Evaluation of Anatomically Designed Flocked Rectal Swabs for Molecular Detection of Enteric Pathogens in Children Admitted to Hospital with Severe Gastroenteritis in Botswana

Abstract

Two-hundred eighty matched bulk stool and anatomically designed flocked rectal swab samples were collected from children admitted to the hospital with acute diarrhea in Botswana. Their parents were asked about the acceptability of the swab collection method compared with bulk stool sampling. All samples underwent identical testing with a validated 15-target (9 bacterial, 3 viral, and 3 parasite) commercial multiplex PCR assay. The flocked swabs had a 12% higher yield for bacterial pathogen targets (241 versus 212; P = 0.003) compared with that of stool samples, as well as similar yields for viral targets (110 versus 113; P = 0.701) and parasite targets (59 versus 65; P = 0.345). One hundred sixty-four of the flocked swab-stool pairs were also tested with separate laboratory-developed bacterial and viral multiplex assays, and the flocked rectal swabs had a performance that was similar to that seen with commercial assay testing. Almost all parents/guardians found the swabs acceptable. Flocked rectal swabs significantly facilitate the molecular diagnosis of diarrheal disease in children.

INTRODUCTION

Diarrheal disease remains a leading cause of global childhood morbidity and mortality, yet access to diagnostic laboratory testing is rarely available in much of the world. One of the barriers to diagnosing diarrheal disease, either for clinical or surveillance purposes, is the difficulty and time delays in obtaining and transporting a bulk stool specimen. Several investigators have sought to overcome this barrier through the use of rectal swab specimens for culture, molecular, and antigen testing, with variable results (1–5). Flocked swabs designed for respiratory and genitourinary sampling have been shown to acquire better samples than those acquired with more traditional spun fiber swabs (6, 7). We used a specially designed flocked rectal swab (FLOQSwabs; Copan Italia, Brescia, Italy) developed specifically for the diagnosis of diarrheal disease in children (Fig. 1) and then compared matched flocked rectal swabs to bulk stool samples in a clinical setting. The samples were collected from children admitted to the hospital in Botswana with severe acute gastroenteritis and tested using a U.S. FDA-cleared commercial multiplex PCR assay in order to assess performance across a broad number of bacterial, viral, and parasitic pathogens.

FIG 1.

Pediatric rectal enteropathogen-flocked swab.

(These data were presented in part at the 29th Annual Clinical Virology Symposium, Daytona Beach, FL, 28 April to 1 May 2013, and at the Annual Pediatric Academic Society Meeting, Vancouver, Canada, 5 May 2014.)

MATERIALS AND METHODS

Children <13 years of age who were admitted to the hospital with a diagnosis of acute gastroenteritis were enrolled prospectively at the Princess Marina Hospital in Gaborone, Botswana. Princess Marina Hospital is the largest referral hospital in Botswana.

Clinical data were collected, and both the pediatric flocked rectal swab and bulk stool samples were obtained from each child as soon as possible after enrollment. The swab and stool samples were collected simultaneously, if possible; otherwise, bulk stool was collected as soon as possible after rectal swab collection. The stool samples were collected and transported in sterile containers kept in cooler boxes containing ice packs and then stored within 6 h of collection at −80°C. The parents or guardians of children who had both swab and stool specimens collected were asked about the acceptability of rectal swab specimen collection compared with bulk stool collection using a 5-point Likert scale. The parents or guardians gave signed consent, and the research protocol was approved by ethics committees at the University of Botswana, Botswana Ministry of Health, Princess Marina Hospital, University of Pennsylvania, and McMaster University (Hamilton, Ontario, Canada).

The specimens were stored at −80°C in dry swab tubes/cryovials prior to shipment on dry ice to McMaster University for testing. All matched swab-stool pairs underwent identical processing at the same time and by the same technologist. The preanalytical processing methods are shown in Table 1. There was a transition to the easyMAG extraction for the samples collected after 29 January 2013, as this platform became available at the Botswana laboratory, and the establishment of on-site validation and testing was planned. Ten-microliter aliquots of 1.0 × 10⁹ PFU/ml MS2 bacteriophage (catalog no. 0820002; Luminex Molecular Diagnostics) and 6 × 10⁸ CFU/ml Agrobacterium tumefaciens (strain ATCC 33970) were added prior to pretreatment as RNA and DNA internal positive controls, respectively. Reverse transcription, amplification, and detection of 15 pathogen targets (3 viruses, 3 parasites, and 9 bacteria) were performed using the Gastrointestinal Pathogen Panel (GPP) assay (Luminex Molecular Diagnostics, Toronto, Canada) on the MAGPIX system, as per the manufacturer's instructions. The GPP assay has been evaluated (8, 9) and simultaneously detects the following pathogen analyte-specific reagents (ASRs): Giardia, Cryptosporidium, Entamoeba histolytica, Yersinia enterocolitica, Salmonella, Escherichia coli heat-stable (ST) enterotoxin, E. coli heat-labile (LT) enterotoxin, Shigella, Clostridium difficile toxin A, C. difficile toxin B, Campylobacter, Vibrio cholerae, E. coli O157, Shiga toxin 1, Shiga toxin 2, norovirus GI, norovirus GII, rotavirus A, and adenovirus 40/41. The samples collected from 6 September 2012 until 29 January 2013 (n = 164) were also tested in parallel with two laboratory-developed multiplex PCR assays, one targeting the three most prevalent bacterial pathogens (Salmonella spp., Shigella spp., and Campylobacter jejuni-C. coli) (10–12) and the other targeting rotavirus A, norovirus GI/GII, and all adenoviruses (13–16). These assays were adapted from the literature, and the primer and probe sequences are listed in Table 2. Five microliters of extracted nucleic acid from the matched stool and swab samples (processed as described above) was added to the primers, probes, and mastermix reagents, and the QuantiTect multiplex no ROX PCR kit (catalog no. 204743; Qiagen, Mississauga, Ontario, Canada) for the bacterial multiplex and QuantiTect Virus + ROX Vial kit (catalog no. 211033; Qiagen) was used for the viral multiplex. Statistical analysis was performed using Stata statistical analysis software version 11 (StataCorp, College Station, TX). McNemar's test for paired samples was used to assess swab versus bulk stool detection of the target pathogens. The bacterial, viral, and parasitic targets were each analyzed separately. The sensitivity for each pathogen ASR was calculated using the reference standard of the presence of the ASR in either sample. A paired t test was used to compare the threshold cycle (C_T) values in the matched positive samples tested by the LDT assays.

TABLE 1.

Preanalytical processing methods used in the study

Processing step	Matched samples collected from:
	6 September 2012 to 29 January 2013 (n = 164)	30 January to 16 August 2013 (n = 116)
1	Swabs were eluted in 1 ml of eNAT (Copan Italia, Brescia, Italy), and 100 μl of the matching stool sample was added to 900 μl of eNAT; both tubes contained 1-mm glass beads (BioSpec Products, Bartlesville, OK)	Swabs were eluted in 1 ml of easyMAG lysis buffer (bioMérieux, Durham, NC), and 100 μl of the matching stool sample was added to 900 μl of easyMAG lysis buffer; both tubes contained Bertin SK38 soil mix beads (BioAmerica, Inc., Miami, FL)
2	5 min of lysis via vortex mixing and then held for 10 min at room temp, followed by centrifugation at 14,000 rpm for 2 min	5 min of lysis via vortex mixing and then held for 10 min at room temp, followed by centrifugation at 14,000 rpm for 2 min
3	Crude lysates stored at −80°C until nucleic acid extraction performed using 200 μl of cleared supernatant via QIAsymphony (Qiagen, Germantown, MD, USA) using the DSP virus/pathogen minikit with an elution vol of 70 μl	Crude lysates stored at −80°C until nucleic acid extraction performed using 200 μl of cleared supernatant via NucliSENS easyMAG using extraction specific B protocol and an elution vol of 70 μl

TABLE 2.

Pathogen primers and probes used in the laboratory-developed multiplex assays

Pathogen target	Forward primer	Reverse primer	Probe sequence	Target	Reference
Salmonella	CTCACCAGGAGATTACAACATGG	AGCTCAGACCAAAAGTGACCATC	CAC CGA CGG CGA GAC CGA CTT T	ttr gene	10
Shigella	CCTTTTCCGCGTTCCTTGA	CGGAATCCGGAGGTATTGC	CGC CTT TCC GAT ACC GTC TCT GCA	ipaH gene	11
Campylobacter	CTGCTTAACACAAGTTGAGTAGG	TTCCTTAGGTACCGTCAGAA	TGTCATCCTCCACGCGGCGTTGCTGC	16S rRNA	12
Rotavirus A	GGAKGTYCTGTACTCMTTGTCA	CCAGTTTGRAASTCATTTCC	GAATATAAT/ZEN/GTACCTTCRACAATTTTGTCYCTAGCATC	VP6 gene	13
Norovirus GI	CGYTGGATGCGNTTYCATGA	CTTAGACGCCATCATCATTYAC	AGATYGCGRTCYCCTGTCCA	RNA polymerase/capsid	14, 15
Norovirus GII	CARGARBCNATGTTYAGRTGGATGAG	TCGACGCCATCTTCATTCACA	TGGGAGGGCGATCGCAATCT	RNA polymerase/capsid	14, 15
Adenovirus	CAGGACGCCTCGGRGTAYCTSAG	GGAGCCACVGTGGGRTT	CCGGGTCTGGTGCAGTTTGCCCGC	Hexon	16

RESULTS

Specimens were collected from 6 September 2012 to 16 August 2013. A total of 338 flocked rectal swab specimens were collected, of which 280 (83%) also had a matched bulk stool specimen collected. The parents or guardians of 279 of the 280 subjects answered the questionnaire regarding the acceptability of rectal swab sampling as follows: 266 (95%) responded “acceptable,” 8 responded “slightly acceptable,” 3 responded “neutral,” none responded “slightly unacceptable,” and 2 responded “unacceptable.” Only one sample showed frank inhibition of MS2 (bulk stool sample), but no samples showed inhibition of A. tumefaciens DNA detection. One child did not have a rectal swab collected due to an imperforate anus, and one child's guardian refused the rectal swab collection. The median time from swab collection to bulk stool collection was 5 min longer than that for a swab sample (interquartile range, 0.03 to 2.7 h). A comparison of the GPP and LDT assay results on 164 stool samples revealed a relatively close concordance between the assays for 3 bacterial and 2 viral targets (see Table 3). Adenovirus was not compared, given that GPP targets serotypes 40/41, while the LDT targets all adenoviruses. The matched swab-stool pair GPP testing results are shown in Table 4. The Y. enterocolitica and V. cholerae targets were not detected in any sample. Assessing the total of all the pathogen targets, the flocked swab samples detected a total of 410 targets, and the bulk stool samples yielded 390 pathogen targets (P = 0.113). Among the 280 flocked swab samples tested with the GPP assay, 110 had a single pathogen detected, 73, 33, 10, and 3 had two, three, four, and five pathogens, respectively, and 51 had no pathogens detected. Among the 280 stool samples, 113 had a single pathogen detected, 63, 35, 9, and 2 had two, three, four, and five pathogens, respectively, and 58 stool samples had no pathogens detected. There was an average of 1.46 pathogens detected per patient for the swab samples and 1.39 pathogens per patient with stool sample testing. The swab samples detected 12% more bacterial targets than did matched stool sample testing (241 versus 212, respectively; P = 0.003). There was no significant difference in the detection of viral pathogens (110 versus 113; P = 0.701) or protozoal pathogens (59 versus 65; P = 0.345). Focusing on the pathogens for which antimicrobial treatment is generally recommended in the context of severe gastroenteritis requiring admission to the hospital (Shigella, Campylobacter, enterotoxigenic E. coli [ETEC], Cryptosporidium, Giardia, and E. histolytica), the flocked swab samples identified 226 pathogens, and the stool samples identified 203 pathogens (P = 0.009, McNemar's test). The testing results from the laboratory-developed multiplex PCR assays on 164 matched flocked swab-stool pairs are shown in Table 5. There was a total of 189 pathogen targets detected in the flocked swab samples and 167 detected in the matched stool specimens. The C_T values for the matched concordant positive swab and stool samples are shown in Table 6. The C_T values were similar for Shigella, Salmonella, and adenovirus, and they were lower in the swab samples concordant for Campylobacter and higher for swab samples concordant for norovirus.

TABLE 3.

Comparison of performances of gastrointestinal pathogen panel assay and laboratory-developed test on shared bacterial and viral targets using 164 bulk stool samples

Pathogen target	Result (no.) for indicated testa				Concordance (%)	McNemar P
	Both positive	GPP positive only	LDT positive only	Both negative
Shigella	42	0	2	120	98.8	0.50
Campylobacter	23	1	7	133	95.1	0.07
Salmonella	12	1	3	148	97.6	0.62
Rotavirus	13	0	0	151	100	1.00
Norovirus GI/GII	28	0	4	132	97.6	0.12

^aGPP, gastrointestinal pathogen panel; LDT, laboratory-developed test.

TABLE 4.

Comparison of pathogen target detection in 280 matched bulk stool and flocked swab sample pairs using xTAG GPP assay

Pathogen target	Result (no.) for indicated test					Swab sensitivity (%)	Stool sensitivity (%)	McNemar P
	Either sample positivea	Both samples positive	Swab positive only	Stool positive only	Both samples negative
Bacteria
Shigella	82	63	16	3	198	96.3	80.5	<0.01
Campylobacter	56	43	11	2	224	96.6	80.4	0.02
Salmonella	52	29	13	10	228	80.8	75	0.68
ETEC LT/ST	36	25	9	2	244	94.4	74.3	0.06
E. coli O157	10	7	2	3	268	75.0	83.3	1.00
STEC stx1/stx2	11	6	2	3	269	72.7	81.8	1.00
C. difficile toxin A/B	23	8	7	8	257	65.2	69.6	1.00
All bacteria combined	272	181	60	31	1,688	88.6	77.9	<0.01
Viruses
Norovirus GI/GII	58	41	5	12	222	79.3	91.4	0.14
Rotavirus A	32	29	2	1	248	96.9	93.8	1.00
Adenovirus 40/41	35	28	5	2	245	94.3	85.7	0.45
All viral combined	125	98	12	15	715	88.0	90.4	0.70
Protozoa
Cryptosporidium	47	33	6	8	233	83.0	87.2	0.79
Giardia	24	15	0	9	256	62.5	100	<0.01
E. histolytica	5	0	5	0	275	100	0	0.06
All protozoa combined	76	48	11	17	764	77.6	85.5	0.34

^aEither sample positive is set as the reference for the calculation of sensitivity.

TABLE 5.

Pathogen target detection in 164 matched bulk stool and flocked swab sample pairs using laboratory-developed multiplex assays

Pathogen target	Result for indicated tests					Swab sensitivity (%)	Stool sensitivity (%)	McNemar P
	Either sample positivea	Both samples positive	Swab positive only	Stool positive only	Both samples negative
Bacteria
Shigella	54	42	12	0	110	100	77.8	<0.01
Campylobacter	43	27	13	3	121	93.0	69.8	0.02
Salmonella	29	11	13	5	135	82.8	55.2	0.10
All bacterial combined	126	80	38	8	366	93.6	69.8	<0.01
Viruses
Norovirus GI/GII	32	25	0	7	132	78.1	100	0.02
Rotavirus A	13	12	0	1	151	92.3	100	1.0
Adenovirus	36	32	2	2	128	94.4	94.4	1.0
All viral combined	81	69	2	10	411	87.7	97.5	0.04

^aEither sample positive is set as the reference for calculation of sensitivity.

TABLE 6.

Mean threshold cycle values for matched positive samples and C_T value differences by laboratory-developed real-time PCR^a

Pathogen target (n)	CT for:		Difference in CT values (95% CI)b	P value by paired t test
	Rectal swab	Stool sample
Shigella (42)	25.96	26.34	−0.39 (−1.69 to 0.91)	0.55
Salmonella (11)	33.69	34.88	−1.19 (−3.5 to 1.14)	0.28
Campylobacter (27)	27.98	30.33	−2.35 (−4.05 to −0.064)	<0.01
Rotavirus (12)	26.86	25.60	1.27 (−1.33 to 3.86)	0.31
Adenovirus (32)	28.01	27.42	0.60 (−0.98 to 2.17)	0.20
Norovirus GI/GII (25)	27.82	23.79	4.03 (2.18 to 5.89)	<0.01

^aC_T, threshold cycle.

^bCI, confidence interval.

DISCUSSION

We found that the samples collected using specifically designed flocked rectal swabs from children admitted with severe acute gastroenteritis in Botswana allowed for significantly higher bacterial pathogen detection when using multiplex PCR assays than that for the same testing on the matched bulk stool samples. Our population had a high prevalence of pathogens detected via molecular multiplex assays, which is in keeping with other studies done on pediatric gastroenteritis in developing countries (17).

As outlined in Materials and Methods, during the study, we changed our extraction method from the QIAsymphony platform to the easyMAG platform, as the easyMAG platform had become available at the laboratory in Botswana, and we planned to transition testing on site using this method. Thirty-seven paired bulk stool-swab samples were processed using both extraction platforms, and there was no clear difference in the results (data not shown). easyMAG extraction using the Bertin ceramic beads did detect three additional E. histolytica-positive flocked swab samples that were not detected using glass bead lysis. As shown in Table 3, we also found 5 additional E. histolytica positives, all detected in flocked swab samples and all in samples processed using the easyMAG platform using ceramic bead lysis with easyMAG extraction. Although not proven, we surmise that the additional positives were a result of better cyst lysis with ceramic beads, but this requires further study. However, this potential difference in E. histolytica detection by preanalytical method is unlikely to affect the results of the swab and bulk stool comparison, as all matching samples were processed identically, and neither viral nor bacterial target amplification was affected by extraction methods.

A similar study carried out with children presenting with diarrhea in Rwanda compared regular flocked swabs to bulk stool PCR and found similar yields for the qualitative detection of multiple bacterial, viral, and parasitic pathogens (18). Our swab showed similar detection rate for most pathogen targets but actually had a higher yield for several bacterial targets. There are several plausible explanations for this finding. The rectal FLOQSwabs were specially designed such that they would sample just beyond the anal canal at the columnar epithelium. The swab has a lengthened flocked surface and a stopper at the 3.2-cm mark, which extends just proximal to the surgical anus of most children <3 years of age (19). Many of the bacterial and some of the protozoal pathogens of diagnostic interest reside in this anatomic location, and therefore, we hypothesized that this would be the ideal location to sample. Conversely, bulk stool samples contain more contents derived largely from the small intestine, which may in fact dilute the cellular material of interest contained in the colonic mucosal surface. Given that we used molecular diagnostics, another potential explanation is that bulk stool samples contain more inhibitory material than do flocked swab samples. We included MS2 phage and A. tumefaciens as internal RNA and DNA controls, respectively, and found frank RNA inhibition with only one bulk stool sample. However, these controls may not rule out relative inhibition, which may have affected bulk stool to a greater extent than rectal swabs.

Our group and others have found relatively high rates of mortality in children in sub-Saharan Africa presenting with moderate to severe acute gastroenteritis (20, 21). Many of these children who die from gastroenteritis are found to have treatable enteropathogens detected in their stool (20, 22). Given that in our study, we were not able to collect a matching bulk stool sample from 17% of the enrolled children prior to discharge or death, and that for an additional 25% of children, it took ≥2.7 h to collect a matching stool sample, our data suggest that point-of-care diagnostics using bulk stool samples would be a challenge for a large proportion of children even in the in patient setting. The combination of rapid sample acquisition with sensitive rapid detection methods, such as PCR, may allow for targeted treatment and the potential for significantly improved outcomes for this common and, in many places, deadly infection.