Multicenter Evaluation of an ELISA for the Detection of Cryptosporidium spp. Antigen in Clinical Human Stool Samples

Human cryptosporidiosis remains underdiagnosed, and rapid/accurate diagnosis is of clinical importance. Diagnosis of the Cryptosporidium oocyst in stool samples by conventional microscopy is labor-intensive, time-consuming, and requires skillful experience. Thus, we aimed to evaluate the usefulness of a coproantigen enzyme-linked immunosorbent assay (ELISA) test in detecting Cryptosporidium spp. from fecal specimens. For this aim, we evaluated the performances of a commercial ELISA (CoproELISA Cryptosporidium kit, Savyon Diagnostics, Israel) for the detection of Cryptosporidium spp. in random clinical stool samples through a multicenter study. The sensitivity and specificity for coproantigen ELISA were 98.86% and 94.32%, respectively. The coproantigen ELISA results indicate that the simple, rapid, reliable, and standardized immunoassay test is sensitive and specific for routine diagnosis, and may be useful for large-scale epidemiological studies of cryptosporidiosis.


Introduction
Cryptosporidiosis, an emerging infectious disease of public health significance worldwide, is caused by the protozoan Cryptosporidium. Currently, twenty-three species and 61 valid genotypes of Cryptosporidium spp. have been described from a wide range of vertebrates, including humans, mammals, wildlife, domestic livestock, reptile, birds, amphibians, and fish, causing asymptomatic or mild-to-severe gastrointestinal disease in its host species [1].
Cryptosporidium parvum and the anthroponotic Cryptosporidium hominis are the major causes of human cryptosporidiosis. Although infection is asymptomatic, sometimes it causes diarrhea with a large number of oocysts present in the stools, as the acute infection resolves and the patient becomes asymptomatic, the number of oocysts decreases. Cryptosporidium can also cause chronic diarrhea, abdominal pain, weakness, weight loss, which can result in death in immunosuppressed individuals [2,3]. The diagnosis of cryptosporidiosis is usually made by microscopic detection of the parasite oocysts. However, this method is labor-intensive, time-consuming, requires skillful experience, and has low sensitivity [4,5]. Molecular biology techniques have become diagnostic tools that have been used to understand the epidemiology of Cryptosporidium spp. However, the accessibility to this technique is limited in some laboratories and totally absent in others. In addition, the expense and requirement for technical expertise have limited their use, particularly in high-prevalence regions, such as developing countries.
Enzyme immunoassay (EIA) for detecting antigen has been introduced successfully for Cryptosporidium screening in feces [5,6] Demonstration of cryptosporidial antigen in stool using ELISA is useful for screening large numbers of specimens. Several commercial immunochromatographic assays are currently available; however, previous comparisons have revealed limitations in their sensitivities according to Cryptosporidium species [7]. Potential advantages of commercially available ELISA kits for coproantigens are thus standardization of reagents, potential automatable process, and reproducibility in the context of administrative validation procedures. However discrepant data are available concerning this method [8].
The aim of the present study is to evaluate the performances of a commercial ELISA (CoproELISA Cryptosporidium kit, Savyon Diagnostics, Israel) for the detection of Cryptosporidium spp. in random clinical stool samples through a multicenter study.

Materials and Methods
The study was conducted in three medical parasitology laboratories from the University Hospitals of Rouen (Lab#1), Nantes (Lab#2), and Dijon (Lab#3) with recognized proficiency in the detection of Cryptosporidium spp.
Fifty stool samples containing Cryptosporidium oocysts were provided by the French Cryptosporidium National Network [9] and separately investigated by Lab#1 and Lab#2.
Cyclospora and Cystoisospora oocysts' strong autofluorescence properties render fluorescence microscopy useful for identification [11]. In epifluorescence microscopic examination, using a 330-380-nm ultraviolet filter, C. cayetanensis oocysts can be easily visible in clinical samples [12]. Oocysts of Cystoisospora can be differentiated by their roundish appearance, their thin, smooth wall and, after sporulation, by the number of sporocysts [13]. None of the samples contained other Coccidia, i.e., Cyclospora or Cystoisospora.
To assess human clinical stool preservation's effects on the CoproELISA performance, we used five fresh stool samples with C. parvum oocyst counts ranging from 4 to 76 per 50 microscopic fields (MF).
In each laboratory, ELISA were performed in triplicates according to the manufacturer's instructions (Savyon Diagnostics, Ashrod, Israel) by the same experienced staff member to minimize handling risks and reading errors. The Cryptosporidium antigen negative and positive internal control preparations were provided by the manufacturer. Results were expressed as absolute 450/605 nm 3,3 ,5,5 -Tetramethylbenzidine (TMB) product optical density (OD) values. Stool samples were considered ELISA-positive when the corresponding mean OD value was ≥ to the mean Cryptosporidium negative internal sample OD + 0.300 OD. Optical density of positive internal controls was checked for each series of experiments.
The sensitivities, specificities, positive predictive values and negative predictive values were calculated according to Loong [14].

Intra-Assay (Test-Retest) Assay
Intra-assay variability was investigated sequentially on the same day with one ELISAnegative and 2 ELISA-positive clinical stool samples diluted 1:10 in PBS (one sample containing C. parvum oocysts and 1 containing C. chipmunk genotype oocysts). For each sample, 6 independent ELISA were performed in triplicates (different plates, same reagents).

Inter-Assay ELISA Reproducibility
The replicate (inter-assay) ELISA reproducibility was assessed using one ELISAnegative sample and one ELISA-positive sample (containing C. hominis oocysts), tested independently in triplicates on 4 different days.

Statistical Evaluation of Results
Contingency analysis was performed using exact Fisher's test. Distribution comparisons between series of results were performed using t tests, thus assuming normal-like distribution of data. Correlation trends were estimated by calculating r correlation coefficient values.

Effects of Human Clinical Stool Preservation on Cryptosporidium spp. Antigen ELISA Detection
Most available clinical samples being preserved in K2CRO4 and/or frozen buffers, preliminary experiments were aimed at evaluating the effects of K2CRO4 treatment and freezing on ELISA antigen detection. As shown in Table 1, no difference in OD values was observed between undiluted stool, 1:10 stool dilution in PBS, 1:1 stool dilution in 2.5% K2CRO4 PBS, and undiluted stool kept frozen at (−80 • C) (p > 0.05). Specifically, all samples were found antigen positive. Results expressed as 405/605 nm mean OD values of triplicates. The coefficient of variation (CV), which is the measure of relative variability; the ratio of the standard deviation (SD) to the mean, accounted for less than 10%. #: sample number.

Influence of Cryptosporidium spp. oocyst Concentration in Clinical Stools on ELISA Antigen Detection
ELISA detection cut-off value for clinical samples was estimated in a pilot study according to C. parvum oocyst counts in stools. As shown in Table 2, OD exhibited moderate variation (from 3.690 to 4.407) from <1 oocyst to 76 oocysts/50 MF, which suggests that ELISA may detect antigens in clinical samples containing very low oocyst concentrations and considered microscopically negative.
As summarized in Table 4, ELISA detection of Cryptosporidium spp. antigens in 176 oocyst-negative clinical samples resulted in 166 true negative and 10 false-positive results Among the negative controls, 6 and 1 were found ELISA-positive (OD range: 0.671-3.549) by Lab#1 and Lab#2, respectively. In Lab#1, 4/6 samples were controlled negative by microscopic examination, but positive by PCR. The false-positive sample found by Lab#2 was positive for strongyloides stercoralis. At first, it was suspected cross-contamination between Cryptosporidium and nematode antigens, but this was ruled out as Strongyloides stercoralis ova found in few other samples were negative with the ELISA. Lab#3 found three false-positive samples that, after microscopy and PCR control, turn out to be negative to Cryptosporidium. No other intestinal parasites were found during this control.

Conclusions
In summary, ELISA detection of Cryptosporidium spp. coproantigens performed in 88 oocyst microscopy-positive clinical samples and 176 microscopy-negative samples resulted in 87 true positive, a false negative, 166 true negative, and 10 false negative results in at least one lab leading to sensitivity, specificity, positive and negative predictive value of 98.86%, 94.32%, 89.69%, and 99.40%, respectively. These data show that the method is efficient to identify Cryptosporidium oocysts in stool samples whatever the species with the exception of C. cuniculus, which seems to give lower OD values, but this result has to be confirmed as few samples were evaluated. Interestingly, the quantity of oocysts did not interfere with the result as sample with few or many oocysts were detected with similar OD values as previously described [8].
Coproantigen detection using ELISA method requires minimal training thus appeared to be easy to perform, as well as accurate for epidemiological studies and diagnostic purposes of Cryptosporidium infection, compared with conventional microscopic methods.