Introduction

Cryptosporidiosis is an emerging disease caused by apicomplexan parasites belonging to the genus Cryptosporidium. Twenty-seven recognized species and more than 40 genotypes have been described in the literature [1–3], but C. parvum and C. hominis represent the most prevalent agents infecting humans [4, 5]. Some other species are considered to be pathogens in both immunocompetent and immunocompromised individuals, but their prevalence remains much lower. Infection is most commonly spread by the fecal–oral route, through oocyst-contaminated water or food or by direct contact with infected humans or animals as well as contaminated non-living objects (fomites). Transmission by aerosol inhalation of oocysts has also been reported, but it remains very uncommon [6]. The severity of illness depends on both host and parasite factors, which include immune and nutritional status, age, and species/genotype involved. In healthy subjects, cryptosporidiosis is a transient self-limiting gastroenteritis-like syndrome that usually lasts 1–2 weeks [4]. But immunocompromised patients (especially with HIV/AIDS) can suffer from profuse, watery diarrhea, resulting in malabsorption, wasting, and reduced life expectancy [6, 7]. Extra-intestinal manifestations, such as biliary involvement, pancreatitis, and respiratory tract disease, may also occur in deeply immunosuppressed individuals with severe cryptosporidiosis [8]. Nevertheless, the treatment options are limited, and no antiparasitic drug has been proven to possess clear activity against Cryptosporidium [9].

The lessons learnt from the numerous waterborne and foodborne outbreaks that have occurred worldwide during the 2 past decades [10, 11] have elevated Cryptosporidium spp. to the rank of parasites of major public health importance. In 2012, a joint Food and Agriculture Organization (FAO) and WHO expert committee ranked Cryptosporidium in fifth place among the 24 most important food-borne parasites [2]. In addition, the parasite has been recently identified as one of the main causes of mild-to-severe diarrhea in children <5 years old in developing countries [12]. Information regarding the overall incidence of cryptosporidiosis is limited. In France, cryptosporidiosis cases have been notified to the ANOFEL Cryptosporidium National Network (ACNN) since 2006, and ≈ 100 cases per year (≈0.15/100,000 population/year) have been reported [13]. However, cryptosporidiosis probably continues to be underestimated, primarily because sampling from patients with gastrointestinal impairments and specific requests for diagnostic tests is insufficient.

Biological diagnosis relies on the detection of Cryptosporidium oocysts in stool samples by microscopic examination of smears prepared from formalin–ether concentrates and stained with the historical modified Ziehl–Neelsen (ZN) method [14], but sensitivity is low (~75 %) [15, 16]. Increased sensitivities of 90–100 % can be obtained with fluorescent auramine O and immunofluorescent assays using monoclonal antibodies [16–18]. But the sensitivity of these microscopic methods remains dependent on the clinical status of the patient, yielding false-negative results among poorly symptomatic or asymptomatic subjects with low parasitic loads and light infections. Immunochromatographic assays (ICA) have also been developed to detect Cryptosporidium antigens. However, a recent French blind multicenter study demonstrated limited interest, since the sensitivity of the four tested ICAs ranged from 50.1 % to 86.7 % for C. parvum and C. hominis, and was <35 % for the other species [19].

More recently, molecular methods, which are required for species identification and furthermore allow DNA quantification when real-time assays are used, have been developed. The higher sensitivity reported for Cryptosporidium PCR in several studies suggests that molecular tools could improve the diagnosis of cryptosporidiosis [20, 21], and they are increasingly used as first-line in clinical diagnostics [22]. 18S rDNA is the most frequently used target [23]. PCRs targeting the Cryptosporidium outer wall protein (COWP) [24], 60-kDa glycoprotein (gp60) [25], actin [26], LAXER [27], beta-tubulin [28] or HSP-70 [29] have also been developed. However, the small amount of target DNA combined with the low efficiency of total DNA extraction due to oocyst cell wall sturdiness, and the presence of PCR inhibitors in the samples to be amplified, may result in detection of false negatives. Moreover, only a few publications have focused on the optimization of Cryptosporidium DNA extraction in stools [30–34]. In this context, we aimed to (i) develop an optimal procedure for Cryptosporidium DNA isolation, detection and quantification, by comparing the performance of two commercial DNA extraction kits and the accuracy of two real-time quantitative PCR (qPCR) methods targeting 18S rDNA or LAXER locus, and (ii) evaluate the contribution of qPCR to the diagnosis and follow-up of cryptosporidiosis by comparing their performances to those of conventional microscopic techniques [ZN, auramine–phenol (AP), and glycerin] in a prospective study enrolling patients at risk for Cryptosporidium infection.

Materials and methods

Fecal sample collection

The fecal samples analyzed in this study included: (i) four human negative fecal samples of distinct origin (one child and three adults) and consistency (two soft, one liquid and one greasy stools, that were identified as S1, S2, S3, and S4 respectively), selected on the basis of negative results for cryptosporidia after ZN staining, and for other parasite cysts or ova after formalin–ether concentration, that were spiked with C. hominis oocysts isolated from a human diarrheal stool containing 2.105 oocysts per milliliter (P1) for comparison of two DNA extraction methods, (ii) 29 Cryptosporidium-positive fecal specimens from diarrhoeal patients (P2 to P30) stored at +4 °C in tubes containing 2.5 % potassium dichromate, and selected from the French ACNN collection [13], which included 13 C. hominis (P2 to P14), 12 C. parvum (P15 to P26) and 4 C. felis (P27 to P30) isolates, according to molecular identification performed as described previously [13], and (iii) 69 fecal specimens from 56 patients presenting gastro-intestinal symptoms, prospectively collected in the Parasitology–Mycology Laboratory of Lille University Hospital between September 2012 and April 2013. Fecal samples from subjects hospitalized in Lille University Hospital, considered at risk for cryptosporidiosis, were included in the study based on the following criteria: (i) liquid consistency (all adult or pediatric immunocompetent and immunosuppressed patients), or (ii) soft consistency if age was < 5 years, or if the patient had undergone a transplant, suffered from hematological malignancy, or an infectious disease. Stools submitted to the Parasitology–Mycology Laboratory of Lille University Hospital by other hospital centers for specific Cryptosporidium detection were also included.

Authorization for utilization of the stool isolates that were collected in Lille University Hospital was obtained from the French Ministry of Research (N°DC-2008-642). Furthermore, as required by French regulations, the ACNN collection was declared to the French Consultative Committee on Information Treatment of the Ministry of Research.

Light microscopic detection of Cryptosporidium

Microscopic detection of Cryptosporidium was performed after diphasic concentration for all samples, except those from the ACNN collection, for which the diagnosis of cryptosporidiosis had previously been established by microscopy in the ACNN laboratory of origin. The samples were subjected to formalin–ether concentration using a Parasep Fecal Parasite Concentrator system (DiaSys, Wokingham, UK) and were extemporaneously examined after glycerin preparation [35]. Two air-dried smears were prepared from stool concentrate, fixed in methanol, and stained by ZN (RAL, Paris, France) and AP (Sigma–Aldrich, Saint-Quentin Fallavier, France) techniques. Glycerin preparations and ZN-stained smears were observed at 400x magnification and under oil immersion at 1,000x under a light microscope respectively, whereas AP-stained smears were screened at 400x magnification and checked at 1,000x under the oil immersion lens of a fluorescent microscope.

Comparison of DNA extraction methods

Firstly, two extraction methods using the ZR Fecal DNA MiniPrep™ Kit (ZR) and the automated NucliSens® easyMAG® method (EM) were compared using aliquots from four Cryptosporidium-negative stools (S1 to S4) that were seeded with 104 C. hominis oocysts (S5 suspension) provided by P1 stool. DNA extractions were performed in triplicate from 150 mg of stool accordingly to the maximal starting amount of material compatible with ZR extraction kit. The automated EM method was adapted from the protocol previously defined by Mary et al. [32] [lower initial quantity of stool, i.e., 150 mg instead of 400 mg, and mechanical lysis using the MagNA Lyser instrument (Roche Diagnostics, Meylan, France) instead of the FastPrep-24 grinder, at 7,000 rpm for 70 sec with 425–600 μm beads (Sigma–Aldrich)], while the manual ZR extractions were carried out following strictly the steps in the manufacturer’s instructions.

Thereafter, the ZR and EM methods were further compared using positive stool samples from the ACNN collection. For each DNA extraction method, 1 ml of the stool that had been conserved in potassium dichromate was first washed with sterile water, and then centrifuged six times at 2,000 g for 10 min. DNA extraction was then performed on the pellet, as described above.

Lastly, the ZR kit was used for DNA extraction from the 69 stool specimens that had been prospectively collected between September 2012 and April 2013, and stored at −20 °C before DNA extraction.

For each DNA extract, the DNA yield (ng/μl) was determined using the Thermo ScientificTM NanoDropTM 3300 fluorospectrometer (Thermo Fischer Scientific, Inc., Illkirch, France).

Quantification of Cryptosporidium DNA using real-time qPCR

Two qPCR assays were performed on all DNA extracts. The first assay (18S qPCR), which was developed by Mary et al. [32], targets 18S rDNA, and is a pan-Cryptosporidium assay. The second assay (LAXER qPCR), which can detect C. parvum, C. hominis, and C. meleagridis, was designed to detect a 138-bp fragment positioned inside a specific 452-bp fragment sequenced by Laxer et al. [27, 36]. The qPCR technical conditions used in this study were those described by Mary et al. [32] and Fontaine & Guillot [36], except for the following points: (i) the 18S qPCR DNA template and mixture final volume were 2 μl and 20 μl respectively, instead of 1 μl and 25 μl, and (ii) the LAXER qPCR primers and probe concentrations were 0.4 μM and 0.1 μM respectively, instead of 0.3 μM and 0.2 μM, whereas the DNA template and mixture final volume were half of those described in the original publication.

18S and LAXER amplifications were performed with the LightCycler® 480 Thermal Cycling System (Roche Diagnostics) and ABI7500 Real-Time PCR System (Applied Biosystems®, Life Technologies, Saint Aubin, France) respectively. The detection limit of each qPCR method, which was determined using 10-fold serial dilutions of plasmid DNA carrying the target genes, reached 1 and 0.37 copies per microliter of DNA extract, i.e., 2.0 and 1.85 copies per reaction for the 18S and LAXER qPCRs respectively. Considering five copies of 18S [37] and one copy of LAXER [27] per sporozoite, i.e., 20 copies of 18S and four copies of LAXER per oocyst, these quantities were equivalent to 0.1 and 0.46 oocysts per reaction respectively, which corresponded to five and 23 oocysts in the total DNA extract (taking into account a 100-μl elution volume), or in the 150-mg initial amount of stool respectively. The qPCR efficiencies reached 94.5 % and 100 % respectively. In order to quantify Cryptosporidium DNA, plasmid suspensions containing 103 to 105 copies/μl or 3.68.103 to 3.68.105 copies/μl were included as standards in each 18S or LAXER qPCR series respectively. PCR inhibitor detection was performed using the TaqMan Exogenous Internal Positive Control reagent kit (Applied Biosystems). A negative control without a DNA template was included in each qPCR run. Each DNA extract was tested pure and diluted 10 times. Cryptosporidium detection was considered positive if at least one replicate was positive. The parasitic loads that are presented in tables and figures were calculated from the undiluted replicate, except when this replicate was negative. This exception only concerned one sample (see Table 1, stool P25), for which DNA quantification was obtained by multiplying by 10 the detected quantity.

Table 1 Comparison of the efficiency of two different commercial DNA extraction methods for the detection of Cryptosporidium DNA in 29 positive stool samples. ZR, ZR Fecal DNA MiniPrep™ Kit (Zymo Research); EM, NucliSens® easyMAG® (bioMérieux);
Full size table

In order to check the coherence between the Cryptosporidium DNA amounts detected with the 18S (5 gene copies [37]) and LAXER (single copy gene [27]) qPCRs, a 18S/LAXER ratio was calculated from Cryptosporidium DNA quantities (in copies/μl). These ratios were calculated for the 29 Cryptosporidium positive clinical specimens (P2 to P30, Table 1).

Nested-PCR detection and 18S rDNA genotyping of Cryptosporidium

Nested PCR was performed for all Cryptosporidium-positive stool specimens (ACNN collection and samples obtained from the prospective study), according to the protocol developed by Xiao et al. [23]. The isolates obtained from the prospective study were further identified by 18S rDNA sequencing using the 3500 Dx Genetic Analyser (Applied Biosystems). The sequences were compared to the GenBank database, using the BLAST program [38]. The DNA extracts from the prospective study were further tested with a typing 18S qPCR using C. parvum and C. hominis-specific probes, as described previously [32].

Statistical analysis

The mean total DNA and Cryptosporidium DNA quantities were compared by the Student's t- test, using GraphPad Prism software (version 6.01, GraphPad Software Inc., California, USA). Cryptosporidium DNA quantities in positive stool samples were compared using a Wilcoxon signed rank test. A p-value ≤ 0.05 was considered to be statistically significant.

Results

Comparison of ZR Fecal DNA MiniPrep™ Kit (ZR) and NucliSens® easyMAG® (EM) methods for molecular detection of Cryptosporidium oocysts in spiked human stools

The total DNA quantification in DNA extracts obtained from the suspension containing 104 oocysts of C. hominis (S5) and from the four spiked stool samples (S1 to S4) revealed significantly higher DNA quantities with the ZR kit for S3, S4, and S5 samples (p ≤ 0.01)  (Fig. 1a). Nevertheless, the DNA yields were similar for the EM and ZR extraction methods performed on S1 and S2.

Fig. 1
figure 1

Comparison of absolute DNA and Cryptosporidium DNA yields obtained from four spiked stool samples (S1-S4) and a suspension (S5) using two extraction methods. Seeded samples included two softs (S1-S2), one liquid (S3) and one greasy (S4) stools from one child and three adults. Absolute DNA quantities (a) were measured using NanoDropTM. Cryptosporidium DNA was quantified using 18S (b) or LAXER (c) qPCR. ** p ≤ 0.01. ZR, ZR Fecal DNA MiniPrep™ Kit (Zymo Research); EM, NucliSens® easyMAG® (bioMérieux)

Full size image

Detection of Cryptosporidium DNA by 18S qPCR was positive for all the samples. Except for greasy stool S4, Cryptosporidium DNA quantities detected were significantly higher with ZR than EM (Fig. 1b). Quantities were higher in the liquid stool S3 and in the oocyst suspension S5, with up to 57,800 copies/μl being detected with ZR (Fig. 1b). The weakest signals were obtained in the child soft stool S1 (where 110 copies/μl and 458 copies/μl were detected for the EM and ZR methods respectively), and in the greasy stool S4 (33 and 45 copies/μl respectively). Detection of PCR inhibitors was positive in the three DNA extracts from the paediatric soft stool S1 with the EM method. However, despite low DNA quantities were found in the greasy stool S4, no PCR inhibitors were detected in this stool. Furthermore, none of the ZR DNA extracts presented PCR inhibition.

When the number of Cryptosporidium LAXER copies was determined by qPCR, the ZR kit still yielded the best performance for stools S2, S3, and S5, for which 824, 7,600, and 5,540 copies/μl were detected with ZR and 0, 2,110, and 1,600 copies/μl with EM extracts respectively (Fig. 1c). ZR performances were also better on stool S1, for which a very weak signal was observed for one ZR replicate, whereas the three EM replicates were negative. However, the LAXER qPCR was negative in S4 ZR extracts, while Cryptosporidium DNA was detected in only one out of three EM replicates (five copies/μl) for this stool. Globally, fewer samples were positive using LAXER when compared to 18S qPCR for Cryptosporidium detection. Furthermore, around 10-fold lower DNA yields were systematically found with LAXER qPCR, and no amplification was observed in the S1 EM extracts containing PCR inhibitors (Fig. 1c).

Overall, the combination of any of the two DNA extraction methods with 18S qPCR yielded adequate detection of Cryptosporidium. Although only four seeded samples were tested, our data indicate that the performances of DNA extraction could be influenced by the origin and consistency of the stool. Since higher Cryptosporidium DNA yields and lower frequency of PCR inhibitors were obtained using the ZR kit, this method was selected for further evaluation of Cryptosporidium positive stool samples and comparison with automated extraction using EM.

Performance of ZR and EM methods for Cryptosporidium DNA extraction from positive stool samples

The total DNA quantities obtained from the 29 Cryptosporidium-positive human stool samples from the ACNN collection reached 0 to 70.7 ng/μl and 5.0 to 111.7 ng/μl with the ZR and EM methods respectively (Table 1). PCR inhibition was detected in no ZR or EM extracts. All the samples were positive with 18S qPCR for both DNA extraction methods, but Cryptosporidium DNA amounts were significantly higher in ZR than EM extracts (p = 0.002). Nested PCR was also positive for all ZR extracts, but surprisingly, four EM extracts from P2, P4, P15, and P27 stools, corresponding to two C. hominis, one C. parvum, and one C. felis isolates respectively, were negative. The Cryptosporidium LAXER qPCR was positive for all C. hominis and C. parvum samples using the EM method, but ZR extraction yielded two negative C. hominis samples (P13, P14), which corresponded to samples with very low DNA amounts using 18S qPCR (42 and 45 copies/μl respectively). Despite these two false negative samples with ZR, when DNA quantities were compared between ZR and EM extracts in the 23 samples that were positive with LAXER qPCRs, the Cryptosporidium DNA amounts that were detected in ZR extracts were found to be significantly higher than the ones detected in EM extracts (p = 0.004). The LAXER qPCR was negative for the four C. felis isolates (P27 to P30) with both the EM and ZR extracts. Moreover, when both LAXER and 18S qPCRs were positive, the latter yielded higher Cryptosporidium DNA amounts (Table 1), with an \( \frac{18S}{LAXER} \) median value reaching 5.8 and 6.6 for the ZR and EM kits respectively.

Prospective comparison of microscopy and molecular tools for the detection of Cryptosporidium in patients at risk

The stool samples collected for parasitological examination between September 2012 and April 2013 included 58 specimens from 47 patients hospitalized at Lille University Hospital, and 11 specimens from nine patients at other hospitals. The M:F gender ratio was 1.15, and underlying diseases included hematological malignancy (three stem-cell transplants, two myelodysplastic syndromes, one acute lymphoid leukemia, one acute myeloid leukemia, one multiple myeloma), kidney transplantation (n = 6), human immunodeficiency virus infection (n = 2), primary immunodeficiency (hyper-IgM syndrome, n = 1), auto-immune hepatitis (n = 1), and pancreatic adenocarcinoma (n = 1). Thirty-three patients (15 children and 18 adults) were immunocompetent. The immune status of four persons was unknown.

Cryptosporidiosis was diagnosed by microscopic detection in 12, 13, and 14 samples from 7/56 (12.5 %) patients after glycerin, ZN, and AP staining respectively. Among patients with cryptosporidiosis, two were hospitalized at Lille University Hospital and five at other hospitals. Therefore, the incidence of cryptosporidiosis in patients at risk reached 4.3 % (2/47) at Lille University Hospital. Taking into account all Lille University Hospital patients who underwent parasitological stool examinations between September 2012 and April 2013, the incidence of cryptosporidiosis was 0.3 % (2/651). The patients included three immunocompromised (two kidney transplants and one primary immunodeficiency) and four immunocompetent patients (Table 2). The molecular identification yielded five C. parvum and two C. hominis isolates, with full agreement between the DNA sequencing and typing qPCR.

Table 2 Characteristics of patients with cryptosporidiosis diagnosed at Lille University Hospital between September 2012 and April 2013
Full size table

In comparing the microscopic and molecular methods, 48 samples were negative with all techniques. All samples from patients with cryptosporidiosis were positive with both AP staining and 18S qPCR (n = 14, Table 3). For the latter, six supplementary samples were later positive (Ct > 40). The LAXER qPCR detected only 12 of the 14 microscopically positive samples. One supplementary sample was positive for a patient previously diagnosed with cryptosporidiosis (patient no. 3, Table 2).

Table 3 Comparison of microscopic and molecular methods for the detection of Cryptosporidium in stools (n = 69)
Full size table

Contribution of qPCR to the follow-up of cryptosporidiosis

When evaluating qPCR for the follow-up of cryptosporidiosis in two patients with biliary dissemination (patient no. 1, Table 2) or acute gastroenteritis (patient no. 2, Table 2), both the 18S and LAXER Cryptosporidium DNA loads were found to be correlated to disease severity. Firstly, for patient no. 1, an 18-year-old male with hyper-IgM syndrome suffering from sclerosing cholangitis, a dramatic increase in parasitic DNA loads was detected simultaneously with bioclinical worsening, revealed by strong elevation of hepatic enzymes combined with reinforcement of abdominal pains, despite treatment with nitazoxanide (Fig. 2a). Identically, for patient no. 2, a 12-year-old kidney transplant child with choleriform diarrhea, the Cryptosporidium DNA quantities rose simultaneously with both hepatic and pancreatic involvement despite nitazoxanide administration. Clinical improvement and a gradual decrease in the parasite DNA loads (until negativation by LAXER qPCR) were observed after treatment with intravenous immunoglobulins (IVIg), which was prescribed to treat a hypogammaglobulinemia at 3.3 g/l (Fig. 2b).

Fig. 2
figure 2

Follow-up of Cryptosporidium oocysts and DNA loads in two patients with chronic sclerosing cholangitis (a) and abrupt watery diarrhea secondarily complicated by acute pancreatitis (b). Considering the date of the first stool diagnosing cryptosporidiosis as day 0 (D0), nitazoxanide treatment (arrowhead) was initiated at D45 and D2 for patients nos. 1 and 2 respectively, while IVIg injections (arrow) were used only for patient no. 2 (D25 and D26). ALP, alkaline phosphatase (n = 35-–30 IU/l); GGT, γ-glutamyl-transpeptidase (n = 5–45 IU/l); LIP, lipase (n = 13–60 IU/l). Cryptosporidium oocysts semi-quantification using microscopy (glycerin staining): (+++), abundant (>1 oocysts/field at 400× magnification); (++), moderately abundant (>10 oocysts/slide); (+), scarce (1–5 oocysts/slide); (−), absent

Full size image

Discussion

When evaluating microscopy and molecular tools for the diagnosis of cryptosporidiosis, our prospective study confirmed the higher sensitivity of AP [16, 17]. Although two samples that were positive with AP were not detected with glycerin (yielding a lower sensitivity of 85.7 %), glycerin was positive in all the fecal specimens at the time of diagnosis, and only became negative in the course of cryptosporidiosis. Like the negative staining technique of Heine [39], glycerin is poorly known by medical biologists, and few laboratories use it for the routine diagnosis of cryptosporidiosis. However, it is a fast, reliable, and cost-effective method compared to ZN or AP. Thus, it could be widely used in patients with diarrhea in non-specialized laboratories. The high sensitivity of 18S qPCR confirmed the usefulness of this method for cryptosporidiosis diagnosis. As reported in previous publications revealing a significant increase in cryptosporidiosis cases (up to 2.5-fold) detected with 18S nested PCR in HIV-positive patients, or children with diarrhea [17, 20, 40], 18S qPCR positivity in six supplementary samples could correspond to those that were misdiagnosed using microscopic methods. By contrast, Calderaro et al. showed that real-time PCR, ICA, and fluorescent microscopy offered similar performance despite a large cohort (1,040 stools from 533 patients with a suspected intestinal parasitosis) [21]. However, this study was conducted in a low-prevalence area where an elevated proportion of light infection is expected, and the sensitivity of the molecular method that was used (multiplex qPCR with Cryptosporidium hominis/parvum detection using a LAXER targeting assay) could have been too low to increase the diagnosis of cryptosporidiosis in such a setting. In all, this study suggests that repeated sequences such as 18S rDNA could be a more appropriate target, but further large prospective studies are needed to confirm our data.

Our evaluation of the ZR Fecal DNA MiniPrep™ Kit and NucliSens® easyMAG® performed on spiked stools confirmed the usefulness of the EM automated method for Cryptosporidium DNA extraction [30, 32]. Furthermore, even if high performance has been reported for Blastocystis DNA extraction from stools with the ZR kit (94 % vs 48 % sensitivity for QIAamp® DNA Stool Mini Kit (QIA) in [41]), our data provides the first evidence of the efficiency of this method for Cryptosporidium DNA extraction from human stools. The significantly higher Cryptosporidium DNA quantities we obtained using ZR in most spiked stools and in stools from the ACNN collection could result from the mechanical pre-treatment step (bead-beating) which is systematically used prior to DNA isolation in this kit, and allow both optimal stool homogeneisation and oocyst disruption, whereas it was specifically added for EM extraction in our study, accordingly to Mary et al. who reported significantly enhanced DNA yields when adding bead-beating (by 2.17-fold, p < 0.0001) [32]. Bead-beating was also found by Halstead et al. [31] to improve yields for parasitic targets. Thermal treatment has also been reported to increase the sensitivity of Cryptosporidium detection from 60 % to 100 % in stools positive by microscopy after extraction with QIA [30]. These results underline the importance of mechanical or thermal pre-treatment for lysis of Cryptosporidium oocysts and optimal DNA extraction. If not included in the DNA extraction method provided by the supplier (such as for EM), it should be included as a supplementary step. Moreover, our results on S1 stool (Cryptosporidium-negative, seeded with oocysts) confirm that an efficient removal of PCR inhibitors, which was optimal with ZR kit, is essential. Although the presence of inhibitors in all S1 EM replicates did not completely prevent DNA amplification by 18S qPCR in this stool, it probably led to an underestimation of the quantities that were detected. Furthermore, the greater impact of PCR inhibitors on LAXER qPCR (which was negative for all EM DNA extracts) is consistent with its lower sensitivity (5-fold fewer target genes per sporozoite than 18S). It could also result from differences in inhibition strength according to the targeted locus.

Although Cryptosporidium DNA quantities were higher with ZR, both the EM and ZR kits allowed the efficient detection (100 %) of the parasite with 18S qPCR in Cryptosporidium-positive stool samples from the ACNN collection. 18S nested PCR was also 100 % sensitive with the ZR kit, but was not able to detect the parasite in four of the 29 EM extracts in our study. Such false-negative results have been reported previously by several authors [32, 42]. In our study, since nested PCR targets long DNA fragments (820 bp), the false negatives we observed could be related to DNA fragmentation resulting from mechanical pretreatment, which could have a lower impact on 18S qPCR since this assay targets a smaller DNA fragment (178 bp). As expected, the LAXER qPCR yielded lower sensitivity. These results were consistent with the intrinsic performance of this assay, which is specific for the detection of C. parvum and C. hominis, and does not cross-react with C. felis [36]. Additionally, two ZR DNA extracts from C. hominis isolates, which were found to be very weakly positive with 18S qPCR, were negative with the LAXER qPCR (Table 3). Once again, these data are consistent with the higher sensitivity of 18S qPCR, resulting from the higher number of 18S copies than LAXER copies (five vs one) in the sporozoite genome [27, 37]. Our results are in line with this statement, since the proportion of these two targets were respected, with an \( \frac{18S}{LAXER} \) median value reaching 5.8 and 6.6 for the ZR and EM extraction performed on Cryptosporidium-positive samples.

Beyond the differences in performance for DNA recovery and co-purification of PCR inhibitors, the choice of a DNA extraction method depends on several features including processing/hands-on time, cost, and the variety of validated sample types (Table 4). The NucliSens® easyMAG® is an IVD-labeled system allowing DNA extraction from a wide panel of specimens and volumes, well adapted to high-functioning laboratories. Moreover, automation enables up to 24 extractions simultaneously, releasing non-negligible hands-on time. The system also incorporates sample and reagent traceability, improving the safety of the process analysis. However, the main disadvantage of the EM platform remains the price of the robot, inadequate for small laboratories. The overall cost of the manual ZR Fecal DNA MiniPrep™ Kit is lower and provides high quality/quantity DNA from stools, particularly since bead-beating is processed prior to DNA extraction and inhibitors are efficiently removed. However, it requires a higher hands-on time and achieves lower standardization.

Table 4 Comparison of ZR Fecal DNA MiniPrep™ Kit (ZR) and NucliSens® easyMAG® (EM) based on technologies, hands-on and automated processing times, validated samples, input quantities, elution volume and pretreatment options
Full size table

Our prospective study revealed a 4.3 % incidence of cryptosporidiosis in patients at risk who were hospitalized at Lille University Hospital between September 2012 and April 2013, and a 0.3 % incidence among patients who underwent parasitological stool examination in Lille during this period. Taking into account all the patients with cryptosporidiosis who were diagnosed at Lille University Hospital (including the five patients staying at other hospitals), the incidence reached 1.1 % and was similar to the 1.1-2.3 % national prevalence estimated by the ACNN [13]. But our data shows an unexpected excess of Cryptosporidium cases in the North Region in this period (six, versus two or one cases reported in 2010 and 2011 respectively). Such an observation was also reported at the national level. Indeed, a 2-fold increase in notification of cryptosporidiosis cases to the ACNN was observed between September and December 2012, with 55 cases versus 29 and 22 cases notified in the same period in 2010 and 2011 respectively. Interestingly, this situation emerged simultaneously with a dramatic (1.8- to 4.9-fold) increase in cryptosporidiosis cases in three European countries, i.e., the Netherlands, the United Kingdom and Germany [43]. Increased reporting in 2012 compared with 2011 was also observed in Spain (268 %), Finland (127 %), Belgium (103 %), Ireland (35 %) and Slovenia (20 %) [44]. These observations support the value of systematic Cryptosporidium screening for patients presenting gastrointestinal symptoms, which is still insufficiently implemented in French laboratories, and leads to an underestimation of the cryptosporidiosis burden [13].

Although real-time PCR was found to be a reliable tool for the quantification of microsporidia and the monitoring of treatment efficacy [45], reports on Cryptosporidium infection follow-up with qPCR were lacking. In the present study, the Cryptosporidium DNA loads were found to be correlated with the disease severity in two cases of persistent cryptosporidiosis. The increase in DNA loads despite treatment with nitazoxanide in the two patients was consistent with the insufficient anti-Cryptosporidium activity of this drug among severely immunosuppressed patients [9]. For the kidney transplant child, the dramatic decrease of DNA loads after initiation of immunotherapy suggested that IVIg, which has already been reported for the treatment of several bacterial or viral infections [46], could also be useful in treating cryptosporidiosis in immunocompromised patients. These new data open the way for larger studies to evaluate the efficacy of IVIg for the treatment of cryptosporidiosis.

Altogether, these results indicate that, although little known, glycerin is a sensitive method for Cryptosporidium detection. Furthermore, the combination of the ZR Fecal DNA Kit and 18S qPCR was found to be an adequate procedure for Cryptosporidium DNA detection in human stools. qPCR was more sensitive than microscopy in patients at risk, and was also more sensitive and accurate in estimating the oocyst shedding in the course of infection. Furthermore, the 18S qPCR typing method was an accurate tool in providing rapid identification of the two most prevalent Cryptosporidium species. Using this procedure in patients at risk (i.e., children and immunocompromised individuals) could be helpful in preventing the misdiagnosis of cryptosporidiosis and quickly identifying the sources of infection in the case of outbreaks.