**1. Introduction**

Human cases of cryptosporidiosis were first reported in the 1970s in children and immunosuppressed adults [1]. In 2015, the Global Enteric Multicenter Study (GEMS) described *Cryptosporidium* spp. as the second leading cause (5–15%) of moderate to severe diarrhea among infants in countries of sub-Saharan Africa and South Asia, after rotavirus [2]. At the same time, *Cryptosporidium* spp. were found to be responsible for more than 8 million cases of foodborne illnesses in 2010, and they were ranked fifth out of 24 potentially foodborne parasites in terms of importance [3,4]. In 2017 in France, the National Reference Center-Expert Laboratory (CNR-LE) for cryptosporidiosis was set up, allowing the collection and interpretation of epidemiological data thanks to the participation of members of the network. Published data from the French CNR-LE for cryptosporidiosis show that: i) even with around 250 notified cases each year, cryptosporidiosis is still largely underestimated in France, ii) cryptosporidiosis is predominant in immunocompetent individuals and especially in young children and young adults, and iii) cryptosporidiosis is over-represented in the summer [5,6]. The routine diagnosis of cryptosporidiosis still relies on light microscopy examination for many laboratories [7–10]. However, light microscopy

**Citation:** Costa, D.; Soulieux, L.; Razakandrainibe, R.; Basmaciyan, L.; Gargala, G.; Valot, S.; Dalle, F.; Favennec, L. Comparative Performance of Eight PCR Methods to Detect *Cryptosporidium* Species. *Pathogens* **2021**, *10*, 647. https:// doi.org/10.3390/pathogens10060647

Academic Editor: Hans-Peter Fuehrer

Received: 29 April 2021 Accepted: 21 May 2021 Published: 23 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

examination lacks sensitivity, is time-consuming, and requires skilled technicians, making it an inefficient method for laboratories which are able to switch to PCR analysis [8,10–12]. Currently, several PCR methods are available to screen *Cryptosporidium* spp. DNA, both "inhouse" and commercial methods, sometimes incorporated into multiplex panels [13–17]. Consequently, more and more laboratories are opting for such methods based on the practicalities of economic management. However, disparities exist in the performance of these methods. DNA extraction is essential to obtain good performance in PCR analysis and especially regarding parasitological investigations on stool samples. Some studies reported different performances of DNA extraction methods, and regarding the extraction of *Cryptosporidium* oocysts, a mechanical treatment of stool samples seems essential [18–23]. In addition to the extraction method, the removal of inhibitory substances and the gene locus targeted by related primers plays a major role in the performance of the method. One of the tasks of the CNR-LE for cryptosporidiosis is to assess the performance of available diagnostic tools. A previous work already compared performances of various extraction methods on *C. parvum* oocysts from stool samples [22]. In continuity of this work and based on the most effective extraction method, we propose a comparison of the limit of detection of eight real-time PCR methods (commercial or not) on the DNA of *Cryptosporidium* species. The main aim was to provide data to select the best methods for DNA amplification in terms of sensitivity and ability to detect human pathogenic *Cryptosporidium* species (even rare ones) in routine diagnosis.

#### **2. Results**

The results obtained from the four "in-house" PCR methods are summarized in Table 1. Except for the most concentrated *C. parvum* extract (105 oocysts/gram), significant differences in threshold cycle (Cq) values were observed when applicable (on ANOVA test). All four "in-house" methods detected *C. parvum* DNA and *C. hominis* DNA with a limit of detection of 10<sup>3</sup> oocysts/gram and 10<sup>4</sup> oocysts/gram, respectively. The most sensitive "in-house" PCR method for both *C. parvum* and *C. hominis* was the method developed by the CNR-LE Cryptosporidiosis Collaborating Laboratory (University Hospital of Dijon) and described by Valeix et al. 2020 [22]. Cq values obtained with the method described by Mary et al. 2013 [15] were lower than other tested methods but analysis performed in triplicate was insufficient to detect 10 oocysts/gram for *C. parvum*, contrary to the method described by Valeix et al. PCR efficiencies were satisfactory only on *C. parvum* DNA amplification for the methods described by Fontaine et al. 2002 and Valeix et al. 2020 [13,22]. R<sup>2</sup> values were satisfactory (>0.99) only to detect *C. parvum* and for the methods described by Fontaine et al. 2002, Hadfield et al. 2011 and Mary et al. 2013 [13–15].

The results obtained from the four multiplex commercial methods are presented in Table 2. All four commercial methods detected *C. hominis* DNA and *C. parvum* DNA with a limit of detection of 103 oocysts/gram. The best performance was obtained with the FTD Stool parasites method, with a detection of DNA corresponding to 1 oocyst/gram for *C. parvum* and 10 oocysts/gram for *C. hominis*. The Allplex GI Parasite Assay kit was the second-best method, with a detection of DNA corresponding to 100 oocysts/gram for *C. hominis* and 10 oocysts/gram for *C. parvum* but requiring a triplicate to reach the limit of detection (only 1/3 triplicates was positive to detect *C. parvum* at 10 oocysts/gram). PCR efficiencies and R2 values varied greatly depending on the studied method but overall were unsatisfactory (PCR efficiency < 90% or > 110% and R2 value < 0.99).


**Table 1.** Limit of detection of tested "in-house" methods on *C. parvum* and *C. hominis*. PCR efficiencies were calculated based on obtained results from corresponding ranges of dilutions.

**Table 2.** Limit of detection of tested commercial methods on *C. parvum* and *C. hominis.* PCR efficiencies were calculated based on obtained results from corresponding ranges of dilutions.



**Table 2.** *Cont.*

The ability to detect rare species implicated in human pathologies for each tested method is summarized in Table 3. All tested methods were able to detect the species *C. cuniculus*, *C. meleagridis*, *C. felis*, *C. chipmunk*, and *C. ubiquitum*, except the methods described by Mary et al. 2013 and Fontaine et al. 2002 [13,15]. Specificity tests performed in triplicate per condition, as described in the Methods section, revealed cross-reactivity only for the method described by Hadfield et al. 2011 [14] with *Encephalitozoon intestinalis* DNA.

**Table 3.** Detection of rare species of *Cryptosporidium* implicated in human cases by tested methods.


#### **3. Discussion**

This study was designed to address questions regularly raised within the framework of scientific exchanges of the CNR-LE for cryptosporidiosis. It compared the performance of eight PCR methods to detect *Cryptosporidium* species (even rare) implicated in human infection, and their limit of detection. At first, the subject appeared to be well-investigated within the scientific community. However, in most cases, PCR performances to detect *Cryptosporidium* DNA were evaluated in cohorts from microscopically positive stool samples (probably relatively highly concentrated in oocysts), or not specifically through multiplex panels and from various extraction methods, or sometimes from DNA extracts stored for a long time [24–33]. In this study, thanks to a standardized extraction procedure (selected among the best methods regarding specific *Cryptosporidium* DNA extraction from stool samples [22]), observed PCR performances were exclusively due to the DNA amplification step. The limit of each studied PCR method was determined by assessing titrations of *Cryptosporidium* oocysts in stool samples as well as testing rare species implicated in human infection. The main interest of the study was to provide data on efficient methods for

the routine diagnosis of cryptosporidiosis as a complement to extraction methods already assessed [22,23].

The results obtained generally showed similar performances between commercial and "in-house" methods in terms of limit of detection, with variations between each tested kit. Regarding *C. parvum* and *C. hominis* respectively, limits of detection generally reached at least 100 and 1000 oocysts/gram regardless of the method. Nevertheless, the limit of detection appeared optimal with the FTD® method considering both *C. parvum* and *C. hominis*. Variations in limits of detection may first be explained by the genes targeted by PCR methods. Three of the four "in-house" methods target the 18S rRNA gene whose expression is estimated at 5 copies/genome (20 copies/oocyst) [15]. The "in-house" method described by Fontaine et al. 2002 targets a gene whose expression is estimated at 1 copy/genome, and indeed, its observed performance in terms of limit of detection was generally poorer than that of the three methods targeting the 18S rRNA gene. Regarding commercial methods, targeted genes were only available for FTD® (DNA J-like protein, number of copies per genome not known) and Amplidiag® methods (COWP gene; 1 copy/genome). Of note, the observed performance of the Amplidiag® method was close to that of Fontaine et al.'s "in-house" method targeting a gene also expressed in 1 copy/genome. Comparing Tables 1 and 2, the limit of detection of the Amplidiag® method appeared slightly better than that of Fontaine et al.'s "in-house" method (for both *C. parvum* and *C. hominis*) but this was only due to DNA detection in one replicate at the very end of the PCR program. It could be explained by the heterogeneous distribution of DNA in elution volume when parasite concentrations are low. To limit this bias, and to obtain optimized performance, we recommend running each DNA extract in several replicates (at least in triplicate) or until exhaustion if possible.

Regarding the results obtained to detect rare species of *Cryptosporidium* implicated in human infections, most tested methods were able to detect rare species except the "in-house" methods of Fontaine et al. 2002 and Mary et al. 2013 [13,15]. However, a limitation of this study was the use of only triplicate of each tested *Cryptosporidium* subtype due to the amount of available positive stools. The use of more numerous rare strains could potentially improve the observed results. For the method described by Fontaine et al. 2002, they highlighted the use of a specific primer-probe set supposed to be specific for a *C. parvum* genomic DNA sequence. No cross-reactivity with other *Cryptosporidium* species was expected; however, they initially reported cross-reaction with the *C. meleagridis* genotype, which was confirmed in our study [13]. For the method of Mary et al. 2013, no rare species was detected in this study. In the original article, tests on *C. felis*, *C. bovis*, *C. cuniculus*, *C. canis*, and *C. chipmunk* were evoked in the discussion. However, in reality, corresponding results were not shown [15]. Consequently, primers and probes described in the article of Mary et al. 2013 are probably very specific to *C. parvum* and *C. hominis*. Regarding specificity in this study, performances obtained were highly satisfactory for each tested condition in concordance with the literature [13–15,27,34]. Cross-reactivity with *Candida albicans* DNA was tested since Mary et al. 2013 reported potential cross-reactivity with the *C. albicans* 18S rRNA gene (based on an in silico approach) and primers and probes of the PCR method described by Hadfield et al. 2011 [14,15]. For the method described by Hadfield et al. 2011, no cross-reactivity was observed with *C. albicans* but cross-reactivity was observed with *E. intestinalis*.

Finally, out of a total of 784 PCRs performed, varying results were obtained from the same DNA samples. Commercial methods (especially FTD® and Allplex®) appeared to be valuable options for large screening to detect *Cryptosporidium* species. We recommend testing each DNA extract at least in triplicate to optimize the detection of small amounts of DNA. However, if commercial methods are able to detect rare species, results are expressed exclusively as positive or negative for *Cryptosporidium* spp. DNA detection. Consequently, to discriminate species, we recommend the use of "in-house" methods, and especially the method described by Valeix et al. 2020 [22], due to the results obtained in terms of limit of detection and the ability to detect rare species. In addition, the method described by

Valeix et al. 2020 appeared to be strongly replicable, since performances in terms of limit of detection were similar to those described here, even using different stool samples [22].

#### **4. Materials and Methods**

#### *4.1. Strains*

*Cryptosporidium* spp. tested strains were obtained from the French cryptosporidiosis CNR-LE stools collection. *C. parvum* IIaA15G2R1 (n = 3), *C. hominis* IbA10G2 (n = 3) gp60 subtypes, *C. cuniculus* (n = 3), *C. meleagridis* (n = 3), *C. felis* (n = 3), *C. chipmunk* (n = 3), and *C. ubiquitum* (n = 3) were tested, all previously isolated from human clinical samples. In total, per studied method, six ten-fold range points (105-1 oocysts/gram) were studied for both *C. parvum* and *C. hominis*. Ranges of dilutions were done from highly concentrated natural stool samples that we diluted subsequently. Ranges of dilutions were performed in liquid stool matrix exempt of *Cryptosporidium* species. Each sample was vortexed for 20 s before performing dilutions. Oocyst numeration was done microscopically using Kova cells and confirmed by immunofluorescence as described in Section 4.2.

Regarding the studied *Cryptosporidium* rare species, stools were selected from the CNR collection with oocyst concentrations that varied between 103 and 10<sup>4</sup> oocysts/gram to be easily detectable.

Other positive stool specimens were obtained from the CNR-LE collection to evaluate specificity: *Giardia intestinalis* (n = 3), *Blastocystis hominis* (n = 3), *Enterobius vermicularis* (n = 3), *Chilomastix mesnilii* (n = 3), *Entamoeba histolytica* (n = 3), *Entamoeba dispar* (n = 3), *Encephalitozoon intestinalis* (n = 3), and *Enterocytozoon bieneusi* (n = 3) positive stool samples were tested. Exact stool concentrations of these other pathogens were not calculated. Positivity was objectified by microscopy exclusively assuming relatively high concentrations. Additional tests were performed on *Candida albicans* (n = 3) and on negative stool samples (n = 20). A total of 784 PCRs were performed (98 per tested method).
