No CrossRef data available.
Article contents
A novel method for the isolation of gastro-intestinal nematode eggs that allows automated analysis of digital images of egg preparations and high throughput screening
Published online by Cambridge University Press: 28 November 2001
Abstract
A simple and robust method for the isolation of gastro-intestinal nematode eggs from faeces is described that uses both salt- and sugar solutions for flotation. Application of this ‘salt–sugar’; isolation method to large numbers of faecal samples of adult dairy cows indicates a 3- or 4-fold reduction in the proportion of e.p.g.-negative cows relative to studies that used other techniques for egg isolation. The procedure detects more eggs than the Wisconsin flotation method in replicate samples and in spiked egg-free faeces. The number of recovered eggs in spiked faecal samples is linear over a range of egg concentrations, and the transparent faecal preparations that result from the protocol can be stored as digital images which can be used as input for an efficient automated egg-counting procedure. The increased rate of processing of faeces combined with the large reduction of the percentage of e.p.g.-negative cows allows more accurate analysis of large numbers of adult or resistant animals for studies of nematode parasitism.
Information
- Type
- Research Article
- Information
- Copyright
- © 2002 Cambridge University Press
INTRODUCTION
The concentration of nematode eggs in faeces (eggs per gram of faeces; e.p.g.) is one of the most important parameters in parasitology. In contrast to other parasitological parameters such as worm counts, e.p.g. values can easily be determined for individual animals as well as for herds and, apart from being labour intensive, the assays that are commonly used are inexpensive. The methods that are used for the investigation of faeces such as the McMaster technique (MAFF, 1977) and Wisconsin flotation (Cox & Todd, 1962) are well established and have not changed much during the past decades. However, both methods have serious drawbacks for the determination of the e.p.g. of adult animals. The theoretical detection limit of the standard McMaster technique is 50 e.p.g. (Guttieres, Todd & Crowley, 1979; Burrows et al. 1980) and combined with the need for instantaneous microscopical examination this precludes its use for the study of large numbers of adult stock. The latter argument holds also true for the Wisconsin flotation method (Agneessens et al. 2000). Taken together, these factors prohibit the routine analysis of large numbers of faecal samples with low numbers of eggs using traditional methods.
Here we present a simple and robust, partially automated egg isolation protocol that permits high throughput screening of faeces with low concentrations of eggs. This method is compared to the Wisconsin sugar flotation method using egg-free faeces spiked with known numbers of gastro-intestinal nematode eggs as well as faecal samples from naturally infected cows.
MATERIALS AND METHODS
Faecal samples
Faeces from non-infected calves was used as negative faeces in spiking experiments. Eggs for spiking experiments were obtained from faeces of calves that had been artificially infected with 30000 infective larvae of Cooperia oncophora. The nematode egg isolation protocol described below was used to monitor naturally infected adult dairy cattle. To this end, a total of 1400 faecal samples were collected between June and August 2000 from approximately 300 commercial dairy farms that are distributed over The Netherlands. On most farms only 2 or 3 cows were sampled. Most cows had been on pasture from the end of April or the beginning of May. The faecal samples of the cows were collected by veterinarians and were processed at the Parasitology division at Utrecht University on the day of arrival.
Salt–sugar flotation (SSF)
One to three grams of faeces were suspended in 25 ml of a 13% NaCl solution (δsalt = 1.06). After the suspension was filtered over a coarse metal sieve the filtrate was spun for 5 min at 4000 g (RT). After addition of an equal volume of tapwater to the supernatant and mixing, the sample was spun for 1 min at 4000 g (RT). The supernatant was removed with a water pump, and the pellet was resuspended in 5 ml of a 17% sucrose solution. Upon centrifugation in a 15 ml centrifuge tube for 5 min at 4000 g (RT), the supernatant was diluted once with tap water. After thorough mixing the sample was spun again for 1 min at 4000 g (RT). After centrifugation, the supernatant was removed to reduce the final volume of the egg preparation to approximately 200 μl. The amount of faeces and volume of sugar and salt solutions used for flotation can be scaled up or down to suit the centrifuge capacity. Only the 2-fold dilutions need to be strictly adhered to.
Subsequently, 3 drops of saturated sucrose solution (≈70%w/v) were added to the 200 μl of egg preparation. After thorough mixing, the samples were transferred to a flat-bottom 96-well microtitre plate. Prior to this, the wells of the microtitre plate had been filled with 3 drops of mineral oil. The samples were pipetted from underneath the mineral oil to ensure a positive meniscus. Subsequently, samples were allowed to settle for at least 15 min during which eggs migrated to the top of the meniscus. In this way, the gastro-intestinal nematode eggs were concentrated at the top of the sugar solution. In the sugar solution, the eggs remain undamaged for at least 4 h.
The egg suspensions were scanned using a Coolpix 950 digital camera (Nikon) attached to a Nikon binocular (SMZ800) at 10× (automated counting; see below) or 30× magnification (manual counting). Scans were stored at normal density in JPEG format which resulted in files with a size of 350–400 kb. Most standard graphics programs are capable of importing these JPEG files.
To be able to further increase the speed of processing, an automated egg counting procedure was developed. For this procedure, a 24-well microtitre plate was used and a larger volume (700 μl) of saturated sucrose solution was added to the egg suspension. Mineral oil was not added to the wells. After thorough mixing, the scans of these suspensions in these larger wells were made at 10× microscopic magnification. Automated egg counts were obtained by importing the graphics into the image analysis package KS 400 (Carl Zeiss Vision, Germany). A dynamic discrimination procedure based on the local neighbourhood grey value (matrix size 31×31 pixels) was used to select objects in the image. These comprised gastro-intestinal nematode eggs, small straws, and other artifacts. The selection criteria for the detection of nematode eggs were determined using a trial and error approach using a sample of approximately 50 images of faecal samples that contained a wide range of eggs. The following identifiers of gastro-intestinal eggs were used. Objects smaller than 40×40 pixels and larger than 230×230 pixels were rejected. Finally, objects were rejected on the basis of a circular shape below 0.39 (e.g. pieces of straw) and above 0.93 (e.g. air bubbles).
The efficiency of the automated counts was evaluated by comparing the image of the egg preparations on a PC monitor to the same image that was overlaid with the particles recognized as gastro-intestinal nematode eggs by the algorithm on another monitor. Occasional false negatives were, if necessary, corrected by visual inspection on a computer screen. The number of false negatives was used to determine the efficiency of the automated counting procedure.
Wisconsin flotation method
The Wisconsin method for the isolation of gastro-intestinal nematode eggs involved 2 steps. First, 1 g of faeces was suspended in 25 ml of tap water, filtered over cheese-cloth, and resuspended in 10 ml of saturated sucrose. This suspension was transferred to a 10 ml centrifuge tube. Subsequently, a cover-slip was put on top of the tube and the suspension was spun for 5 min at 4000 g at room temperature (RT). During centrifugation, the gastro-intestinal nematode eggs were concentrated against the cover-slip that was removed from the tube for microscopical examination at 40× magnification. The centrifuge tube was refilled with saturated sucrose solution, a cover-slip was put onto the centrifuge tube, and the sample was centrifuged again. The total number of nematode eggs on the 2 cover-slips slides was determined.
Efficiency of the salt–sugar isolation method (SSF) versus Wisconsin sugar flotation
The efficiency of the SSF protocol was compared to the most widely used egg isolation protocol for analysis of faeces of adult cattle (Wisconsin flotation; Cox & Todd, 1962) using 2 approaches. First, the number of eggs recovered from a 1 g duplicate sample of the same faeces was determined by parallel processing using the SSF as well as the Wisconsin method. This was done for 31 faecal samples. Second, both methods were also used to determine the recovery after spiking an amount of egg-free faeces with different volumes of egg preparations with a known concentration of gastro-intestinal nematode eggs. The volumes that were added to the faeces ranged from 0.02 to 1 ml and from 0.01 to 0.8 ml for the SSF and the Wisconsin method, respectively. For the Wisconsin sugar flotation 0.93 to 2.17 g of faeces were spiked with 1.7, 3.4, 5.1, 6.8, 8.5, 17.0, 34.1, 68.1 and 136.3 eggs. For the SSF flotation 1.07 to 2.14 g of faeces were spiked with 3.4, 5.1, 8.5, 12.8, 17.0, 34.1, 51.1, 68.1, 85.2, 102.2, 136.3, and 170.3 eggs. For both flotation techniques 24 faecal samples were spiked.
RESULTS
Examples of digital images taken at 10× (automated counting) and 30× magnification (manual counting) are shown in Fig. 1. The clustering of gastro-intestinal nematode eggs at the top of the egg preparation that is overlaid with mineral oil (left) allowed accurate counts of eggs in the entire faecal sample if manual counting is preferred. Alternatively, if automated egg counts are desired, egg preparations were put in large wells that resulted in a dispersed pattern of eggs (Fig. 1). This method is recommended when egg concentrations are high.
Fig. 1. Examples of egg suspensions obtained using the 2 variants of the gastro-intestinal nematode egg isolation protocol. (A) Nematode egg isolation in a sucrose solution overlaid with mineral oil in a 96-well microtitre plate at 30× magnification. (B) Nematode egg isolation in a sucrose solution in 24-well microtitre plate at 10× magnification.
The number of eggs recovered from replicates of 1 g of faecal sample using the Wisconsin and the SSF method was compared using faecal samples in which the e.p.g. values ranged from 0 to 37 as determined with Wisconsin flotation and from 0 to 36 e.p.g. when using SSF. This range is typical for adult cattle. The SSF found significantly higher e.p.g. values than Wisconsin sugar flotation in replicate samples of faeces of adult cattle (n = 31; difference between the e.p.g. values averaged over 31 samples was 2.81, P<0.005; paired t-test). The median of the distribution using the SSF and Wisconsin method was 9 and 5 e.p.g., respectively.
The efficiency of the SSF egg isolation procedure was determined using the recovered number of eggs after spiking an egg-free faecal sample with a known number of eggs. The absence of eggs in a large number of egg-free faecal samples as determined using SSF indicated the absence of false positives. After spiking egg-free faecal samples with a range of egg numbers, the correlation between the expected and observed number of eggs was high (Fig. 2A; R2 = 0.95). The slope of the least squares linear regression is approximately 0.50, and the averaged percentage recovery over expected e.p.g. values was 36%.
Fig. 2. Correlation between the observed and expected e.p.g. value obtained by the SSF (A) and Wisconsin sugar flotation (B) method using egg-free faeces that were spiked with a known number of gastro-intestinal nematode eggs.
The slope of the regression line indicates a substantial loss of eggs during the SSF isolation procedure that is not surprising given that filtration and centrifugation are involved in the isolation protocol. In our hands, however, the accuracy of the Wisconsin flotation is even lower as determined by the first derivative of the least squares linear regression (0.25), and the percentage recovery averaged over all 24 isolates (24.5%). In addition, the higher regression coefficient of the SSF method relative to the Wisconsin technique (R2 = 0.84; Fig. 2B) suggests that the power of variance-based statistical analyses is higher when SSF is used. In conclusion, the good fit between the expected and recovered number of eggs indicates that precise estimates (Sokal & Rohlf, 1981) of the e.p.g. can be obtained with the SSF, and that this method is suited to obtain statistically reliable estimates of the e.p.g. over a range of egg concentrations.
The software that was developed for automated analyses of gastro-intestinal nematode eggs retrieved the majority of the eggs in the preparation using the settings described in the Materials and Methods section. The algorithm that was used to detect gastro-intestinal eggs included parameters such as sphericity and size of the eggs, and the surface that was used to scan the image of the egg preparation. Obtaining egg counts from a digital image took approximately 2 sec. Using these settings, all but a few eggs were detected by the algorithm. As a consequence, the regression coefficient between the observed number of eggs in the automated counting procedure and the actual number of eggs was very high (R2 = 0.98; Fig. 3).
Fig. 3. Correlation between the observed number of gastro-intestinal nematode eggs that was detected by the automated egg counting algorithm and the expected number of eggs as determined by visual examination of eggs.
The SSF was applied to faeces from 1400 naturally infected adult dairy cattle from commercial farms. The e.p.g. ranged from 0 to 168.88. The arithmetic mean of the e.p.g. values was 5.7, and the geometric mean (Smothers, Sun & Dayton, 1999) was 2.38. The frequency distribution of the e.p.g. values of these cows had a variance of 215.41, it was positively skewed (skewness 5.61) with kurtosis 39.58. While a positive skewness is typical for the distribution of e.p.g. values of gastro-intestinal nematodes in host populations, the variance and kurtosis of the e.p.g. distribution have not been reported before for large populations of dairy cattle. Finally, in the faecal samples of approximately 21% of the cows no eggs were detected.
DISCUSSION
The SSF method allows animals with low egg counts to be routinely investigated using standard equipment at a relatively high throughput. It has amply been demonstrated that due to long-term exposure to gastro-intestinal parasites, adult dairy cattle are (partially) immune to gastro-intestinal nematodes and as a result have a low egg output via the faeces. As a consequence, it is important to effectively analyse maximal amounts of faeces per host in order to avoid the need for extrapolation in estimating the e.p.g. Because the number of eggs in 1 g of faeces forms the basis of the e.p.g., the number of nematode eggs in an amount of faeces that is effectively screened has to be corrected. In particular when small amounts of faeces are examined, the multiplication factor renders e.p.g. estimates less precise. The SSF method involves successive flotation and concentration steps that lack the extrapolation that is typical for the standard and modified McMaster techniques. Instead, 1–3 g is effectively evaluated for the presence of gastro-intestinal nematode eggs which is much more than the modified McMaster that has also been used for adult dairy cattle (Thienpont, Rochette & Van Parijs, 1979; Agneessens et al. 2000). When comparing methods that do not require extrapolation, i.e. Wisconsin flotation versus SSF, the latter method detected more eggs. This was found with spiked samples as well as in parallel isolates of equal amounts of naturally infected faecal samples. These parallel isolates comprised only 2 and 3 negative samples with the SSF and the Wisconsin method, respectively, which indicates that the different numbers of eggs isolated with the 2 methods are due to a different efficiency of egg isolation. The retrieval of a related procedure that strongly resembles the Wisconsin method and that is referred to as a direct centrifugal flotation (DCF) technique by Levine et al. (1996) varied between 22 and 51% when Sheather's sugar solution was used. Thus, the efficiency of recovery of the related DCF procedure was in the range of percentages of recovery of the Wisconsin method as well as the SSF method. A very important difference between the DCF and the SSF techniques pertains to the number of faecal samples that can be processed. Using the SSF method e.p.g. values of more than 100 animals per day can routinely be determined. This is not possible when instantaneous microscopical examinations are required such as for Wisconsin sugar flotation, or, more generally, DCF methods.
Studies of the distribution of e.p.g. values of adult dairy cattle that were similar with regard to age and sex of the animals have been conducted recently (Agneessens et al. 2000; Borgsteede et al. 2000) and these studies form a suitable reference for the interpretation of our results. A possible difference among these studies and ours most likely results from health differences of the animals because the 2 former studies used slaughterhouse material. Although nematode infections need not be causal to health problems, a bad physical condition of a limited number of animals may explain the higher arithmetic e.p.g. mean as found by Agneessens et al. (2000) and Borgsteede et al. (2000). A few cows with very high egg counts have a big impact on the arithmetic mean (Smothers et al. 1999). The geometric mean of the e.p.g. values is less sensitive to outliers and is comparable among studies (2.4 in Borgsteede et al. (2000) but Agneessens et al. (2000) did not report the geometric mean of their e.p.g. values). The most striking difference between the studies, however, is the frequency of cows with very low e.p.g. values. For example, the percentage of e.p.g.-negative cows in our study is approximately 21%, whereas 86% and 70% has been reported by Agneessens et al. (2000) and Borgsteede et al. (2000), respectively. It is very unlikely that the difference of the proportion of e.p.g.-negative cows is caused by differing physiological conditions of the animals because these low e.p.g. values do not cause clinical symptoms. Moreover, in both studies a high percentage of cows was positive for gastro-intestinal nematode infection when larval cultures were examined. The drastically lower percentage of e.p.g.-negative cows obtained using the SSF method is therefore interpreted to be primarily due to an improved method of egg isolation relative to the modified McMaster method (Agneessens et al. 2000; Borgsteede et al. 2000). This substantiates the belief that the modified McMaster technique is not particularly suited for the examination of faeces of adult dairy cattle (Agneessens et al. 2000). The much lower number of faecal samples without eggs as found using the SSF method is expected to improve the power of statistical analyses because of the reduced number of null values and, more generally, more precise estimates (Sokal & Rohlf, 1981).
On the other hand, in the spiking experiments a limited number of samples with low egg concentrations seemed to have a lower observed e.p.g. with SSF than with the Wisconsin method. The regression line of SSF between 0 and 5 e.p.g. contains a substantial number of negatives, and perhaps somewhat more than the Wisconsin method. This probably resulted from random error of the number of eggs that was used for spiking because especially at these low concentrations the impact of random error is substantial. Although other studies of adult dairy cattle (Agneessens et al. 2000; Borgsteede et al. 2000) did not use the Wisconsin method, a substantial increase of the proportion of positive cows using SSF relative to these studies appears to support this interpretation.
Ordinary least-squares linear regression was used to compare the two methods of egg isolation that are subject to random error of both observed and expected e.p.g. (see the Materials and Methods section). Subject to the specification of the squared analytical error ratio, other regression methods that are designed for the comparison of analytical methods, may result in less systematic bias relative to ordinary least squares. However, we did not have duplicate determinations which are required for the estimation of the error ratio. Further, the standard deviation of residuals across a regression line is important for distinguishing whether the observed spread across the regression line is caused by analytical imprecision alone or whether sample-related effects also contribute. The relatively even spread of residuals along the regression line of SSF using ordinary least squares relative to the Wisconsin method suggests that deviations of the Wisconsin method are due to analytical imprecision (Stockl, Dewitte & Thienpont, 1998). Finally, in view of the high regression coefficients, different regression methods are likely to arrive at marginally different statistics.
The most important technical difference between the traditional isolation methods such as Wisconsin flotation and modified McMaster on the one hand, and the SSF method on the other hand, involves the concentration of the salt- and sugar solutions. McMaster and Wisconsin techniques use solutions that are near saturation (δsugar = 1.20–1.30; δsalt≈1.20). In contrast, in our protocol the salt- and sugar solutions are diluted (δsugar≈1.06; δsalt≈1.06) and the less dense solutions invariably resulted in cleaner preparations. This is expected when the difference of the density of solutions that are used for successive flotation and precipitation steps is narrowed (0.03 in salt–sugar flotation and 0.10–0.15 in saturated salt- and sugar solutions, respectively).
The most important advantage of the SSF method is the possibility to process large number of samples. Agneessens et al. (2000) suggested the use of Wisconsin sugar flotation to obtain more precise estimates of the e.p.g. but they considered this isolation method to be too labour intensive. The possibility to postpone the visual examination of egg preparations, and the use of an automated egg counting procedure allow an increased rate of processing of faecal samples. In addition, counting eggs is less tiring and time-consuming using a computer screen than with a microscope. Finally, the possibility of re-examination of egg images allows the use of other parameters of interest such as species-specific egg features and the re-examination of critical specimens. Taken together, we believe that the SSF protocol allows many samples to be processed with a high level of precision. These are necessary conditions for studies that involve examination of many animals in a relatively short time-span such as large surveys or studies that aim at identification of genetic components underlying nematode resistance in cattle.
This research was supported by the Technology Foundation STW, applied science division of NWO and the technology programme of the Ministry of Economic Affairs.
References
AGNEESSENS, J., CLAEREBOUT, E., DORNY, P., BORGSTEEDE, F. H. M. & VERCRUYSSE, J. (2000). Nematode parasitism in adult dairy cows in Belgium. Veterinary Parasitology 90, 83–92.CrossRefGoogle Scholar
BORGSTEEDE, F. H. M., TIBBEN, J., CORNELISSEN, J. B. W. J., AGNEESSENS, J. & GAASENBEEK, C. P. H. (2000). Nematode parasites of adult dairy cattle in The Netherlands. Veterinary Parasitology 89, 287–296.CrossRefGoogle Scholar
BURROWS, R. O., BEST, P. J. & PRESTON, J. M. (1980). Trichostrongylid egg output of dairy cows. Veterinary Record 107, 399–401.CrossRefGoogle Scholar
COX, D. D. & TODD, A. C. (1962). Survey of gastro-intestinal parasitism in Wisconsin dairy cattle. Journal of the American Veterinary Medical Association 141, 706–709.Google Scholar
GUTTIERES, V., TODD, A. C. & CROWLEY, J. W. (1979). Natural populations of helminths in Wisconsin dairy cows. Veterinary Medicine 74, 369–374.Google Scholar
LEVINE, N. D., MEHRA, K. N., CLARK, D. T. & AVES, I. J. (1960). A comparison of nematode egg counting techniques for cattle and sheep faeces. American Journal of Veterinary Research 21, 511–515.Google Scholar
MAFF, . (1977). Manual of Veterinary Parasitological Laboratory Techniques. Technical Bulletin 18. Her Majesty's Stationery Office, London, UK.
SOKAL, R. R. & ROHLF, F. J. (1981). Biometry. Freeman and Company, New York.
SMOTHERS, C. D., SUN, F. & DAYTON, A. D. (1999). Comparison of arithmetic and geometric means as measures of a central tendency in cattle nematode populations. Veterinary Parasitology 81, 211–224.CrossRefGoogle Scholar
STOCKL, D., DEWITTE, K. & THIENPONT, L. M. (1998). Validity of linear regression in method comparison studies: is it limited by the statistical model or the quality of the analytical input data? Clinical Chemistry 44, 2340–2346.Google Scholar
THIENPONT, D., ROCHETTE, F. & VAN PARIJS, O. F. J. (1979). Diagnosing Helminthiasis by Coprological Examination. Janssen Research Foundation, Beerse, Belgium.
Fig. 1. Examples of egg suspensions obtained using the 2 variants of the gastro-intestinal nematode egg isolation protocol. (A) Nematode egg isolation in a sucrose solution overlaid with mineral oil in a 96-well microtitre plate at 30× magnification. (B) Nematode egg isolation in a sucrose solution in 24-well microtitre plate at 10× magnification.
Fig. 2. Correlation between the observed and expected e.p.g. value obtained by the SSF (A) and Wisconsin sugar flotation (B) method using egg-free faeces that were spiked with a known number of gastro-intestinal nematode eggs.
Fig. 3. Correlation between the observed number of gastro-intestinal nematode eggs that was detected by the automated egg counting algorithm and the expected number of eggs as determined by visual examination of eggs.
You have
Access
- 33
- Cited by