Helicobacter ISSN 1523-5378

Diagnosis of Helicobacter pylori Infection Cliodna A. M. McNulty,* Philippe Lehours†‡ and Francis Me´graud†‡ *

Consultant Medical Microbiologist and Head, Health Protection Agency Primary Care Unit, Microbiology Department, Gloucestershire Royal Hospital, Gloucester GL1 3NN, UK, †INSERM U853, 33076 Bordeaux, France, ‡Universite´ de Bordeaux, Laboratoire de Bacte´riologie, 33076 Bordeaux Cedex, France

Keywords Culture, rapid urease test, histology, PCR, serology, urea breath test, stool antigen test. Reprint requests to: Professor Francis Me´graud, Laboratoire de Bacteriologie, C.H.U. Pellegrin, Place Amelie Raba-Leon, 33076 Bordeaux Cedex, France. E-mail: [email protected]

Abstract When an endoscopy is performed, it now becomes easier to observe indirect evidence of the presence of a Helicobacter pylori infection, given the progress of new methods including magnifying narrow band imaging or confocal laser endomicroscopy. Out of the biopsy-based tests, the novel original method proposed concerned culture in a broth medium with or without antibiotics and ELISA detection of H. pylori. New stool antigen tests are still appearing with no major improvement in comparison with the monoclonalbased tests already on the market. The combination of pepsinogen detection to H. pylori serology is now more and more evaluated to detect preneoplastic lesions.

While few new methods have been proposed for Helicobacter pylori diagnosis, there are still a number of articles evaluating the current methods and trying to improve their accuracy.

Invasive tests Endoscopy Attempts to diagnose Helicobacter pylori infection directly during endoscopy have been made in the past. While the observation of H. pylori per se is usually not possible, indirect evidence of its presence can be found. This year, using standard endoscopy in children, Hidaka et al.[1] were able to show that the absence of regular arrangement of collecting venules at two sites, indicated the absence of H. pylori infection with an excellent sensitivity (100%) and specificity (90%). The respective values for antral nodularity, the usual criterion, were 84% and 100%. Three studies evaluated narrow band imaging (NBI), a technique enhancing the mucosal and capillary patterns of the gastric surface. Three hundred patients were explored with conventional NBI and 5 mucosal patterns were identified, corresponding to different grades of histological gastritis [2]. Magnifying NBI was also used to investigate the changes in gastric mucosal patterns before and 12 weeks after H. pylori eradication. In patients without severe atrophy and intestinal metaplasia (IM), the changes in pit size and density of fine irregular vessels were observed with a perfect sensitivity

10

and specificity. In contrast, no change occurred in patients with severe atrophy and IM, regardless of H. pylori eradication [3]. The authors emphasized that this method predicts histological conditions and pepsinogen levels and may be cost effective in gastric cancer surveillance [4]. Confocal laser endomicroscopy, a magnifying endoscopy technique, allows an in depth analysis of the gastric mucosa. In a first study, Ji et al.[5] identified H. pylori gastritis with a 92% accuracy. The mean kappa value for inter-observer agreement was 0.78. In a second study, the same group focused on the severity of H. pylori-associated gastritis, especially atrophy and IM, with a rather good diagnostic accuracy [6].

Histology The ever well-known OLGA (Operative Link on Gastritis Assessment) staging system was highlighted again under the aspect of histology reporting of gastritis. It is especially valuable as it allows a prediction of the gastric cancer risk [7]. A long-term follow-up (12 years) of 93 patients confirmed that all invasive or intra-epithelial gastric neoplasia were consistently associated with high-risk (III ⁄ IV) OLGA stages [8]. Immunohistochemistry can be used to assess the presence of H. pylori with more certainty. However, its systematic use in the routine of a pathology laboratory does not seem necessary because it is of no value in cases with an absence of pathological abnormality nor reactive gastropathy [6]. However, it allows the

ª 2011 Blackwell Publishing Ltd, Helicobacter 16 (Suppl. 1): 10–18

H. pylori Diagnosis

McNulty et al.

diagnosis of clusters of predominantly coccoidal forms of H. pylori which are difficult to identify by standard staining. This aspect may be frequent in gastric resection specimens, possibly resulting from hypoxia or other stress conditions [9]. Fluorescence in situ hybridization (FISH) is the other method with high specificity. A set of peptide nucleic acid (PNA) probes was developed to detect H. pylori and its resistance to clarithromycin. The optimized PNAFISH proved to be a reliable method when used in clinical specimens [10]. When H. pylori-like organisms are observed on histological preparations after standard staining but not with anti-H. pylori antibody stains, it is a challenge to identify the bacteria present. Pyrosequencing was used in Korea and, surprisingly, identified Campylobacter hyointestinalis, a gastric bacterium from pigs, seldomly cultured from humans. Among the 20 discrepancies observed, H. pylori was confirmed in four cases, while the other cases detected two Helicobacter cinaedi, one Campylobacter upsaliensis, and 12 C. hyointestinalis [11]. In another report, a urease-negative Helicobacter-like organism was identified as Selenomonas sp., but this time by culture [12]. The value of endoscopic surveillance to detect histologically premalignant gastric lesions, especially IM, as a gastric cancer preventive measure has been highlighted. Both targeted and nontargeted biopsies are required [13]. A decision analysis model was constructed to compare a strategy with a yearly endoscopy versus no surveillance. Over a 10-year period, endoscopic surveillance would decrease the number of detected cancers by 58–84% and appear to be cost effective [14].

Rapid Urease Test (RUT) It is important to remember that formalin contamination of biopsy forceps may be a cause of false-negative RUT (and culture) [15]. Vaira et al. compared three RUT in 375 consecutive patients. The reference was positivity both from urea breath test (UBT) and histology. The three RUT under study were the CLO-test (Kimberley-Clark; Ballard Medical Products, Roswell, UT, USA), PyloriTek (Serim Lab, Elkhart, IN, USA) and a new test: UFT300 (ABS Cernusco, sul Naviglio, Italy). The sensitivity of the new test and PyloriTek at 1, 5, and 60 minutes were comparable and significantly better than CLO-test [16]. As observed previously, testing dual specimens (antrum + body) represents a way to enhance the sensitivity of RUT as was shown with ProntoDry and newly developed RUT Helictec UT (Strong Biotech Corp, Taipei, Taiwan)[17].

ª 2011 Blackwell Publishing Ltd, Helicobacter 16 (Suppl. 1): 10–18

In an experimental model of infected pigs, a new RUT containing 5% urea and bromocresol purple as an indicator (instead of 2% urea and phenol red) was sprayed on the gastric mucosa after sacrifice showing an improved detection and localization of the H. pylorilike organisms present [18]. Lee et al. measured the urease activity in gastric antrum and corpus samples of 54 patients. They observed that in the antrum, there was a good correlation between urease activity, H. pylori density, and inflammation. In contrast, this correlation was not found in the body where the mean urease activity was twice as high as in the antrum. This difference may be explained by the high amiE expression in the gastric body compared with the antrum [19]. amiE (HP0294) is an amidase-hydrolyzing short-chain amide. Positivity of the RUT performed over a 20-year period in a pediatric clinic in Greece (530 infected patients, 1060 controls) also correlated with higher bacterial density and severity of gastritis in the antrum but the body was not tested [20]. Finally, the pros and cons of using RUT were reviewed by Zullo et al. [21].

Culture Culture remains a reference method, but its limited sensitivity leaves room for improvement. Patient factors which could affect culture results were studied in Poland. High gastritis activity, low bacterial load, alcohol drinking, and use of anti-H2 reduced culture efficacy in infected subjects [22]. Miendje Deyi et al. compared two commercial media, Pylori agar (bioMe´rieux, Marcy l’Etoile, France) and BD Helicobacter agar (Becton Dickinson, Franklin Lakes, NJ, USA) to their in-house medium. The three media had the same capacity to grow H. pylori, but the selective property of commercial media was inferior [23]. To speed up the results of culture and susceptibility testing, an original approach was developed by Perna & Vaira. The principle is 1) to culture H. pylori in a broth medium allowing its growth within 20 hours, 2) to include the critical concentration of antibiotics in different tubes, and 3) to detect growth using an ELISA. The new medium tested on 111 H. pylori-positive patients could detect 105, like the standard culture method, and correctly identified clarithromycin and metronidazole susceptibility with two and 10 exceptions, respectively [24]. As shown before, culture of H. pylori from stools is extremely difficult. Kim do et al.[25] used the specific conditions of the colonoscopy preparation to look for viable H. pylori in rectal and ileal fluids. They cultured H. pylori in nine and 11 samples of 20 H. pylori positive

11

H. pylori Diagnosis

patients, respectively, confirming the princeps results of Parsonnet et al.[26]. Numerous studies have tried to identify H. pylori pathovars, but it has not been possible yet to link a specific characteristic of the strain to the disease outcome. The antioxidant protein alkylhydroperoxide reductase (AhpC) from H. pylori was found to correlate with the extent of inflammatory damage in tissues. Huang et al.[27] found AhpC in higher amounts in H. pylori strains isolated from patients with gastric cancer than in patients with gastritis; in addition, high-molecularweight AhpC was more likely to be recognized by antibodies from patients with gastric cancer. Detection of this protein in stools by immunoblotting has also been proposed as a stool antigen test [28]. Other information gathered this year concerns the possibility to maintain viable H. pylori grown in agar stabs for prolonged periods of time (56 days) when a temperature of 37 C with 10% CO2 atmosphere was used whereas the bacteria did not survive at room temperature [29].

Molecular methods Molecular methods have the advantage of their rapidity and the limited influence of the transport conditions. Real-time PCR formats have led to the best results in terms of sensitivity and specificity. Furthermore, they may allow concurrent detection of clarithromycin resistance. Another kit, MutaREAL Helicobacter pylori (Immundiagnostik, Bensheim, Germany), appeared on the market. It was tested after DNA extraction with NucliSens magnetic extraction reagents (bioMe´rieux). Sensitivity and specificity tested on 106 gastric biopsies from children were 93% and 91%, respectively, for H. pylori detection compared with culture. Sensitivity and specificity for clarithromycin resistance were 91% and 96%, respectively, compared with the Etest [30]. It may be interesting to know H. pylori’s viability, especially in environmental samples. A propidium monoazide-based quantitative PCR was developed for this purpose with success [31]. It was again shown that H. pylori cagA and vacA genotypes, determined by PCR on biopsy specimens by reverse hybridization onto a line probe assay, were predictors of progression of preneoplastic lesions in 312 patients endoscoped 20 years apart. Infection with both cagA positive and vacA s1m1 strains was associated with progression of gastric precancerous lesions with an OR of 4.80 (95% CI 1.71–13.5) versus infection with cagA negative ⁄ vacA s2m2 strains [32]. The main limitation in detecting H. pylori DNA in feces is the presence of inhibitors of the Taq polymerase

12

McNulty et al.

used, which have been shown to be complex polysaccharides [33]. Until now, all of the DNA extraction methods proposed have failed to lead to a good sensitivity of the PCR. A new method adapted from extraction of Mycobacterial DNA in clinical samples was proposed, based on a selective hybridization of target DNA with biotin-labeled probes, followed by DNA isolation with streptavidin-coated magnetic beads. It was tested in the model of H. pylori-infected gerbils with fecal samples analyzed 1, 4, and 10 days postinfection. The detection limit obtained was one bacterial cell per 100 mg of stool after heating, i.e. a 10-fold increase in sensitivity compared with a commercially available stool DNA extraction kit [34]. Detection of H. pylori in dental plaque is even more challenging for another reason, i.e. other members of the Epsilonproteobacteriaceae can be present and lead to false positivity. Using two genes versus one as a target, Chaudhry et al. decreased the rate of positivity from 73% to 52% [35]. In another study using PCR and Southern blotting, there was a positive correlation between H. pylori positivity in gastric biopsies and the oral cavity, suggesting the existence of an oral reservoir [36].

Noninvasive tests Urea breath tests There were very few papers in this area this year. Petrovic et al. evaluated a 14C UBT (Nuclear Sciences, Vinca, Serbia) undertaken in fasting Serbian patients 30 minutes after a urease capsule containing a 37 kBq ⁄ dose of 14C. A positive test, defined as a 80% rise in test values compared with the baseline breath pre 14C dose, when compared with histology and biopsy urease test had high sensitivity (94.9%), 100% specificity and thus high positive (100%) and negative (96.3%) predictive values [37]. In another study, using the 13C-UBT, Delta Over Baseline values did not correlate with H. pylori antibiotic resistance [38]. The UBT has for some time been considered the gold standard noninvasive test. A 2009 systematic review by Nocon et al. of 30 studies that directly compared the 13C-UBT to biopsy-based tests as the gold standard confirmed this viewpoint. The 13C-UBT showed higher sensitivity and specificity than the IgG serology and stool antigen tests in the majority of studies [39]. In comparison with the biopsy urease test, results for sensitivity were inconsistent, but the specificity was slightly higher for the 13 C-UBT [39]. There were insufficient results for comparisons between the 13C-UBT and the 14C-UBTs, histology and PCR to determine any significant differences [39].

ª 2011 Blackwell Publishing Ltd, Helicobacter 16 (Suppl. 1): 10–18

H. pylori Diagnosis

McNulty et al.

Stool antigen tests Many of the evaluations of the stool antigen tests (SATs) reported this year from Eastern Europe and beyond in adults, found the SATs to be less accurate than in previous reports. Da Silva et al. evaluated a polyclonal-based cassette HpSA (ACON laboratories Inc, San Diego, CA, USA) in Brazil and found the test to have sensitivity and specificity under 90%; however, they suggested this test may be useful in laboratories with few facilities [40]. Kesli et al. compared histology and rapid urease test to monoclonal SAT enzyme immunoassays (EIAs), Premier Platinum HpSA Plus EIA (Meridian Diagnostics Inc, Cincinatti, OH, USA) and H. pylori Antigen test (Dia.Pro Diagnostic Bioprobes Sri, Milano, Italy) and one immunochromatographic assay (Vegal Farmaceutica, Madrid, Spain) for the diagnosis of H. pylori infection in 168 Turkish adults with dyspepsia before eradication therapy. All had a similar specificity of about 92%, but the Premier Platinum EIA had the highest sensitivity at 90% (Table 1) [41]. Falaknazi et al.[42] determined the value of the HpSA polyclonal Premier Platinum EIA in Iranian patients with chronic renal failure undergoing renal dialysis pre-and postH. pylori eradication treatment. The pre-eradication sensitivity (87%) was lower, and specificity (94%) was higher than in a previous comparative study [43]. In this small series, two of the seven remaining positive by UBT post-treatment gave a false-negative HpSA [42]. Several groups have evaluated the different diagnostic tests in their local populations. Zalabska in a series of 300 Czech patients, and Kalem et al. in a Turkish series of 103, found that the Meridian HpSA EIA (Meridian Diagnostics) detected more positive H. pylori patients than other invasive biopsy-based tests including culture, a rapid urease test, and histology [44,45]. In a small series of 59 Japanese patients postdistal gastrectomy, Yan et al.[46] found the Premier Platinum HpSA test (Meridian Diagnostics) to be more accurate than the UBT with a sensitivity and specificity of 100% and 90.5%. They suggested that the 59.1% specificity obtained with the UBT in this setting may well be due to the altered intragastric environment. Children Kalach et al. evaluated the rapid in-office monoclonal enzyme immunoassay stool antigen test (Rapid HpStAR; Oxoid Ltd.) in 108 children (16 H. pylori positive). The overall sensitivity, specificity, and positive and negative predictive values were higher than in previous studies reported in the review published last year [47] 87.5%, 97.8%, 87.5%, and 97.8%, respectively,

ª 2011 Blackwell Publishing Ltd, Helicobacter 16 (Suppl. 1): 10–18

with an accuracy of 96.2%. Prell et al.[48] also evaluated this Rapid Hp STAR test (Oxoid Ltd.) using culture plus histology and RUT as the gold standard. They found a higher sensitivity but lower specificity in this larger series (Table 1). Raguza et al.[49] found that the Amplified IDEIA Hp STAR (Dako Cytomation Ltd, Hamburg, Germany) had 100% sensitivity but a lower specificity of 76.2% using the manufacturer’s cut-off; therefore, they suggested increasing the cut-off to 0.400 after re-analysis using a receiver operating characteristic (ROC) curve, leading to a specificity of 97.7% in this large series. However, Graham[50] criticized this approach indicating that the new cut-off should be validated in another series of patients, we will await next year’s articles with interest to see how this debate plays out! Tiryaki et al.[51] evaluated the HpSA monoclonal EIA (Diagnostic Bioprobes Sri) in a series of 87 Turkish children, but they found its performance (Table 1) and a blood-based serological test (H. pylori immunoglobulin G, HpIgG; Radim, Pomezia-Rome, Italy) to be poor compared with the UBT pre and posteradication. Post-treatment Calvet et al. evaluated three monoclonal stool tests in 88 patients compared with breath test or histology at least 8 weeks posteradication treatment. The rapid inoffice test RAPID Hp STAR (Oxoid Ltd.) and the laboratory-based Enzyme Immunoassay Amplified IDEIA HpStAR (Oxoid Ltd.) both had 100% sensitivity and therefore, can reliably indicate eradication; however, the ImmunoCard STAT1 HpSA (Meridian Diagnostics) had a lower sensitivity of 90%. All the tests had a specificity of over 92%, but all the tests gave some false positives post-treatment with quite low positive predictive values (62–69%) [47]. Shimoyama et al. sought to evaluate the Premier Platinum HpSA Plus EIA (HpSA ELISA II; Meridian Diagnostics) which uses multiple monoclonal antibodies in Japanese patients postH. pylori eradication [52]. They found this test could detect fewer numbers of H. pylori organisms in spiked fecal specimens than another stool antigen with a single monoclonal antibody (Testmate pylori antigen EIA; Wakamoto Pharmaceutical Co. Ltd., Tokyo, Japan), but interestingly they found that both tests produced ten false negatives and the same negative predictive value. The Testmate produced more false positives (six compared with one), and therefore, the positive predictive value of the Premier Platinum was higher at 97% when it was recalculated (Table 1) [52]. The alkyl hydroperoxide reductase protein AhpC gene was amplified by Pourakbari et al. and used as the basis of a new stool antigen test. This test had similar

13

14 –

ª 2011 Blackwell Publishing Ltd, Helicobacter 16 (Suppl. 1): 10–18 Culture & 2 tests; Post-treatment:UBT

UBT

Culture or histology + RUT

RAPID Hp StAR oxoid in-office, mean 2 observers [48] IDEIA Hp StAR Oxoid EIA [48]

HpSA monoclonal EIA Diagnostic Bioprobes, cut-off 0.1 [51]

Amplified IDEIATM Hp STAR Dako, cut-off 0.4 [49]