Development, Optimisation and Validation of a Novel Multiplex Real-Time PCR Method for the Simultaneous Detection of Cryptosporidium spp., Giardia duodenalis and Dientamoeba fragilis

The enteric protozoan parasites Cryptosporidium spp., Giardia duodenalis and Dientamoeba fragilis are—to various extents—contributors to the burden of gastrointestinal illness in high-income countries. Detection of these pathogens by microscopy examination is challenging because of the limited sensitivity and need for specific staining procedures. We developed and optimised a new multiplex real-time PCR assay for the simultaneous detection of Cryptosporidium spp., G. duodenalis and D. fragilis in clinical (stool) samples. The diagnostic performance of the assay was evaluated against a large panel of well-characterised DNA samples positive for Cryptosporidium spp. (n = 126), G. duodenalis (n = 132) and D. fragilis (n = 49). The specificity of the test was assessed against a DNA panel from other intestinal or phylogenetically related parasites (n = 105) and faecal DNA from individuals without clinical manifestations (n = 12). The assay exhibited a diagnostic sensitivity of 0.90–0.97 and a diagnostic specificity of 1. The limit of detection was estimated for Cryptosporidium (1 oocyst) and G. duodenalis (5 × 10−4 cysts). The method allowed the detection of four Cryptosporidium species (C. hominis, C. parvum, C. meleagridis and C. cuniculus) and five G. duodenalis assemblages (A–E) without cross-reacting with other parasites belonging to the phyla Amoebozoa, Apicomplexa, Euglenozoa, Microsporidia, Nematoda and Platyhelminthes. This newly developed multiplex real-time PCR assay represents a novel alternative for the rapid and accurate detection of Cryptosporidium, G. duodenalis and D. fragilis in clinical settings.


Introduction
Diarrhoea is a major health problem worldwide. Only in 2017, almost 1.6 million people died from diarrheal diseases globally (https://ourworldindata.org/diarrheal-diseases#, accessed on 7 October 2022). Those most affected by diarrhoea are children and immunocompromised individuals living in developing countries with limited or no access to safe water and adequate sanitation. Even though diarrhoea-related mortality rates in resourcepoor settings have decreased considerably, morbidity remains high [1]. In most instances, the aetiologies of diarrhoea are related to viral, bacterial and parasitic infections, differing in severity and public health impact according to endemic level and geographical area.
Among parasitic pathogens, the protozoa Cryptosporidium spp. and Giardia duodenalis are major contributors to the burden of diarrhoeal disease globally [2]. Cryptosporidium spp., with C. parvum and C. hominis accounting for most documented cases of human cryptosporidiosis, have been recognised as the cause of large water-and food-borne outbreaks of gastroenteritis [3][4][5]. Giardia duodenalis infections are very common throughout the world and are regarded as a major cause of gastrointestinal complaints in western countries [6]. Another intestinal protozoan with a cosmopolitan distribution, although with highly variable colonisation/infection rates, is Dientamoeba fragilis. Although its pathogenicity and clinical significance are still controversial, most clinicians agree that D. fragilis can cause abdominal pain and diarrhoea as it is frequently found in patients suffering from these disorders [7]. Remarkably, the perception of the clinical significance of D. fragilis differs greatly among geographical areas. This protozoon is a common finding in Danish children attending day-care centres and, therefore, is regarded as a commensal organism [8]. In contrast, D. fragilis is less prevalent, but more frequently associated with gastrointestinal manifestations, in other high-income countries such as Australia [9]. These data suggest that implementing routine (molecular) screening of D. fragilis might be useful in those parts of the world where the parasite has received little attention to gain information on prevalence across various age groups and phenotypes (symptomatic vs. asymptomatic). This information would be extremely useful to help in ascertaining the true clinical significance of D. fragilis.
Classically, diagnosis of protozoan enteroparasite infections has been achieved by microscopical examination of faecal samples. Light microscopy of faecal concentrates stands as the preferred routine diagnostic method in many clinical settings. Although this technique is labour-intensive, lacks sensitivity and requires skilled technicians, its simplicity and low cost outweighs the above-mentioned limitations and makes microscopy suited for resource-limited laboratories, particularly those in endemic and high-prevalence areas [10]. In contrast, typical epidemiological scenarios in high-income countries involve low parasite prevalence rates and burdens. In this context, highly sensitive and specific diagnostic methods are required [11,12]. Accurate and fast diagnosis of intestinal protozoa is also important in treatment decision-making schemes, as well as for early detection in outbreak investigations. Not surprisingly, molecular techniques are progressively replacing microscopy as the first-line diagnostic method for diagnosis of intestinal parasites in industrialised countries [12].
In recent years, a wide diversity of in-house and commercial real-time PCR (qPCR) assays (see Table S1) have been developed for detecting enteric parasitic pathogens, with the trend moving quickly from single to multiple pathogen detection [6,13]. Validation and standardisation of such novel assays are important tasks that must be carried out to accurately assess their diagnostic performance and practical suitability for routine use in clinical settings. Here, we describe the development, optimisation and validation of a fast and practical in-house multiplex qPCR for the simultaneous detection of Cryptosporidium spp., Giardia duodenalis and Dientamoeba fragilis in human stool samples.

Study Design
This was a retrospective observational study specifically conducted to develop, validate and standardise a novel multiplex qPCR method for the simultaneous detection of the diarrheagenic protozoan enteroparasites Cryptosporidium spp., G. duodenalis and D. fragilis.

DNA Reference Panel
A total of 424 well-characterised genomic DNA samples extracted and purified from biological (mostly stool) specimens of clinical patients were used (Table 1). These samples were initially diagnosed by in-house singleplex PCR protocols and, when possible, Sanger sequencing at reception at the Spanish National Centre for Microbiology (SNCM). DNA samples positive for Cryptosporidium spp. (n = 126; median cycle threshold (C T ) value, 32.6; range, 25.5-40.0) and Giardia duodenalis (n = 132; median C T value, 31.0; range, 24.1-40.0) were initially detected by a commercial multiplex qPCR platform (Allplex™ Gastrointestinal-Parasite Assay, Seegene Inc., Seoul, South Korea). DNA samples positive for D. fragilis (n = 49) had initially been identified by an in-house singleplex qPCR (median C T value, 32.2; range, 21.4-41.3) [14]. DNA samples (n = 105) positive for enteric or phylogenetically related (e.g., Apicomplexan) parasitic species other than G. duodenalis and D. fragilis and those belonging to Cryptosporidium (and used in cross-reaction evaluation in the present study) were initially diagnosed by a variety of in-house single-round, semi-nested and nested PCR assays in place at the SNCM. DNA (n = 12) from stool samples of apparently healthy individuals was included for the same purpose. Whenever possible, species and genotypes/sub-genotypes were determined by PCR and Sanger sequencing methods at the time of initial diagnosis. The full dataset including all relevant information of the original DNA samples used in this study can be found in Table S2. The DNA reference panel included DNA samples of the four Cryptosporidium species most frequently found causing human cryptosporidiosis (C. hominis, C. parvum, C. meleagridis and C. cuniculus) [15,16] and of five different G. duodenalis genetic variants (assemblages A-E). For cross-reactivity evaluation, we purposely selected representative DNA samples of various taxonomic groups of pathogens, including amoebozoan (n = 38), apicomplexan (n = 41), euglenozoan (n = 13), microsporidian (n = 3), nematode (n = 8) and platyhelminthic (n = 2) parasites ( Table 1).
All DNA samples (except those positive for D. fragilis) were part of the DNA collection of the Parasitology Reference and Research Laboratory of the SNCM. Genomic DNA was extracted from 200 mg of faecal material using the QIAamp DNA Stool Mini Kit (QIAGEN, Hilden, Germany) according to the manufacturer's instructions. Some DNA samples were from in vitro cultures or of animal origin, particularly those belonging to animal-adapted species/genotypes or rarely found circulating in humans (Table S2). DNA samples from D. fragilis-positive patients were provided by the Department of Bacteria, Parasites and Fungi, Infectious Disease Preparedness, Statens Serum Institute (Copenhagen, Denmark). All DNA samples used in this study had been extracted during the period 2014-2019 and stored at -20 • C until testing.

Primers and Probes
PCR primers and detection probes were chosen using Primer Express software (Applied Biosystems, CA, USA) based on the known oocyst wall protein 1 (cowp1) gene sequence for C. hominis (GenBank accession No. DQ388389), the small subunit ribosomal RNA (ssu rRNA) gene sequence for G. duodenalis (GenBank accession No. M54878) and the internal transcriber spacer (ITS) gene sequence for D. fragilis (KY554844). Selected primer and probe sequences are detailed in Table 2 Table 2. Primers and probes used in the multiplex real-time PCR assay for the simultaneous detection of Cryptosporidium spp., Giardia duodenalis and Dientamoeba fragilis in the present study.

Optimisation
Singleplex or multiplex qPCR reactions were carried out in volumes of 25 µL with PCR buffer (2× Brilliant III Ultra-Fast QPCR master mix, Agilent Technologies, Santa Clara, CA, USA) containing mutant Taq DNA polymerase, dNTPs and 5.5 mM of MgCl 2 . Mixes also included 50 pmol of each specific primer, 25 pmol of double-labelled probes and 5 µL of template DNA. Cycling conditions consisted of 10 min incubation at 95 • C, followed by 45 cycles of 15 s at 95 • C and 1 min at 60 • C, with fluorescence being measured after each cycle. Amplification was performed with the BioRad CFX96 Dx System (BioRad, Hercules, CA, USA).

Performance Assessment
All samples forming part of the DNA reference panel were blindly analysed in duplicates to avoid bias. Samples positive for Cryptosporidium spp. (ROX fluorophore), G. duodenalis (FAM fluorophore) and D. fragilis (HEX fluorophore) typically generated fluorescence curves with qPCR C T values < 45.

Limit of Detection and PCR Efficiency
Ten-fold serial dilutions (down to 10 −7 ) of individual DNA samples positive for C. hominis, G. duodenalis and D. fragilis were used to estimate the relative limit of detection of the qPCR assay in both singleplex and multiplex formats. DNA isolated from 10,000 C. hominis oocysts and 500 G. duodenalis cysts (enumerated from human faecal samples using direct fluorescence microscopy) were used as starting concentrations for this purpose. No quantified starting material was available for D. fragilis, as cyst forms from this protozoon have not been unambiguously described in human clinical specimens [17]. Additionally, the efficiency (E) of the multiplex qPCR assay was calculated from the slope(s) of the standard curve according to the following formulas: Slope= −1/log2 (after multiplying both sides by (slope/log2) (4) Slope = −3.32 (5) Therefore, for a 10-fold dilution series, the slope is −3.33 when E = 100%. A good reaction should have an efficiency between 90% and 110%, which corresponds to a slope between −3.58 and −3.10 [18].

Statistical Analyses
The Cohen's Kappa test was estimated to assess the agreement of the diagnostic results obtained with the multiplex qPCR detection assay and the reference singleplex qPCR methods used during routine analyses at initial diagnosis (see Table S2). Cohen's Kappa ranges between 0 (no agreement between the two raters) and 1 (perfect agreement between the two raters). A Cohen's Kappa value between 0.81 and 0.99 was considered as "near-perfect agreement". Clinical diagnostic sensitivity and specificity, and negative and positive predicted values (with 95% confidence intervals), were calculated using Meta-DiSc 1.4. software [19] based on the following formulas: Positive predictive value (PPV) = [a/(a + b)] × 100 samples with a negative result in the novel multiplex assay but a positive result in the subsequent confirmatory singleplex qPCR were considered true false negatives.

Limit of Detection and Efficiency
The analytical sensitivity and efficiency of the qPCR assay for the detection and differentiation of Cryptosporidium spp., G. duodenalis and D. fragilis (both in singleplex and multiplex formats) are summarised in Table 3. The assay was able to detect the equivalent of a single Cryptosporidium oocyst in both the singleplex and multiplex formats. However, the latter performed slightly poorer than the former in terms of efficiency (110% vs. 100.3%, respectively). These results are consistent with the facts that: (i) a single Cryptosporidium oocyst contains four sporozoites, each with a haploid genome, resulting in an oocyst ploidy of 4N, and its cowp1 gene exists in a single copy, (ii) a fully differentiated Giardia cyst contains four nuclei, each with a ploidy of 4N, resulting in a cyst ploidy of 16N, and its ssu rRNA gene exists in multiple copies, and (iii) Dientamoeba is unknown with respect to ploidy and genome size, but the targeted gene (ITS) should be multi-copy, one for each ribosomal set of genes.

Sensitivity and Specificity
The diagnostic performance of the multiplex qPCR assay is summarised in Table 4. Out of the 126 Cryptosporidium-positive samples included in the DNA reference panel, the multiplex assay was able to detect 119 (diagnostic sensitivity, 94.4%). Samples belonged to C. hominis (gp60 genotype families Ib, Id, Ie and If), C. parvum (gp60 genotype families IIa and IId), C. meleagridis (gp60 genotype family IIIg) and C. ubiquitum (unknown gp60 genotype family) (see Table S1). The seven Cryptosporidium-positive samples that failed to be amplified in the multiplex qPCR assay included one C. hominis, one C. parvum, three C. meleagridis and two Cryptosporidium spp. isolates. All seven Cryptosporidium-positive samples yielded C T values > 35 when retested in singleplex qPCR and were, therefore, regarded as true false-negative results in multiplex qPCR. Of note, 37 Cryptosporidiumpositive samples were known to be concomitantly infected by D. fragilis (n = 26), G. duodenalis (n = 10) and D. fragilis + G. duodenalis (n = 1), as previously determined during routine initial diagnosis. All these samples were also scored positive for G. duodenalis and D. fragilis by the multiplex qPCR assay and, therefore, not considered as false-positive results (Table S1).
Out of the 132 G. duodenalis-positive samples included in the DNA reference panel, the multiplex assay was able to detect 129 (diagnostic sensitivity, 97.7%). Samples belonged to G. duodenalis assemblages A, B, C, D and E (see Table S1). The three Giardia-positive samples that failed to be amplified in the multiplex assay included two assemblage B isolates, whereas the assemblage type of the third isolate was unknown. The multiplex assay also detected co-infections by D. fragilis (n = 50) and D. fragilis + Cryptosporidium spp. (n = 1), all of them previously identified during routine initial diagnosis (Table S1).
Out of the 49 D. fragilis-positive samples included in the DNA reference panel, the assay was able to detect 44 (diagnostic sensitivity: 89.8%). No attempts were made to determine the genotype of the D. fragilis isolates because of the limited intra-genetic diversity previously observed in this protozoon [20,21]. A total of 18 D. fragilis-positive samples were co-infected either with G. duodenalis (n = 17) or Cryptosporidium (n = 1). All these concomitant infections were identified at initial diagnosis and therefore were not considered as false-positive results (Table S1). To discard potential cross-reactivity with phylogenetically closely related members of the genus Trichomonas, we aligned our D. fragilis sequence primers (both forward and reverse) with known partial ssu rRNA sequences from T. vaginalis (GenBank accession number JX943584), T. gallinae (GenBank accession number EU215373) and T. tenax (GenBank accession number D49495). In all cases, mismatches were found in 36-41% of the positions, making the amplification of Trichomonas spp. with our D. fragilis primer set highly unlikely.
None of the 105 DNA samples positive for parasites other than Cryptosporidium spp., G. duodenalis and D. fragilis yielded positive results when tested in the multiplex assay, nor did the 12 DNA samples from stool samples of individuals without clinical manifestations produce any signal (Table 4).
Finally, a very good agreement (Kappa test values ranging from 0.925 to 0.976) was observed between the results obtained by multiplex qPCR assay and those previously generated by the reference qPCR methods at initial diagnosis (Table 5).

Discussion
Here, we developed, optimised and evaluated a new multiplex qPCR assay for the simultaneous detection of the three most common and clinically relevant protozoan enteroparasites (Cryptosporidium spp., G duodenalis and D. fragilis) in western countries. To do so, a large, carefully selected panel of well-characterised DNA samples was used. This is the strategy adopted in our laboratory for evaluating and validating PCR-based diagnostic tests [22,23]. This approach has two additional benefits, as it allows (i) the testing of infrequent species/genotypes and animal-adapted genetic variants with zoonotic potential, and (ii) the testing of species/genetic variants from two European countries (Denmark and Spain), reflecting different geographic areas and epidemiological scenarios. In contrast, most previous studies preferred using prospectively collected stool samples submitted to clinical laboratories for parasite investigation, for which the above-mentioned molecular information is often unavailable [24][25][26][27][28][29][30][31][32].
Our multiplex qPCR assay achieved a diagnostic sensitivity of 0.94 for the detection of Cryptosporidium spp. This figure is slightly lower than those (ca. 1) reported using commercially available multiplex qPCR assays, including the Allplex GI-Parasite (Seegene, Soul, South Korea) [33,34], the EasyScreen Enteric Parasite (Genetic Signatures, Sydney, Australia) [27], the G-DiaParaTrio (Diagenode Diagnostics, Liège, Belgium) [34] and the NanoCHIP (Savyon Diagnostics, Ashdod, Israel) [30] assays. The slightly lower sensitivity of our multiplex qPCR assay is likely due to the single-copy nature of the targeted gene (cowp1) compared to the multiple-copy nature of ssu rRNA and ITS. In contrast, cowp1 has increased specificity over ssu rRNA, reducing the rate of false-positive amplifications. Still, lower values of 0.86-0.92 have been reported with the RIDAGENE Parasitic Stool Panel II (R-Biopharm AG, Darmstadt, Germany) [33,34] and the Tib MolBiol (Roche Diagnostics, Basilea, Switzerland) [33] assays. Remarkably, our multiplex qPCR assay was able to detect the four Cryptosporidium species (namely C. hominis, C. parvum, C. meleagridis and C. ubiquitum) causing the bulk of human cryptosporidiosis cases globally.
In many western countries, the trichomonadida parasite D. fragilis is a very common finding in human stool samples and is second only to Blastocystis sp. in terms of positive rates [36]. Since D. fragilis is particularly difficult to identity using conventional microscopy examination [37], qPCR has become the gold standard for the diagnosis of this (and other) gastrointestinal protozoan parasites. Indeed, several qPCR methodologies for the simultaneous detection of diarrhoea-causing protozoan pathogens, including D. fragilis, have been developed and commercialised in the last decade (see Table S1). A recent study documented that discrepant results in the detection of D. fragilis were common when the diagnostic performances of laboratory-developed and commercial qPCR assays were compared [38]. The multiplex qPCR assay described here achieved a diagnostic sensitivity of 0.90 for the detection of D. fragilis, well in the range of those reported in the literature using commercial methods (see Table S1). Higher sensitivity values (0.95-1) have been reported using the Allplex (Seegene) [33,34], EasyScreen (Genetic Signatures) [27], FTD (Fast Track Diagnostics) [22,25,33,35], NanoCHIP (Savyon Diagnostics) [26,30] and the Tib MolBiol (Roche Diagnostics) methods [33]. Lower (0.25-0.79) diagnostic sensitivities have been reported using RIDAGENE (R-Biopharm) [33,34] Regardless of the pathogen considered, our multiplex qPCR assay produced diagnostic specificities of 1. This was somehow expected as the panel used for reference purposes included well-characterised DNA samples that were previously diagnosed for the potential presence of concomitant infections, minimizing the risk of obtaining false-positive results.
Despite this effort, we are aware that some relevant pathogenic and commensal protozoan species were missing in our panel. For instance, DNA samples from host-adapted Cryptosporidium species rarely seen infecting humans (e.g., C. canis, C. felis, C. muris) were not available for testing. Other potentially cross-reacting species that were not part of the reference panel included Cyclospora cayetanensis, Entamoeba coli, Endolimax nana and Encephalitozoon intestinalis, among others. Similarly, larger numbers of G. duodenalis assemblages C, D and E should be tested to confirm the diagnostic sensitivity achieved for isolates belonging to assemblages A and B. These limitations remain to be addressed in future works.

Conclusions
The novel multiplex qPCR test described here exhibited excellent linearity and high analytical sensitivity and specificity values, providing a suitable alternative diagnostic tool for the molecular detection and differentiation of Cryptosporidium spp., G. duodenalis and D. fragilis, the most clinically relevant diarrhoea-causing protozoan parasites in highincome countries. The test can also be used either for screening or confirmatory purposes in endemic areas of resource-poor settings. Of note, detection of all three pathogens (but especially D. fragilis) by conventional microscopy examination is challenging because of the intrinsic difficulty of correctly identifying some of their morphological features or the need for specific staining methods (e.g., Ziehl-Neelsen staining for Cryptosporidium spp.) [39,40].
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/pathogens11111277/s1. Table S1: Diagnostic performance of commercially available multiplex real-time PCR assays for the simultaneous detection of Cryptosporidium spp., Giardia duodenalis and D. fragilis. Table S2 Informed Consent Statement: Patient consent was waived because the stool samples used were exclusively intended for routine clinical diagnostic procedures. All samples were anonymised using a unique laboratory identifier code to guarantee the anonymity and confidentiality of the patients. Data Availability Statement: Data are available within the article or its supplementary materials.