Evaluation of a New Device for Simplifying and Standardizing Stool Sample Preparation for Viral Molecular Testing with Limited Hands-On Time

Abstract

Sensitive molecular assays have greatly improved the diagnosis of viral gastroenteritis. However, the proper preparation of stool samples for clinical testing remains an issue. bioMérieux has developed a stool preprocessing device (SPD) that includes a spoon for calibrated sampling and a vial containing buffer, glass beads, and two filters. The resulting stool filtrate is used for nucleic acid extraction. The purpose of this study was to evaluate the performance of the SPD for the quantification of human adenovirus (HAdV) DNA in stool samples collected from hematopoietic stem cell transplant (HSCT) recipients. HAdV DNA was quantified with the Adenovirus R-gene kit. The suitability of the device to reproducibly quantify HAdV DNA in stools using different extraction platforms (easyMAG and QIAsymphony) was determined using archived HAdV-positive stool samples. Coefficients of variation of HAdV DNA quantifications ranged from 1.79% to 1.83%, and no difference in quantification was observed between the two extraction systems. The HAdV DNA limit of quantification using the SPD was 3.75 log₁₀ copies/g of stool. HAdV DNA quantification using the SPD was then compared to that of the routine preprocessing technique on 75 fresh stool samples collected prospectively from pediatric HSCT recipients at risk for HAdV infections. Thirty-eight samples were HAdV DNA positive with both the SPD and routine preprocessing methods. HAdV DNA loads were on average 1.14-log₁₀ copies/g of stool higher with the SPD (P < 0.0001) than with routine methods. This new device enabled a standardized preparation of stool samples in <5 min and a reproducible and sensitive quantification of HAdV DNA. The use of the SPD for the detection of other gastrointestinal infections warrants further evaluation.

INTRODUCTION

PCR is a very rapid and reliable tool for molecular biology-based diagnosis of a variety of infectious diseases, using a wide range of clinical sample types (1).

However, stool samples are among the most difficult clinical samples to process for molecular pathogen detection because of the presence of very potent inhibitors of nucleic acid (NA) amplification that are often coextracted along with pathogen NAs (2–4). Bile salts, hemoglobin, polysaccharides, heme, and bilirubin have been identified as factors that inhibit amplification assays (3, 5–8). Inhibitory effects can be reduced by adding amplification facilitators, such as bovine serum albumin, to the PCR mixture (9) by using thermostable polymerases that are more resistant to PCR inhibition (10), or by using more efficient processes for extracting NAs from stool samples (11). The efficiency of NA extraction and purification has been shown to influence the sensitivity, reproducibility, and accuracy of NA amplification-based target detection (12).

Due to the potential presence of amplification inhibitors, sample preprocessing before NA extraction is an essential step to improve amplification-based pathogen detection from stool samples. Currently, stool preparation is not standardized and can be labor-intensive and time-consuming. Few studies have evaluated or compared simple DNA extraction methods that would facilitate and improve the sensitivity of qualitative PCR detection of enteric pathogens from stool specimens (12–15, 27).

In addition to use in qualitative detection of pathogens, quantification in stool may help in patient management. Pediatric recipients of unrelated hematopoietic stem cell transplant (HSCT) are at high risk for disseminated human adenovirus (HAdV) infections. This risk warrants routine HAdV DNA screening, including HAdV quantification in stool (16). Reports have shown that HAdV DNA levels in stool are predictive of disseminated infection (17, 18). However, there is no standardized method for the quantification of HAdV DNA in stool samples, and this hampers any comparison of quantitative analyses between studies (16). Therefore, interlaboratory reproducibility of pathogen quantification in stool samples is a challenging issue. This highlights the need for a standardized technique for stool sample extraction.

bioMérieux has developed a stool preprocessing device (SPD) designed to simplify the extraction of NAs from stool samples. The SPD includes a spoon to aliquot a consistent volume of stool and a vial containing 5 ml of buffer, glass beads to improve sample homogenization, and two filters that clarify the suspension. This system provides a filtered solution that can be used directly for NA extraction. The aim of this study was to evaluate the effect of the SPD on the reproducibility and sensitivity of HAdV DNA extraction and quantification of stool samples collected from HSCT recipients. The SPD was evaluated using two different NA extraction systems and two real-time PCR thermocyclers. The performance of the SPD was compared with that of the laboratory's routine preprocessing technique.

MATERIALS AND METHODS

Patients and stool samples.

A total of 125 clinical stool samples from 40 patients were included in the study, with 50 retrospective samples collected between December 2012 and August 2013 (kept frozen at −80°C) and 75 fresh prospective samples.

Initially, to assess the intra-assay reproducibility of the results after SPD processing, a preliminary retrospective study was conducted using 20 HAdV DNA-positive frozen stool samples from 12 HSCT recipients hospitalized at Robert-Debré Hospital (Paris, France). All samples were preprocessed in triplicate using the SPD. Each triplicate sample was extracted with two extraction systems, the easyMAG (bioMérieux, Marcy l'Etoile, France) and QIAsymphony (Qiagen, Courtaboeuf, France). Each extract was then quantified in triplicate for HAdV DNA using the quantitative real-time PCR Adenovirus R-gene assay (bioMérieux/Argene, Verniolle, France) on an ABI 7500 thermocycler (Applied Biosystems, Carlsbad, CA).

To evaluate the differences in HAdV DNA quantification between the easyMAG and QIAsymphony NA extraction systems and between the ABI 7500 and LightCycler 480 (LC480) thermocycler (Roche Applied Science, Meylan, France) amplification systems, 30 additional frozen stool samples from the follow-ups of 5 HSCT recipients with active HAdV infection were tested in duplicate. This series included for each patient the last HAdV DNA-negative sample before the first HAdV DNA-positive sample and the first HAdV DNA-negative sample after the last HAdV DNA-positive sample, with the aim to evaluate if SPD preprocessing improved the sensitivity of HAdV detection compared to that of the routine preprocessing technique, thereby enabling the detection of HAdV reactivation earlier and for longer time periods.

Finally, 75 fresh stool samples from 23 pediatric HSCT recipients at risk for HAdV infections (cord blood graft or age <6 years) were prospectively tested with the routine preprocessing technique and the SPD. The same NA extraction system (QIAsymphony) and HAdV DNA quantification assay (Adenovirus R-gene assay on ABI 7500) were used for the two methods. All stool samples were qualified using the Bristol scale. The Bristol stool scale classifies the form of human feces into seven categories. Types 1 and 2 indicate constipation, types 3 and 4 are the ideal stools, and types 5, 6, and 7 tend toward diarrhea (19). HSCT patients usually do not have constipation in the weeks after transplantation, especially when an HAdV infection occurs. We thus did not have any type 1 or 2 stools to test.

Pretreatment and extraction. (i) Routine technique.

Before extraction, stool specimens were prepared by diluting 1 ml of diarrheic stool (Bristol scale index, 6 and 7) or 1 g for solid stool in 9 ml of phosphate-buffered saline (PBS). The resulting suspension was subjected to two −20°C freeze-thaw cycles of 30 min and then centrifuged (30 min at 5,000 × g). The supernatant was passed through a 0.45-μm-pore-size filter (Minisart Plus syringe filters; Sartorius Stedim Biotech GmbH, Göttingen, Germany). NAs were purified from 200 μl of preprocessed sample and eluted in 115 μl using the QIAsymphony system and the DSP virus/pathogen minikit (Qiagen). Ten microliters was used for HAdV quantitative real-time PCR. HAdV DNA quantities were expressed in copies per gram of stool, as follows:

$$copies/g=copies/PCR\times 1(\text{g}\hspace{0.17em}\text{of}\hspace{0.17em}\text{stool})\times \frac{115}{10}\left(\frac{\text{volume of QIAsymphony eluate}}{\text{volume used for the PCR}}\right)\times \frac{10,000}{200}\left(\frac{\text{volume of stool suspension}}{\text{volume of stool filtrate used for extraction}}\right)$$

(ii) SPD.

The SPD is composed of a spoon to aliquot a constant volume of stool, glass beads to improve sample homogenization, and two filters (<150 μm for the first and <30 μm for the second), located in the vial cap, to clarify the suspension. The reagents are stable for at least 6 months at room temperature. Briefly, 220 mg of stool was aliquoted with the calibrated spoon and diluted in 5 ml of buffer. The reproducibility of the spoon calibration was validated in separated experiments (data not shown) in which the amount of stool released in the bottle was weighted. Stools representative of different consistencies were used, and 160 weightings were performed. The mean weight of stool released by the spoon was of 220 mg (standard deviation, 21 mg). This suspension was homogenized by vortexing and clarified by squeezing the bottle, thereby pushing the suspension through the two filters into a clean collection tube. With the easyMAG system, NAs were purified from 400 μl of filtrate and eluted in 50 μl, according to the manufacturer's instructions. HAdV DNA quantities were expressed in copies per gram of stool, as follows:

$$copies/g=copies/PCR\times \frac{1}{0.22\left(\text{g of stool}\right)}\times \frac{50}{10}\left(\frac{\text{volume of easyMAG eluate}}{\text{volume used for the PCR}}\right)\times \frac{5,000}{400}\left(\frac{\text{volume of stool suspension}}{\text{volume of stool filtrate used for extraction}}\right)$$

With the QIAsymphony system, NAs were purified from 200 μl of filtrate and eluted in 115 μl, according to the manufacturer's instructions. HAdV DNA quantities were expressed in copies per gram of stool, as follows:

$$copies/g=copies/PCR\times \frac{1}{0.22\left(\text{g of stool}\right)}\times \frac{115}{10}\left(\frac{\text{volume of QIAsymphony eluate}}{\text{volume used for the PCR}}\right)\times \frac{5,000}{200}\left(\frac{\text{volume of stool suspension}}{\text{volume of stool filtrate used for extraction}}\right)$$

(iii) Internal control.

Ten microliters of the IC2, an internal control (IC) provided in the DICO Extra r-gene kit (bioMérieux), was added to each filtrate before extraction to check the entire testing process, including DNA extraction, and possible PCR inhibition during amplification.

Real-time quantitative PCR.

HAdV DNA was detected and quantified with the Adenovirus R-gene kit (bioMérieux), according to the manufacturer's instructions. The assays were performed using an ABI 7500 and/or an LC480 thermocycler. The results are expressed in log₁₀ copies per gram of stool.

LOD determination.

To evaluate and compare the limit of detection (LOD) of the routine preprocessing and the SPD techniques, 10 replicates of six successive dilutions of HAdV DNA-positive stool, prepared using a pool of HAdV DNA-negative stool samples, were tested after preparation using the two techniques. The expected HAdV DNA loads were 5.85, 5.15, 4.45, 3.75, 3.45, and 3.05 log₁₀ copies/g of stool. Each replicate was extracted with easyMAG and quantified on an ABI 7500 thermocycler.

Data analysis.

Statistical analyses were performed on GraphPad Prism version 6.0. The LOD was determined by probit analysis with the Minitab 16 software. All tests were two-sided, with P values of ≤0.05 denoting statistical significance. Comparisons of HAdV DNA quantitative values between the two NA extraction systems and the two real-time PCR amplification systems were done with a paired-samples Wilcoxon test. The agreement between tests was analyzed by Pearson correlation and calculation of Cohen's kappa (20). The comparison of HAdV DNA quantification between the routine technique and that using the SPD on prospective samples was done with a paired-samples Student t test. The bias between the routine preprocessing method and that using the SPD was analyzed by Bland-Altman plots (21). Figures were constructed using GraphPad Prism version 6.0.

RESULTS

Stool preprocessing workflow using the SPD.

A prototype device has been designed to integrate the main steps of stool preprocessing prior to NA extraction. The workflows using both the SPD and the routine laboratory preprocessing technique are presented in Fig. 1. The routine preprocessing used includes many steps, as shown in Fig. 1. The freeze-thaw cycles of the routine preprocessing method provide a stool filtrate that is suitable for both culture and molecular biology testing, as no toxicity was observed in cell culture, and no PCR inhibition was encountered. The routine method, however, required >180 min. The SPD tested in the study was developed mainly for molecular analysis. No virus isolation could be performed with the stool filtrate obtained with the SPD, as the buffer was toxic for cell culture. However, the SPD is very rapid, with one to eight stool filtrates prepared within 5 min. Stool filtrates can be stored at 4°C for 24 h before nucleic acid extraction (data not shown).

FIG 1.

Comparative workflows of stool sample processing with the routine technique and the SPD. Top, routine technique. Stool specimens were prepared by dilution of 1 ml or 1 g of stool in 9 ml of phosphate-buffered saline (PBS) (step 1). The resulting suspension was homogenized using a vortex for 30 s (step 2), subjected to two freeze-thaw cycles (freezing at −20°C for 30 min) (step 3), and centrifuged (30 min, 5,000 × g) (step 4). The supernatant was then passed through a 0.45-μm-pore-size filter (step 5). Bottom, SPD technique. Two hundred twenty milligrams of stool was collected using the calibrated spoon (step 1). The spoon was placed into the bottle containing glass beads and 5 ml of buffer (step 2). The suspension was homogenized by vortexing the bottle for 30 s (step 3). The suspension was then clarified by squeezing the suspension through the two filters into a clean tube (step 4).

Reproducibility and limit of detection.

A total of 20 stool samples found positive for HAdV DNA with the routine technique (average, 5.20 log₁₀ copies/g of stool; range, 3.42 to 7.74 log₁₀ copies/g of stool) were tested in triplicate to assess the reproducibility of results after preprocessing with the SPD technique. The tested stool samples had scores ranging from 4 to 7 using the Bristol scale. Each SPD triplicate was extracted with the QIAsymphony and easyMAG extraction systems, and each extract was quantified for HAdV DNA in triplicate using Adenovirus R-gene in the same assay.

For each sample, coefficients of variation (CV) between triplicate samples were calculated. The average CV was 1.79% (95% confidence interval [CI], 1.07 to 2.52%) for the easyMAG and 1.83% (95% CI, 0.94 to 2.71%) for the QIAsymphony. There was no significant difference in assay reproducibility when extracted using either the easyMAG or QIAsymphony (P = 0.84).

For stool samples with Bristol scale score equal or inferior to 5 (n = 10), the average CV was 1.18% (95% CI, 0.76 to 1.60%) for the easyMAG and 1.33% (95% CI, 0.75 to 1.90%) for the QIAsymphony, while for samples with a Bristol scale score superior to 5 (n = 10), the average CV was 2.41% (95% CI, 0.99 to 3.81%) for the easyMAG and 2.33% (95% CI, 0.53 to 4.14%) for the QIAsymphony. A comparison of the average CVs showed no significant difference according to their Bristol score (stool consistency) for easyMAG (P = 0.08) or QIAsymphony (P = 0.25).

Probit analysis predicted LOD of 3.75 and 4.30 log₁₀ copies/g of stool for the SPD and routine preprocessing, respectively (see Table S1 in the supplemental material).

Differences in HAdV DNA quantification.

Differences in HAdV DNA quantification after NA extraction with the easyMAG and QIAsymphony were tested retrospectively using 30 successive stool samples collected from 5 HSCT recipients in the context of surveillance for HAdV infection. All SPD samples were tested in duplicate and were previously tested with the routine preprocessing technique. HAdV DNA was detected in 28 SPD samples using easyMAG and in 26 using QIAsymphony/ABI 7500 (Table 1). There was no significant difference in HAdV DNA quantification between the easyMAG (mean HAdV DNA load, 5.15 log₁₀ copies/g of stool; 95% CI, 4.23 to 6.07 log₁₀ copies/g of stool) and QIAsymphony (mean HAdV DNA load, 4.92 log₁₀ copies/g of stool; 95% CI, 3.90 to 5.94 log₁₀ copies/g of stool). In addition, we tested the differences in quantification of HAdV between the ABI 7500 and LC480 amplification systems (see Table S2 in the supplemental material). There was no significant difference in HAdV DNA quantification between the two systems, regardless of whether extraction was performed using the easyMAG or the QIAsymphony (see Fig. S1 in the supplemental material).

TABLE 1.

Retrospective comparison of HAdV DNA quantification between easyMAG and QIAsymphony for nucleic acid extraction on 30 stool samples from 5 HSCT recipients^a

Patient	Stool sample	Bristol scale score	HAdV results by technique (log10 copies/g of stool)
			Routine (QIAsymphony)	SPD
				easyMAG	QIAsymphony
1	1	6	Undetb	3.44c	4.45
	2	6	3.95	4.76	3.37
	3	5	4.02	4.25	3.57
	4	6	4.26	4.33	Undet
	5	6	Undet	3.73	3.61
	6	4	Undet	3.12	3.74c
	7	6	4.08	4.51	4.33
	8	7	4.66	5.00	5.00
	9	6	3.44	4.48	4.02
	10	7	Undet	3.55	4.15
2	1	7	Undet	2.77	Undet
	2	6	7.28	8.44	8.37
	3	6	8.66	9.63	9.63
	4	6	9.80	10.12	10.05
	5	6	7.00	8.05	8.12
	6	6	Undet	4.75	4.76
3	1	4	Undet	5.88	5.98
	2	4	3.85	5.77	5.83
	3	4	3.46	5.30	5.26
	4	4	Undet	5.53	5.66
4	1	6	Undet	Undet	Undet
	2	6	4.06	4.94	4.79
	3	6	5.91	7.52	7.47
	4	6	4.56	5.31	4.75
	5	6	Undet	Undet	2.18
5	1	4	Undet	1.24c	Undet
	2	7	5.55	8.00	7.97
	3	6	5.98	7.87	7.98
	4	5	4.02	6.49	6.74
	5	6	Undet	5.71	5.76

^aStool samples were preprocessed before nucleic acid extraction with the technique used in the laboratory (HAdV routine technique) or with the stool preprocessing device (HAdV SPD technique). The samples were extracted and tested in duplicate. All samples were quantified with the Adenovirus R-gene kit (bioMérieux) on an ABI 7500 thermocycler.

^bUndet, undetectable.

^cSingle value.

Compared to the routine preprocessing method, HAdV reactivation was detected earlier using the SPD in two patients (patients 1 and 3) (Table 1). In two other patients (patients 2 and 5), the detection of HAdV DNA by the SPD on the last negative HAdV DNA sample before the first positive HAdV DNA (with routine preprocessing) was inconsistent, depending on the extraction method used (easyMAG positive, QIAsymphony negative). For these two samples, the HAdV DNA load at this time point was below the LOD. In four out of five patients, the first negative HAdV DNA samples after the last positive HAdV DNA samples (routine preprocessing) were HAdV DNA positive using the SPD. These results suggest that the SPD might increase the sensitivity of HAdV DNA detection in stools and might enable the detection of HAdV reactivation earlier.

Prospective analysis.

Seventy-five fresh stool samples (Bristol scale index range, 4 to 7, including 50 samples at 6 or 7) from 23 pediatric HSCT recipients at risk for HAdV infections were prospectively tested with our routine preprocessing and the SPD. The same NA extraction system (QIAsymphony) and HAdV DNA quantification assay (Adenovirus R-gene assay on ABI 7500) were used for the two methods.

The overall concordance was good (kappa = 0.78, P < 0.01). Out of 75 samples, 38 samples were HAdV DNA positive with the two preprocessing techniques. In addition, five stool samples that were HAdV DNA negative with routine preprocessing were HAdV DNA positive with the SPD (HAdV loads, 0.72, 3.59, 4.44, 4.86, and 5.03 log₁₀ copies/g of stool), and 3 stool samples that were HAdV DNA negative with the SPD were HAdV DNA positive with routine preprocessing (HAdV loads, 2.88, 3.23, and 3.34 log₁₀ copies/g of stool). There was a correlation between the viral loads of the 38 samples that were HAdV DNA positive with both preprocessing techniques (r² = 0.95; P < 0.0001). Among samples that were HAdV DNA positive with both preprocessing techniques, the viral loads measured with the SPD were significantly higher (average, 1.14 log₁₀ copies/g of stool; P < 0.0001) than those measured with the routine preprocessing technique (Fig. 2).

FIG 2.

Bland-Altman analysis of HAdV DNA load measured prospectively with the routine technique and with the SPD. Differences between HAdV DNA loads with routine preprocessing and the SPD are shown for 75 prospectively collected fresh stool samples. The mean difference (bias) is indicated by a solid line and the 95% confidence interval (±1.96 standard deviations) is represented by dotted lines. Adv, adenovirus; gr, gram.

DISCUSSION

We have evaluated a new device for the preparation of stool specimens for NA extraction. The device was very simple to handle and provided a filtrate suitable for extraction in both systems tested within a very limited turnaround time (about 5 min). Our results showed that the device reproducibly quantified HAdV DNA in stool samples when using different extraction and amplification systems. In addition, the performance of the SPD was not affected by the stool form as estimated with the Bristol scale. Compared to the laboratory's routine preprocessing technique, the SPD provided a better LOD and increased slightly the detection rate of HAdV-infected samples. In the routine monitoring of patients, the SPD enables the detection of HAdV replication in the gastrointestinal tract earlier and for longer time periods.

Previous studies have shown the correlation between the HAdV DNA load in stool and the risk of HAdV disseminated infection and suggested that quantitative molecular surveillance of HAdV DNA in stool should be implemented to anticipate HAdV disease in high-risk patients (17, 18). However, quantification results from different studies cannot be compared because of the diversity of techniques used and the lack of an international HAdV quantitative standard by which assays can be equivalently calibrated. The most critical point is the calibration of the quantity of stool that is collected and extracted. The device tested in this study included a calibrated spoon to collect 220 mg of stool samples and glass beads in the bottle with buffer to homogenize the stool suspension. The combination of the two components was likely a key factor for providing the very good reproducibility of quantitative results. Therefore, such a device could be helpful to standardize HAdV quantification in stool.

Although the laboratory preprocessing technique used a larger volume of stool, the SPD detected some HAdV DNA-positive samples earlier and detected more DNA than the laboratory technique. The difference in the pore sizes of the filters between the two techniques was the main reason for the better sensitivity of the device (data not shown). A 0.45-μm-pore-size filter was used in the routine preprocessing technique in order to eliminate any cytotoxic compounds that might hamper virus isolation and any bacteria to avoid cell culture contamination. Although 0.45 μm is larger than the size of HAdV virions, multivesicular bodies and small-membrane parts with embodied viruses may have been retained within the filter. Freeze-thaw cycles used in the routine preprocessing may also have compromised DNA integrity and may have reduced the sensitivity of HAdV DNA detection. Conversely, freeze-thaw cycles are used in cell culture to recover intracellular virus particles from infected cells by disrupting cell membranes and have been found to enhance virus detection from environmental samples (22–24). In addition, freeze-thaw cycles have been found to be efficient in removing PCR inhibitors (11). Three samples that were HAdV DNA negative with the SPD were found to be HAdV DNA positive with routine preprocessing. All three samples had DNA loads below the SPD LOD and thus would not be expected to be detected consistently. Despite a seeming difference in the overall sensitivity of HAdV quantification, HAdV DNA loads for samples that were positive with both techniques were highly correlated.

Aside from HAdV DNA quantification, such a device could be used for the diagnosis of other pathogens involved in gastroenteritis. Indeed, the pore sizes of the SPD filters are large enough to collect bacteria and parasites. Recently, several multiplex molecular assays have been developed to detect several pathogens, including viruses, bacteria, and parasites, in a single assay (25). They demonstrate higher sensitivities than those of conventional methods, with reduced turnaround and hands-on times. However, because of increased sensitivity, positive detection may be observed in asymptomatic individuals, and other independent methods may be required to confirm active infection (26). Few reports have shown an association between the quantity of pathogens in stool specimens and diarrhea (28–30). Both qualitative molecular detection and quantification of viral, bacterial, and parasitic enteric pathogens in stool specimens can refine the diagnosis of gastroenteritis and help to ascribe an etiology to coinfected samples.

To conclude, this stool preprocessing device enables a standardized pretreatment and calibration of stool samples and provides reproducible and sensitive quantification of HAdV DNA. The use of the SPD for the diagnosis of other gastrointestinal infections must be evaluated.