The ¹³C-urea breath test (¹³C-UBT) is a noninvasive, simple, and widely available test for the initial diagnosis of Helicobacter pylori infection and confirmation of H. pylori eradication after treatment.^1,2 This test is considered an ideal test for those in whom endoscopy is not required because it offers the combination of simplicity, accuracy, absence of exposure to radioactivity, and reliability. However, the precise cutoff point to define whether UBT is positive or negative remains controversial.³ One approach to solve this problem is to consider a range of ¹³C-UBT values (2.5‰ to 5.0‰) as a “gray zone” for which results should be considered inconclusive. Calvet et al.^4,5 and our team have reported that most false-positive results are between 2.5‰ and 12‰, suggesting a larger “gray area” of delta values. Several studies have set up the cutoff point between 1.3‰ and 7.4‰, showing high sensitivity and specificity of ¹³C-UBT after H. pylori eradication.^1,6 The selected cutoff value of ¹³C-UBT could depend on several factors, including the dose of urea administered, the indication of ¹³C-UBT, the measuring equipment for detecting δ ¹³CO₂, the presence and composition of a meal, and the time from the ingestion of urease to the expired gas sample of ¹³C-UBT.^2,7,8 Thus, a unique and generally proposed cutoff level may not be possible. Our previous retrospective study has reported lower specificity (67.2%) and despite high sensitivity (96.4%) of ¹³C-UBT for the range of 2.5‰ to 3.0‰ after H. pylori eradication.⁴

To decrease the rate of false positivity, the use of a citric acid meal has been proposed to increase the ¹³C-UBT value in H. pylori-infected patients and reduce this value in uninfected patients.^3,9,10 Citric acid can increase urea hydrolysis by H. pylori¹¹ and decreases intragastric pH, which tends to inhibit non-H. pylori ureases. It may also retard gastric emptying, thus making gastric food contents remain longer in the stomach.³ In the United States, coadministration of citric acid and urea is used to increase the diagnostic accuracy of UBT. Both the United States and European Standard Protocols include citric acid within the test kit.¹² In contrast, the use of a citric acid meal is not a standard part of the Asian Standard Protocol. In addition, severe atrophic gastritis (AG) or intestinal metaplasia (IM) was the cause of false positivity in this test.^4,13 Generally, the prevalences of endoscopic AG and IM were 40.7% and 12.5% in a previous Korean nationwide multicenter study.¹⁴

Based on this background, we hypothesized that a citric acid meal could increase the accuracy of ¹³C-UBT. The aim of this study was to determine the effect of citric acid meal on the diagnostic validity of ¹³C-UBT after H. pylori eradication and compare such ¹³C-UBT with the use of citric acid and endoscopic biopsy-based methods for detecting H. pylori.

1. Study population

Patients with proven H. pylori infection (i.e., positive results for two of the following three endoscopic biopsy methods: histology [the modified Giemsa test], CLO test, and rapid urease test) were recruited prospectively between January 2015 and April 2018 at Seoul National University Bundang Hospital. They were classified into two groups depending on the test day regardless of H. pylori eradication regimen or other clinical setting. The testing day was chosen blindly by subjects. On Tuesdays, subjects received 4 g of citric acid dissolved in 200 mL of water immediately before receiving a ¹³C-UBT tablet (citric acid group). On Mondays, subjects received ¹³C-UBT without a citric acid meal (control group). Exclusion criteria were (1) after gastric operation; (2) administration of antibiotics or consumption of bismuth salts within 4 weeks or administration of a proton pump inhibitor (PPI) within 2 weeks prior to ¹³C-UBT; (3) H. pylori eradication failure because of poor compliance; and (4) H. pylori reinfection where H. pylori status became positive for more than 1 year after successful eradication. The study protocol was approved by the Ethics Committee of Seoul National University Bundang Hospital (SNUBH B-1412/279-004). All participants gave written informed consent. This trial was registered with the UMIN Clinical Trials Registry (number: UMIN000001169).

2. H. pylori eradication

For treatment of H. pylori infection, standard Korean government-approved therapies were used, including PPI-based triple therapy (standard dose of PPI b.i.d. [twice a day], clarithromycin 500 mg b.i.d., and amoxicillin 1 g b.i.d. for 1 week) and sequential therapy (initial 5-day therapy with a combination of PPI b.i.d. and amoxicillin 1 g b.i.d., followed by 5 days of PPI b.i.d., clarithromycin 500 mg b.i.d., and metronidazole 500 mg t.i.d. [three times a day]) as first-line therapies in all study subjects.¹⁵ When these first-line therapies failed, two types of rescue therapies were used, namely, bismuth-containing quadruple therapy (PPI b.i.d., tripotassium dicitrate bismuthate 300 mg q.i.d. [three tablets 30 minutes before meals and one tablet 2 hours after dinner], metronidazole 500 mg t.i.d., and tetracycline 500 mg q.i.d.) for 1 to 2 weeks, or moxifloxacin-containing triple therapy (moxifloxacin 400 mg q.i.d., amoxicillin 1 g b.i.d., and PPI b.i.d.) for 1 to 2 weeks. When second-line therapy failed, other rescue therapies were used.

3. Administration of a citric acid meal and ¹³C-UBT

Before ¹³C-UBT, patients were instructed to stop taking medications such as bismuth salts or antibiotics for 4 weeks and PPI for 2 weeks. They were asked to fast for a minimum of 4 hours. Patients were assigned to receive the test meal (citric acid group) for ¹³C-UBT on Tuesdays or to the control group on Mondays. After washing the oral cavity by gargling, participants in the citric acid group received the citric acid solution (Dongwon, Seoul, Korea; 4 g in 200 mL of water containing 50 g [200 kcal] glucose polymer with artificial sweetener). After consumption of the meal, a predose breath sample was obtained, and then 100 mg tablet of ¹³C-urea (UBiTkit^™; Otsuka Pharmaceutical Co. Ltd., Tokyo, Japan) was administered. In the control group, ¹³C-UBT was performed with prior consumption of a glucose meal without citric acid. Breath samples were collected in the sitting position using special breath collection bags before ¹³C-urea administration (baseline) and 20 minutes after administration. Collected breath samples were analyzed using an isotope-selective, nondispersive infrared spectrometer (UbiT-IR 300^®; Otsuka Pharmaceutical Co. Ltd). Despite the lack of local validation, the ¹³C-UBT cutoff value of 2.5‰ was used as recommended by the manufacturer and a delta ¹³CO₂ of ≥2.5‰ was considered positive.

4. Endoscopic surveillance for the detection of H. pylori infection

Two biopsy specimens were obtained from the antrum and gastric body for histology. The presence of H. pylori was assessed by modified Giemsa staining. Degrees of inflammatory cell infiltration, atrophy, and IM were assessed by hematoxylin and eosin staining. Histological features of gastric mucosae were graded using the updated Sydney scoring system with a 4-point scale (i.e., 0=none, 1=slight, 2=moderate, and 3=marked).¹⁶ Another two biopsy specimens (from the lesser curvature of the antrum and gastric body) were used for the rapid urease test (CLOtest, Delta West, Bentley, Australia). Antral and gastric body biopsy specimens were evaluated separately. All urease tests were monitored for color change for up to 24 hours. Further analysis was performed regarding the IM condition.

5. Statistical analysis

Sensitivity, specificity, and likelihood ratios for a positive test result (LRp) for a negative test result (LRn) of ¹³C-UBT were calculated for the citric acid and control groups. Statistical analysis was conducted using PASW Statistics version 18.0 (SPSS, Chicago, IL, USA). Positive predictive value (PPV) and negative predictive value (NPV) were included in LRs. Student t-test, Pearson chi-square test, and Fisher exact test were used, as appropriate, for univariate analysis of factors affecting the accuracy of ¹³C-UBT. A logistic regression model was used for multivariate analysis. Statistical significance was considered at p<0.05.

1. Comparison of ¹³C-UBT values

Fig. 1 shows the current study flow. Of 1,207 consecutive participants who underwent ¹³C-UBT after H. pylori eradication therapy, 562 participants (46.6%) visited on Tuesdays for ¹³C-UBT and received citric acid before ¹³C-UBT (meal group), while 645 participants (53.4%) visited on Mondays without receiving citric acid (control group). Baseline characteristics of these participants are summarized in Table 1. Overall, 116 participants (20.6%) in the meal group and 139 participants (21.6%) in the control group showed positive results of ¹³C-UBT after H. pylori eradication therapy. The mean δ of ¹³C-UBT value showed a significant difference between the two groups. It was significantly (p<0.001) higher (10.3‰±26.4‰) in the meal group than in the control group (5.1‰±12.6‰) (Fig. 2). The mean δ of the ¹³C-UBT value of positive ¹³C-UBT results was also significantly (p<0.001) higher in the citric acid group (48.3‰±39.7‰ in the meal group vs 22.1‰±19.2‰ in the control group). The mean δ of the ¹³C-UBT value of negative ¹³C-UBT results did not show a significant difference (0.4‰±0.8‰ in meal group vs 0.5‰±0.4‰ in control group, p=0.513). Invasive testing via gastroscopy was recommended when the mean δ of the ¹³C-UBT value was between 2.5‰ and 10‰. Endoscopy was limited to this group, as it was both invasive and expensive. Finally, 60 participants (10.7%) in the meal group and 62 participants (9.6%) in the control group accepted endoscopic biopsy to compare the existence of H. pylori with results of ¹³C-UBT.

2. Effect of citric acid meal on diagnostic accuracy of ¹³C-UBT

A total of 114 participants were evaluated by endoscopic biopsy for histological analysis (modified Giemsa test) of the antrum and gastric body, while 98 participants were evaluated by CLOtest of the antrum and gastric body for determination of H. pylori after ¹³C-UBT. A total of 92 participants were evaluated by both histology and CLOtest. The mean δ of the ¹³C-UBT value showed a significant difference between the two groups (10.9‰±25.6‰ in meal group vs 3.5‰±7.9‰ in the control group, p=0.032). Twenty participants (33.3%) in the meal group and 16 participants (25.8%) in the control group showed positive results of ¹³C-UBT after H. pylori eradication (p=0.429) (Fig. 3). When the diagnostic accuracy of ¹³C-UBT in the meal group was calculated based on endoscopic biopsy results (histology or rapid urease test), its accuracy, sensitivity, specificity, PPV, NPV, LRp, and LRn were 83.3% (95% confidence interval [CI], 71.5% to 91.7%), 91.7% (95% CI, 61.5% to 99.8%), 81.3% (95% CI, 67.4% to 91.1%), 55.0% (95% CI, 39.8% to 69.3%), 97.5% (95% CI, 85.6% to 99.6%), 4.89 (95% CI, 2.65 to 9.03), and 0.10 (95% CI, 0.02 to 0.67), respectively (Table 2). When the diagnostic accuracy of ¹³C-UBT for the control group was calculated based on endoscopic biopsy results, its accuracy, sensitivity, specificity, PPV, NPV, LRp, and LRn were 88.7% (95% CI, 78.1% to 95.3%), 90.9% (95% CI, 58.7% to 99.8%), 88.2% (95% CI, 76.1% to 95.6%), 62.5% (95% CI, 43.5% to 78.3%), 97.8% (95% CI, 87.4% to 99.7%), 7.73 (95% CI, 3.56 to 16.07), and 0.10 (95% CI, 0.02 to 0.67), respectively (Table 2).

3. Diagnostic accuracy of ¹³C-UBT in the gray area based on invasive tests

Of 1,207 participants, 36 (3.0%) in the meal group and 48 (3.9%) in the control group had results in the gray area (2.0‰ to 10.0‰) for ¹³C-UBT (Fig. 1). Fifteen participants in the meal group and 11 participants in the control group were analyzed by endoscopic biopsy methods (histology or rapid urease test). The diagnostic accuracy, sensitivity, specificity, PPV, NPV, LRp, and LRn of ¹³C-UBT for the meal group in the “gray area” were 53.3% (95% CI, 26.6% to 78.7%), 75.0% (95% CI, 19.4% to 99.4%), 45.5% (95% CI, 16.8% to 76.6%), 33.3% (95% CI, 18.6% to 52.2%), 83.3% (95% CI, 44.8% to 96.9%), 1.38 (95% CI, 0.63 to 3.00), and 0.55 (95% CI, 0.09 to 3.88), respectively. They were 45.5% (95% CI, 16.8% to 76.6%), 100.0% (95% CI, 29.2% to 100.0%), 25.0% (95% CI, 3.2% to 65.1%), 33.3% (95% CI, 25.1% to 42.7%), 100.0%, 1.33 (95% CI, 0.89 to 1.99), and 0.00, respectively, for the control group in the “gray area” (Table 3).

4. Risk factors for a discrepancy between ¹³C-UBT and invasive test after H. pylori eradication

Table 4 shows risk factors that caused mismatched results between ¹³C-UBT and endoscopic biopsy-based tests. Except sex (p=0.039), there was no significant risk factor based on univariate analysis.

5. Diagnostic accuracy depending on the degree of gastric atrophy

To exclude the influence of gastric mucosal atrophy on the diagnostic accuracy of ¹³C-UBT, subgroup analysis was performed for the diagnostic validity using 69 patients who were revealed to have no or mild gastric atrophy based on gastric mucosal biopsy (31 subjects in the meal group and 38 subjects in the control group). The diagnostic accuracy was 86.8% (95% CI, 71.9% to 95.6%) for the control group and 80.7% (95% CI, 62.5% to 92.3%) for the meal group (p=0.637). Diagnostic sensitivity, specificity, PPV, NPV, LRp, and LRn were 100.0% (95% CI, 47.8% to 100.0%), 76.9% (95% CI, 56.4% to 91.0%), 45.5% (95% CI, 29.2% to 62.7%), 100.0%, 4.33 (95% CI, 2.15 to 8.74) and 0.0 for the meal group and 100.0% (95% CI, 47.8% to 100.0%), 84.9% (95% CI, 68.1% to 94.9%), 50.0% (95% CI, 30.8% to 69.2%), 100.0%, 6.6 (95% CI, 2.94 to 14.80), and 0.00 for the control group, respectively (Supplementary Table 1). Among those in the gray zone, the diagnostic accuracy was 55.6% (95% CI, 21.2% to 86.3%) for the meal group and 37.5% (95% CI, 8.5% to 75.5%) for the control group (p=0.620). Supplementary Table 2 shows diagnostic validity for the gray area except severe to moderate gastric mucosal atrophy.

6. Diagnostic accuracy depending on gastric intestinal metaplasia

To exclude any influence of gastric mucosal IM on the diagnostic accuracy of ¹³C-UBT, subgroup analysis was performed for the diagnostic validity using 86 patients who were revealed to have no or mild IM based on gastric mucosal biopsy (48 subjects in the meal group and 38 subjects in the control group). The diagnostic accuracy was 86.8% (95% CI, 71.9% to 95.6%) for the control group and 85.4% (95% CI, 72.2% to 93.9%) for the meal group (p=1.000). Diagnostic sensitivity, specificity, PPV, NPV, LRp, and LRn were 90.0% (95% CI, 55.5% to 99.8%), 84.2% (95% CI, 68.8% to 93.9%), 60.0% (95% CI, 41.2% to 76.3%), 97.0% (95% CI, 83.2% to 99.5%), 5.70 (95% CI, 2.66 to 12.22), and 0.12 (95% CI, 0.02 to 0.77) for the meal group and 80.0% (95% CI, 28.4% to 99.5%), 87.9% (95% CI, 71.8% to 96.6%), 50.0% (95% CI, 26.5% to 73.5%), 97.7% (95% CI, 83.3% to 99.4%), 6.60 (95% CI, 2.38 to 18.26), and 0.23 (95% CI, 0.04 to 1.32) for the control group, respectively (Supplementary Table 3). Among those in the gray zone, the diagnostic accuracy was 58.3% (95% CI, 27.6% to 84.8%) for the meal group and 33.3% (95% CI, 4.3% to 77.7%) for the control group (p=0.620). Supplementary Table 4 shows diagnostic validity for the gray area except severe to moderate IM.

The use of 4 g of citric acid as a test meal failed to increase the diagnostic validity of ¹³C-UBT after H. pylori eradication therapy. Although the mean ¹³C-UBT value of the citric acid group (10.3‰±26.4‰) was significantly higher than that of the control group (5.1‰±12.6‰, p<0.001), the sensitivity was only slightly higher in the meal group (91.7% vs 90.9%) while the specificity was higher in the control group (88.2% vs 81.3%). The overall accuracy was also higher in the control group (87.7% vs 83.3%), although the sensitivity of ¹³C-UBT for the meal group was higher than for the control group (90.0% vs 80.0%) in subgroup analysis for those without or with mild gastric atrophy or IM. Gastric atrophy and IM might have blunted the effect of citric acid on the accuracy of ¹³C-UBT.

Generally, moderate to severe gastric atrophy and IM were associated with increased risk of a mismatched result of ¹³C-UBT value after H. pylori eradication, suggesting that the presence of hypochlorhydria or achlorhydria and resulting changes in gastric microbiome might affect the result.^4,13,17 A previous Japanese study demonstrated that five bacterial species with urease activity (Proteus mirabilis, Citrobacter freundii, Klebsiella pneumoniae, Enterobacter cloacae, and Staphylococcus aureus) were subsequently isolated from the oral cavity and/or stomach, and all patients with a false positivity result were suffering from AG.¹⁸ Furthermore, there are marked geographic differences in the frequency of non-H. pylori contamination in duodenal ulcer disease (high in Colombia and low in Korea or United States).¹⁹

Taken together, these results suggest that in regions of the world where AG is common, the diagnostic accuracy of the test needs to be increased, such as by using citric acid meals and different doses or formulations of urea. The original ¹³C-UBT employed a test meal designed to slow gastric emptying and maximize the distribution of the substrate within the stomach to increase the area and time of contact between bacteria and the substrate.²⁰ Different doses (1, 2, or 4 g) of citric acid can produce significant increases in breath ¹³CO₂ activity compared to other meals, such as ascorbic acid, subcutaneous pentagastrin, and glucose polymer.⁹ Citric acid could slow gastric emptying and enhance the intragastric distribution of urea.³ Most studies evaluating the role of citric acid in UBT have shown higher delta values with citric acid in comparison with other pretest meals or no test meals.^12,21,22 Similarly, the standard deviation of δ ¹³C-UBT value in the citric acid group of the present study was significantly higher than that of the control group, while the mean δ ¹³C-UBT value for positive ¹³C-UBT results in the citric acid group was significantly higher. Furthermore, the sensitivity was higher in the citric acid meal group. These results reflect the increase in urease activity by citric acid (i.e., increase of δ ¹³C-UBT value in H. pylori infected patients without a consistent change in the delta value in uninfected patients). Thus, the use of citric acid as a meal theoretically could increase the diagnostic accuracy of ¹³C-UBT and could especially increase the discriminative capacity in the gray zone. However, prior fasting and test meals may not be essential.^23–26 Ng et al.²⁷ reported that sensitivities and specificities of ¹³C-UBT were 97.4% and 95.8%, respectively, in the group with fasting and use of citric acid as a test meal. They were 96.5% and 93.9%, respectively, in the group without fasting and absence of a test meal in a Chinese population. Gisbert et al.²² reported that using citric acid as a test meal might result in higher ¹³C-UBT results at different sampling times (15, 30, and 45 minutes) in H. pylori-positive patients, suggesting that these results did not imply a better discrimination between infected and noninfected patients.

In our study, the use of citric acid as a test meal improved the diagnostic sensitivity after H. pylori eradication therapy in comparison with that of the control group (Table 2). However, there was no benefit for the diagnostic validity in the gray zone (Table 3). Although false-negative tests were rare, false positivity was common, and sensitivity was low even with the use of citric acid. To determine whether such disappointing results could be due to severe gastric mucosal atrophy or IM, we further analyzed those with or without mild atrophy or IM. However, the increase did not reach significant difference (p=0.637 in atrophy and p=1.000 in IM subgroup analysis). Although the sensitivity was equal or higher in the meal group (100.0% both group in atrophy and 90.0% vs 80.0% in IM subgroup analysis), the specificity was higher in the control group (84.9% vs 76.9% in atrophy and 87.9% vs 84.2% in IM group subgroup analysis). In the meal group, the sensitivity was rather low (75.0%) for the gray zone in IM subgroup analysis. There was no significant difference between the two groups, possibly due to the small number of subjects. However, this study was conducted over three years. The accuracy of UBT should apply to all post-eradication subjects regardless of atrophy or IM.

To obtain a statistically significant improvement of diagnostic accuracy from 90% to 95% would likely require an extremely large sample size. However, previous studies could not meet the needed sample size, either, due to limitations in the clinical situation, such as high cost of endoscopy and the invasiveness of endoscopic biopsy-based H. pylori tests.^4,5,22,23 Furthermore, the National Cancer Screening Program in Korea offers either ¹³C-UBT or the rapid urease test for diagnosis after H. pylori eradication with the exception of the endoscopic ¹³C-urea test. These two tests cannot be performed together after H. pylori eradication. Thus, we could only compare ¹³C-UBT with biopsy-based H. pylori tests in 138 subjects (9.9%) after H. pylori eradication therapy over a 3-year period. We were unable to prove whether endoscopic biopsy results were more accurate than ¹³C-UBT results.

Our results are consistent with our previous suggestion that moderate to severe IM is an independent risk factor for a false positivity.^4,28 As the degree of gastric atrophy becomes severe, the environment in the stomach changes to a hypochlorhydric state and causes overgrowth of non-H. pylori urease-positive bacteria.²⁹ CLOtest also has a low detection rate for H. pylori in the presence of mucosal atrophy. IM as this mucosa is not conducive to the growth or attachment of H. pylori. This becomes more prominent in the presence of higher levels of IM and AG.³⁰ A recent Japanese study suggested a possibility that patients with autoimmune gastritis were sometimes misdiagnosed as refractory to eradication therapy for H. pylori because of the presence of urease-positive bacteria other than H. pylori that colonized the stomach.¹³ In addition, we did not give citric acid and urea together, as in Western countries. Confirmatory tests did not follow right away in some cases due to the invasiveness of endoscopy. A previous study reported that, in case of “false-positive ¹³C-UBT results,” re-endoscopic biopsy-based methods showed that all of them had positive histology when multiple antral biopsy specimens were taken.³⁰

In conclusion, our results show that citric acid did not increase the diagnostic accuracy or specificity of ¹³C-UBT after H. pylori eradication therapy. Thus, those with multiple treatment failures as confirmed by ¹³C-UBT should not automatically be considered infected. The diagnosis should be confirmed by another test, such as histology or stool antigen, before starting retreatment.

This work was supported by a grant (number: 2011-0030001) of the National Research Foundation (NRF) for the Global Core Research Center (GCRC) funded by the Ministry of Science, ICT and Future Planning (MSIP), Republic of Korea.

The authors greatly appreciate Prof. David Y. Graham for suggesting this research topic and his endless interest and advice.

No potential conflict of interest relevant to this article was reported.

Guarantor of the article: N.K. Study design, data analysis, statistical analysis, data interpretation and manuscript drafting: Y.H.K. Study design, enrolled the subjects, data interpretation, and critical revision: N.K. Enrolled subjects and edited the manuscript: H.Y., C.M.S., Y.S.P., D.H.L. All authors approved the final version of the manuscript.

See editorial on page 479.