under field conditions, the D. viviparus ELISA showed sensitivities ranging between 69 and > 99% and specificities ranging between 91 and 97% from 8 weeks after turnout until time of housing (21 weeks after turnout ). For this part of the grazing season, positive predictive values estimated for 33, 50, 67 and 80% prevalences of infection ranged between 79 and 99% and negative predictive values between 42 and > 99%. High values (> 90%) for all four test characteristics were observed from 13 to 17 weeks after turnout. Positive predictive factors ranged between 0.90 and 0.97 and negative predictive factors between 0.83 and > 0.99 from 8 weeks after turnout until the end of the grazing season and were still as high as 0.88 and 0.74, respectively, at the time of housing. Hence the ELISA appeared to be a very suitable diagnostic method for epidemiological studies on dictyocaulosis.
INTRODUCTION I n f e c t i o n s w i t h t h e c a t t l e l u n g w o r m D i c t y o c a u l u s viviparus a r e c o m m o n i n cattle on pasture throughout the temperate regions of Europe. The introduction of new anthelmintics, a live vaccine, and control schemes have helped to decrease outbreaks of disease during the last few decades, but non-immune first-year grazing cattle may still be subject to severe disease in wet summers in which large numbers of nematode larvae survive on the herbage. Such outbreaks of dictyocaulosis can cause significant economic losses in the cattle industry. Epidemiological data are needed to develop models that can predict the
Correspondence to: Dr. A.M. Tenter, Institut •r Parasitologie der Tier~irztlichen Hochschule, Biinteweg 17, W-3000 Hannover 71, Germany.
© 1993 Elsevier Science Publishers B.V. All rights reserved 0304-4017/93/$06.00
302
A.M. TENTER ETAL
risk of dictyocaulosis outbreaks under specific environmental conditions and to develop more effective measures to prevent or control the outbreak of disease. However, only a few studies have been carried out with the aim of investigating the epidemiology of D. viviparus infections (Boon et al., 1982, 1984b, 1986; Bos and Beekman, 1985; Wassail, 1991 ). Some of these studies involve the enzyme-linked immunosorbent assay (ELISA) as a diagnostic method to discriminate between D. viviparus-infected and non-infected cattle herds or between areas where D. viviparus is endemic and areas where it is not endemic (Boon et al., 1984b, 1986; Bos and Beekman, 1985). The antigen preparations employed in these diagnostic ELISAs have been derived from D. viviparus larvae ( L 3 and L4) or from adult worms. All of these antigens proved to be suitable to detect D. viviparus-specific antibodies in infected cattle (Marius et al., 1979; Boon et al., 1982; Bos and Beekman, 1985; Wassall, 1991 ). However, it has been suggested previously that while antibody titers against larval antigen reflect the infection doses, i.e. the number of larvae taken up by an animal, antibody titers against adult worm antigen reflect the level of infection, i.e. the worm burden of an infected animal (Bos and Beekman, 1985 ). In addition, it is difficult to obtain the large amount of antigen required for seroepidemiological studies from larval stages of D. viviparus. For these reasons, we intend to employ an ELISA using adult worm antigen for an epidemiological survey on D. viviparus infections in northern Germany. To study the epidemiology of infectious diseases, it is essential to select highly sensitive and specific tests that can discriminate between infected and non-infected animals. For final analysis of the study, animals examined by such diagnostic tests are then usually placed into one of the following four categories: true-positives; true-negatives; false-positives; false-negatives. However, to interpret the test results accurately and to place the animals under study into the correct category it is necessary to know some of the operating characteristics of the test used (Griner et al., 1981; Krbberling, 1982). Hence, the aim of the study presented here was to obtain some information on the operating characteristics of the D. viviparus ELISA and then to use this information in a large-scale epidemiological study on dictyocaulosis in northern Germany. MATERIALS AND METHODS
Source of sera and experimental animals Normal bovine sera were obtained from 52 male and 48 female cattle that were born and raised helminth-free at the Institut f'dr Parasitologie, at outstations of the Tier~irztliche Hochschule Hannover, or at the cattle breeding association of Lower Saxony (Rinderproduktion Niedersachsen). Serum
AN ELISA FOR DICTYOCAULUS VIVIPARUS-SPECIFIC ANTIBODIES IN CATTLE
303
samples were collected from these animals at 4-15 months of age before turnout on pasture. Faecal samples also taken from these animals were checked for the absence of helminth eggs and larvae (Baermann, 1917; Boray and Pearson, 1960; Schmidt, 1971 ). To obtain D. viviparus-positive sera and to monitor the detection of antibodies against D. viviparus in experimentally infected animals, six calves were raised helminth-free. At the age of 6 months, the calves were inoculated perorally with 1000 infective larvae (L3) on each of three consecutive days. Serum samples were collected from these animals twice a week over experimental periods of 72, 86 or 128 days. Faecal samples were taken twice a week from four of these animals and examined for the presence of D. viviparus larvae and the absence of eggs or larvae from helminths other than D. viviparus (Baermann, 1917; Schmidt, 1971 ). To monitor the detection ofD. viviparus-specific antibodies in cattle under field conditions, 16 first-year grazing cattle were turned out onto a pasture that was contaminated with infective stages ofD. viviparus as well as gastrointestinal nematodes. Another 32 first-year grazing cattle were turned out onto a pasture that was only contaminated with infective stages of gastrointestinal nematodes in order to check the ELISA for any false-positive results caused by cross-reactive antibodies (Marius et al., 1979; Boon et al., 1982). The latter pasture was checked to be free of infective stages of D. viviparus by examination of herbage samples and by tracer calves which proved to be free of lungworms upon slaughtering. Pastures were contaminated with lungworms through D. viviparus larvae shed by seeder calves 2-4 weeks before the onset of the study and with gastrointestinal nematodes through overwintered larvae shed by cattle in the previous year. Experimental animals were turned out in mid-May and remained on pasture until early October or November. Serum and faecal samples were collected from all animals in intervals of 2-3 weeks during the whole grazing season (21-25 weeks) and once a month thereafter from animals infected with D. viviparus (up to 35 weeks).
Preparation of antigen Antigen was prepared from adult D. viviparus collected from a calf 35 days after inoculation with 3000 infective larvae. The worms were minced mechanically and then homogenized in 20 m M Tris buffer, pH 8.0. Proteins were extracted for 15 h at 4 ° C with Tris buffer containing 1% sodium deoxycholate and protease inhibitors (100 U ml-1 aprotinin, 2 m M phenylmethylsulphonyl fluoride). The extract was centrifuged at 30 000 g for 45 min to remove insoluble components, distributed into aliquots, and stored at - 20 ° C. The protein concentration of this stock antigen preparation was estimated to be 1.2 mg ml-~ by a modified Lowry assay (Protein Assay Kit No. P 5656; Sigma Chemical Company) using bovine serum albumin as standard. Before use in the ELISA, the D. viviparus antigen preparation was further solubilized
304
A.M. TENTER ET AL.
by adding an equal volume of 8 M urea and incubating the mixture for 15 min at 22 oC. Determination of optimum assay conditions The optimal concentrations of the ELISA reagents were determined by chequerboard assays using serial dilutions of the D. viviparus antigen preparation, of a positive and a negative reference serum obtained from the experimental animals and of a rabbit anti-bovine IgG antibody conjugated to horseradish peroxidase (Affini-Pure T M Whole Molecule; Jackson Immunoresearch Laboratories). The conjugate was specific for the Fc-fragment of bovine IgG and hence reacted only with antibodies of this class of immunoglobulins. To optimize the saturation of free binding sites on the ELISA plates after their coating with D. viviparus antigen 0.1, 0.5, 1 and 2% w/v gelatine, 1, 2 and 5% w/v ovalbumin, 5% v/v horse serum, 5% v/v foetal calf serum and 5% w/v bovine serum albumin (BSA, Fraction V powder; Sigma Chemical Company) were tested as blocking agents. The same blocking agents, as well as polyoxysorbitan-monolaurate (Tween 20; Sigma Chemical Company), were also tested at various concentrations as supplements of the reagent buffer (0.01 M phosphate-buffered 0.15 M saline, pH 7.4) to decrease further non-specific background caused by binding of primary or secondary antibodies to the ELISA plate. Examination of test sera ELISA plates (Immuno-Plate Maxisorp F96; N U N C ) were incubated ( 15 h, 4 ° C) with 100/tl per well of the D. viviparus antigen preparation diluted 1/40 with coating buffer (15 mM carbonate 35 mM bicarbonate, pH 9.6) and were then washed (three times, 5 min) with PBS. The sensitized wells were post-coated ( 1 h, 37 ° C) with 100/A of 5% BSA in PBS and then washed (three times, 5 min) with PBS containing 0.05% v/v Tween 20 (PBS-Tween). Test sera were diluted in 2-fold steps from 1/10 to 1/1280 with 1% BSA in PBS-Tween. The antigen-coated wells were incubated ( 1 h, 37 °C) with 100 #1 of each serum sample and then washed as before. Conjugate was diluted 1/ 2560 with 1% BSA in PBS-Tween, 100/zl was added to each well, and the plates were incubated and washed as before. The substrate solution used was 0.04% chromogen (o-phenylene-diamine dihydrochloride; Sigma Chemical Company) in 50 mM phosphate 25 mM citric acid (pH 5.0) containing 0.012% hydrogen peroxide. The plates were incubated ( 10 min, 22 ° C ) with 50/zl of the substrate solution in the dark and the reaction was stopped by the addition of an equal volume of 2.5 M sulphuric acid. Optical densities (O.D.) were measured at a wavelength of 490 nm against a control well that had not received serum or conjugate. To account for inter-
AN ELISA FOR DICTYOCA ULUS VIVIPARUS-SPECIFIC ANTIBODIES IN CATTLE
305
assay variation, a 1/ 10 dilution of a D. viviparus-specific reference serum was included in all assays performed and the measured O.D. values were converted into indices using the formula: Index = O.D. (test serum)/ O.D. (D. viviparusreferenceserum)
Evaluation of the ELISA All sera were examined twice in two independent assays. To calculate the correlation between the results obtained by ELISA and those obtained by faecal examination, we used regression analyses that were done with a scientific graph software program (Sigma-Plot; Jandel Scientific). This program uses the coefficient of determination R 2 for the assessment of correlation. Sensitivity, specificity and positive and negative predictive values were calculated as described by Griner et al. ( 1981 ) using the following formulae and definitions: Sensitivity equals true-positives/(true-positives+ false-negatives); the probability that the assay will be positive when the disease is present. Specificity equals true-negatives/(false-positives + true-negatives); the probability that the assay will be negative when the disease is not present. Positive predictive value equals true-positives/(true-positives+false-positives); the probability that the disease is present when the assay is positive. Negative predictive value equals true-negatives/(false-negatives+true-negatives ); the probability that the disease is not present when the assay is negative. Positive and negative predictive factors were calculated following K/Sbberling et al. (1984) using the formulae: Positive predictive factor equals sensitivity/(sensitivity plus 1 minus specificity). Negative predictive factor equals specificity/(specificity plus 1 minus sensitivity). RESULTS
Optimization of the ELISA The chequerboard assays were evaluated by calculating the differences between O.D. values measured for the positive and negative reference sera at each of the antigen, serum and conjugate dilutions tested. This evaluation determined the concentration of D. viviparus antigen to be optimal above 7.5 /tg protein per well (Fig. 1 ). To account for batch-to-batch variation of the D. viviparus antigen preparation, twice this amount of protein ( 15/tg per well, i.e. antigen dilution 1/40) was chosen as antigen concentration for the examination of test sera. Saturation of free binding sites on the ELISA plates was most effective when 5% BSA fraction V powder was used as blocking agent. Background reactions on the ELISA plates could be reduced further by supplementing the reagent buffer used for serum and conjugate incubations
306
A.M. TENTERET AL. 2.0~
~>~ cd3
1.5
0.0
•
(1)
1.0
//
L~
0
0.5
/7
q/ •
0.0 .... 0.1
.t,
"
z
/
£,
', ............... 1.0 10.0
tlg
"
100.0
Protein
Fig. 1. Determination of optimal concentration ofD. viviparus antigen. Data are shown for a 1/ 10 dilution of D. viviparus-specific reference serum ( O ) , a 1/ 10 dilution of D. viviparus-negalive reference serum ( A ) and the difference between them ( • ) . Conjugate was used at a dilution of 1/2560. The arrow indicates the antigen concentration chosen for the examination of test sera.
with 1% BSA fraction V powder and 0.05% Tween 20. The optimal dilution of conjugate was found to be 1/2560. The cut-off value between negative and positive ELISA results was calculated from the index values measured for the 100 normal bovine sera at a serum dilution of 1/ 10. This calculation rendered values of0.18 for the arithmetic mean and 0.11 for the standard deviation of reactions given by this group of sera. To minimize the amount of false-positive results, the arithmetic mean plus four standard deviations (0.62) of these reactions was arbitrarily chosen as cut-off value for the discrimination between positive and negative ELISA results. Serum titers were defined as being the last serum dilution showing an index value equal to or greater than this cut-off value ( > 0.62 ).
Detection of antibodies in experimentally infected animals Figure 2 demonstrates the detection of D. viviparus larvae and of IgG antibodies against D. viviparus in experimentally infected cattle. Larvae were first detected between 18 and 22 days post-infection (dpi) and remained detectable until 59-75 dpi. Seroconversion was observed between 30 and 44 dpi. Maximum titers of 1/40 to 1/ 160 occurred between 37 and 68 dpi. Antibody titers decreased to 1/20 and 1/ 10 thereafter and persisted until 107-128 dpi (i.e. end of examination). Linear regression analysis showed no significant correlation between faecal counts of larvae and ELISA results for this group of animals.
AN ELISAFORDICTYOCAULUS VIVIPARUS-SPECIFICANTIBODIESIN CATTLE lOO
°
307
I i
i
•
\oO
/
0.1+ ~'
0
~ ©
O.Oli,
o o
/
1
"
0
~
30
200[:
B
.+.::::° ~
//
•
104
~
+
~%',~~ o o vv•
•*¢•
20
,\AO •
*,*,,".
•
•
-
• "-~ -+
/
EL O) ~f2-"
12O
.
° +•el 0 0
90
60
/
2
,
1 J-
o
/ /
/
/ - - "~-
v',~
so Days
60 after
90
120
infection
Fig. 2. (A) Detection of D. viviparus larvae and (B) of IgG antibodies to D. viviparus in six cattle ( O , O , A, A, ~ , • ) infected experimentally•
Detection of antibodies in animals infected under fieM conditions The detection of D. viviparus larvae and of IgG antibodies against D. viviparus in eight cattle grazing pasture that was contaminated with infective stages ofD. viviparus and of gastrointestinal nematodes is shown in Figs. 3 (A) and 3 (B). Seven of these cattle acquired infections with lungworms and gastrointestinal nematodes (Ostertagia spp. and Cooperia spp. ) as demonstrated by faecal examination. One animal excreted eggs of gastrointestinal nematodes only. Eggs or larvae of helminths other than lungworms or gastrointestinal nematodes were not detected in any of the eight animals. Average egg counts varied between 30 and 410 eggs g-t faeces from 6 weeks after turnout onwards, with a maximum 11 weeks after turnout. D. viviparus larvae were detected between 3 and 11 weeks after turnout (June-July). By contrast, D. viviparus-specific antibodies were first detected 6-8 weeks after turnout (June-July) and reached maximum titers from 1/20 to 1/320 between 8 and 15 weeks after turnout (July-August). Antibody titers persisted until 17 weeks after turnout to the time of housing (September-November) and were still detectable in three of the eight animals 10 weeks after housing (January).
308
A.M. TENTER ET AL. lOO
o (D
lO :,
v
0 ½.
1
L
o~
o.1 o c] >
/
'<<
O.Ol 0
/
o.0ol ,~ 50
o
100
150
200
250
5OO
] I
.+-_
IO0 5O
(D 0 k_ O_
7o
•
•
z,
t
•
ml
7
•
I
I
t
10
•
•
•
el
ii
ip
•
•
•
t
•
•
•
II
ii
t.
•
5
(1) Ek:
............. 0
50
oo
,i
15o
2o0
5OO i
25o I
i
o o C~
10
•
m
•
5
I
] •11:
0
l:l:
"~ - }
i
50
Days
after
~}-L}
1O0
II ~ l Z - J
:
150
turnout
200
on
250
posture
Fig. 3. (A) Detection of D. viviparus larvae and (B) of D. viviparus-specific antibodies in eight cattle ( © , O , ~ , A , V , V , ~ , O ) infected with D. viviparus and gastrointestinal nematodes under field conditions. (C) ELISA results obtained for eight cattle infected only with gastrointestinal nematodes under field conditions: ( [ ] ) seven animals in which cross-reactive antibodies were absent; (11) one animal showing cross-reactive antibodies.
Again, there was no significant correlation between faecal counts of larvae and ELISA results as found by linear regression analysis. Figure 3 (C) shows the ELISA results obtained for eight cattle grazing pasture that was contaminated with infective stages of gastrointestinal nematodes, but not of D. viviparus. All of these animals acquired infections with gastrointestinal nematodes (Ostertagia spp. and Cooperia spp. ) as demon-
AN ELISA FOR
309
DICTYOCAULUS VIVIPARUS-SPECIFIC A N T I B O D I E S I N C A T T L E 0 o~ o0
c5
c:5
o0 o
~
~
~
~
~
c5
c~ 0
e4
o
•~
o
0 o
~
o
~
~
d ~ d
o o o o
c5
ox
Y.
0
c5
e~
8 "4"4
0
<
c~
88~ 0
~.o o~
o
¢~ o
'.o
o o
o d
I ~1
~1
r-- o r~ o dd~dd
I
'q. o
~o
v
v
aaN ~.~ "~a .~
< b"
m
m
~
Z
e~
310
A.M. TENTER ET AL.
strated by faecal examination. No eggs or larvae of other helminths were detected in these animals. Average egg counts varied between 50 and 450 eggs g-1 faeces from 3 weeks after turnout onwards, with a m a x i m u m 8 weeks after turnout. All faecal samples taken from this group of cattle were negative for D. viviparus larvae. However, one of the animals showed infrequent positive reactions with D. viviparus antigen in the ELISA with a m a x i m u m titer of 1/40 8 weeks after turnout (July). The sensitivities, specificities and positive and negative predictive factors calculated for the D. viviparus ELISA over the whole grazing season are shown in Table 1. For these calculations, all 16 cattle grazing pasture contaminated with D. viviparus as well as gastrointestinal nematodes were regarded as positive and all 32 cattle grazing pasture contaminated only with gastrointestinal nematodes were regarded as negative. Positive and negative predictive values were calculated for 33, 50, 67 and 80% prevalences of infection with D. viviparus using data of 32, 16, 8 and 4 control animals, respectively. In total, high values ( > 90%) for all test characteristics were observed from 13 to 17 weeks after turnout (August-September). DISCUSSION
Thus far, the only operating characteristics determined in another study evaluating the ELISA as diagnostic test for infections with D. viviparus have been sensitivity and specificity (Boon et al., 1982 ). These two characteristics are required to determine the accuracy of the test and to decide whether or not the ELISA is suitable for selection as a test procedure in diagnosis or screening (Griner et al., 1981 ). The knowledge of the sensitivity of the ELISA under study provides information on the likelihood that the test result will be positive when the animal tested is infected with D. viviparus, while the knowledge of the specificity of the ELISA provides information on the likelihood that the test result will be negative when the animal is not infected with D. viviparus. In previous studies as well as in the study presented here, the ELISA proved to be a very sensitive test for D. viviparus infections in cattle. Consistent with a previous study on animals continuously infected with various doses of infective D. viviparus larvae (Boon et al., 1984a), ELI SA results in the present study became positive about 2 weeks later than parasitological parameters. Seroconversion in experimentally infected cattle occurred 1-2 weeks after infections became patent, but D. viviparus-specific antibodies persisted up to 8 or more weeks longer than the patency period of the infections. Likewise, although we observed seroconversion up to 3 weeks later than the onset of faecal excretion of larvae in cattle infected under field conditions, D. viviparusspecific antibodies remained detectable for at least 6 weeks, in some cattle even for more than 24 weeks, after the end of patency.
AN ELISA FOR DICTYOCA ULUS VIVIPARUS-SPECIFIC ANTIBODIES IN CATTLE
31 1
Because most cattle in northern Germany are usually kept on pasture during the s u m m e r months it is very unlikely that infections with D. viviparus would be the only helminth infections acquired during the grazing season. At least it can be expected that infections with gastrointestinal nematodes would also be contracted. To reflect these natural conditions as much as possible, we based the calculations of the operating characteristics of the D. viviparus ELISA on data obtained from animals that had mixed infections with D. viviparus and gastrointestinal nematodes rather than data from animals that had monoinfections with D. viviparus. Likewise, the control group used in these calculations did not consist of helminth-free cattle, but of cattle infected with gastrointestinal nematodes. The level of sensitivity or specificity of an ELISA can be influenced by a cut-off value which is chosen arbitrarily. The choice of a high cut-off value usually results in higher specificity but lower sensitivity of the test, whereas the choice of a low cut-off value results in the reverse. Other authors showed that cross-reactions do occur between crude antigen preparations from adult lungworms and antibodies directed against gastrointestinal nematodes (Boon et al., 1982 ) and the infrequency of low antibody titers to D. viviparus antigen observed in three of the control animals examined here also suggests that these reactions are caused by cross-reactive antibodies against gastrointestinal nematodes rather than antibodies specific for lungworms. Infections with gastrointestinal nematodes are frequent infections in cattle on pasture and antigenic cross-reactions due to these parasites may, therefore, interfere significantly with the seroepidemiological determination of areas endemic or not endemic for lungworms. To minimize such interference, we decided to choose the arithmetic mean plus four standard deviations of the reactions given by the negative control group as the cut-off value between negative and positive ELISA results. Although these conditions may appear very stringent, they resulted in only moderately high false-negative rates ( _<6% between 13 and 17 weeks after turnout on pasture; 19-31% thereafter), but were able to keep false-positive rates as low as -< 9% over the whole period of examination (21 weeks ). As outlined above, sensitivity is calculated from the group of infected animals, while specificity is calculated from the group of non-infected control animals. However, when animals from the field are examined serologically, it is usually not known which of these animals are truly infected and which are truly non-infected. Therefore, when seroepidemiological studies are to be carried out, the questions asked are not the ones to which the answers are reflected in sensitivity and specificity, but instead: What is the likelihood that an infection with D. viviparus is present when the test result is positive and what is the likelihood that an infection with D. viviparus is absent when the test result is negative? To answer these questions accurately, it is important to know the positive and negative predictive values of the ELISA (Griner et
312
A.M. TENTER ET AL.
al., 1981; K~Sbberling, 1982 ) and we are not aware that such calculations have been carried out in any of the studies on D. viviparus infections published so far. While sensitivity and specificity are constant operating characteristics of the ELISA that remain unchanged as long as cut-off values and sampling conditions are not altered (Boon et al., 1982), predictive values are dependent on the pre-test probability of infection in the group of animals tested (Griner et al., 1981; KSbberling, 1982; K/Sbberling et al., 1984). Although respiratory diseases in cattle are often observed in northern Germany, it is not known in how many of these cases D. viviparus is present and may be the causative agent of the disease. However, studies in The Netherlands using the ELISA estimated 14-57% of non-grazing milk cows and 75-91% of cattle herds grazing pasture to be infected with lungworms (Boon et al., 1984b, 1986; Bos and Beekman, 1985). We therefore estimated the predictive values for prevalences of D. viviparus infection ranging from 33 to 80%. These calculations showed the positive predictive values of the ELISA to range between 79 and > 99% and the negative predictive values to range between 20 and > 99%. Between 13 and 17 weeks after turnout on pasture, both positive and negative predictive values were 80% or higher (84-99% for positive predictive values, 8 0 - > 99% for negative predictive values). Hence, this time would appear as the optimal sampling time for seroepidemiological surveys as 8 0 - > 99% of the positive and negative test results would reflect true-positives or true-negatives, respectively. Another option for collecting epidemiological data on D. viviparus infections would be to sample cattle herds at the time of housing, as a large number of other data may also be obtained easily at this time (Boon et al., 1986). For prevalences of infection of 50% or higher, the positive predictive values at the time of housing were still 88-97%. However, the relatively low sensitivity of the ELISA observed at the time of housing (69%) resulted in lower negative predictive values ranging between 42 and 74% for the same prevalences of infection. To account for these lower predictive values, an option would be to interpret seroepidemiological results obtained at the time of housing on the basis of the average test result of a herd, as has already been shown by other authors (Boon et al., 1982, 1984b; Bos and Beekman, 1985), and not on the basis of the results given by individual animals. For example, to account for a negative predictive value of 42% (pre-test prevalence 80%), if three animals of one herd are sampled and one of these three animals gives a positive ELISA result, the whole herd should be regarded positive. Because we have no information on the prevalence ofD. viviparus infection in northern Germany at present, we also calculated the positive and negative predictive factors of the ELISA. These test characteristics are independent of the prevalence and describe the diagnostic value of a test with respect to the profit of predictive accuracy gained by the test result (K~Sbbeding et al., 1984).
AN ELISA FOR DICTYOCA ULUS I~TV1PARUS-SPECIFIC ANTIBODIES IN CATTLE
313
The D. viviparus ELISA described here showed high positive and negative predictive factors of > 0.90 and > 0.83, respectively, from 8 weeks after turnout onwards and still had a positive predictive factor of 0.88 and a negative predictive factor of 0.74 at the time of housing. Hence, the ELISA appeared to be a very suitable diagnostic method both for detecting and for excluding infections with D. viviparus. When the predictive factor (c) is known, it is possible to calculate positive predictive values (PPV) for any prevalence (p) using the formula: P P V = p c / [ p c + ( 1 - p ) ( 1 - c ) ], and negative predictive values (NPV) using the formula: N P V = ( 1 - p ) c ' /[ ( 1 - p ) c ' +p( 1 - c ' ) ] (K6bberling et al., 1984). Hence, the knowledge of the predictive factors of the D. viviparus ELISA will now enable us to calculate its predictive values for any prevalence of D. viviparus infection in cattle populations, and will thereby provide valuable information on the correct interpretation of results obtained in seroepidemiological surveys. ACKNOWLEDGEMENTS
We wish to thank E. Brennecke and I. Billerbeck for technical assistance. The study was supported by a research grant from Lower Saxony.
REFERENCES Baermann, G., 1917. Eine einfache Methode zur Auffindung von Ankylostomum-(Nematoden)-Larven in Erdproben. Tijdschr. Diergeneeskd., 57:131-137. Boon, J.H., Kloosterman, A. and Van den Brink, R., 1982. The incidence of Dictyocaulus viviparus infections in cattle in The Netherlands. I. The Enzyme Linked Immunosorbent Assay as a diagnostic tool. Vet. Q., 4:155-160. Boon, J.H., Kloosterman, A. and Breukink, M., 1984a. Parasitological, serological and clinical effects of continuous graded levels of Dictyocaulus viviparus inoculations in calves. Vet. Parasitol., 16: 261-272. Boon, J.H., Kloosterman, A. and Van der Lende, T., 1984b. The incidence of Dictyocaulus viviparus infections in cattle in The Netherlands. II. Survey of sera collected in the field. Vet. Q., 6: 13-17. Boon, J.H., Ploeger, H.W. and Raaymakers, A.J., 1986. Seroepidemiological survey of Dictyocaulus viviparus infections in first-season grazing calves in The Netherlands. Vet. Rec., 119: 475-479. Boray, J.C. and Pearson, J.G., 1960. The anthelmintic efficiency oftetrachlorodifluorethane in sheep infested with Fasciola hepatica. Aust. Vet. J., 36:331-337. Bos, H.J. and Beckman, J., 1985. Serodiagnosis of lungworm infection in calves using ELISA. Proc. Joint IABS/WHO Symp. Diagnostics and Vaccines for Parasitic Diseases. S. Karger, Basel, Develop. Biol. Standard., 62: 45-52. Griner, P.F., Mayewski, R.J., Mushlin, A.I. and Greenland, P., 1981. Selection and interpretation of diagnostic tests and procedures. Ann. Int. Med., 94: 553-600. Kt~bberling, J., 1982. Der pr~idiktive Wert diagnostischer MaBnahmen. Dtsch. Med. Wochenschr., 107: 591-595. K/3bberling, J., Richter, K. and Tillil, H., 1984. The predictive f a c t o r - a method to simplify
314
A.M.TENTERET AL.
Bayes' formula and its application to diagnostic procedures. Klin. Wochenschr., 62: 586592. Marius, V., Bernard, S., Raynaud, J.P., Pery, P. and Luffau, G., 1979. Dictyocaulus viviparus in calves: Quantitation of antibody activities in sera and respiratory secretions by immunoenzymatic analysis. Ann. Rech. Vet., 10: 55-63. Schmidt, U., 197 i. Parasitologische Kotuntersuchung durch ein neues Verdfinnungsverfahren. Tier~irztl. Umsch., 26: 229-230. Wassail, D.A., 1991. Use of an ELISA for serodiagnosis of parasitic bronchitis in cattle. Vet. Rec., 129: 353-355.