A re-testing range is recommended for 13C- and 14C-urea breath tests for Helicobacterpylori infection in China

Background The urea breath test (UBT) is widely used for diagnosing Helicobacter pylori infection. In the Shenzhen Kuichong People’s Hospital, some UBT findings were contradictory to the histology outcomes, therefore this study aimed to assess and compare the diagnostic performance of both 13C- and 14C-UBT assays. Methods We recruited 484 H. pylori-treatment naïve patients, among which 217 and 267 were tested by the 13C-UBT or 14C-UBT, respectively. The cutoff value for H. pylori positivity based on manufacturer’s instruction was 4% delta over baseline (DOB) for the 13C-UBT, and 100 disintegrations per minute (DPM) for the 14C-UBT. Gastric biopsies of the antrum and corpus were obtained during endoscopy for histopathology. Results In patients who were tested using the 13C-UBT kit, histopathology was positive in 136 out of 164 UBT-positive patients (82.9% concordance), and negative in 46 out of 53 UBT-negative cases (86.8% concordance). For the 14C-UBT-tested patients, histopathology was positive for H. pylori in 186 out of 220 UBT-positive patients (84.5% concordance), and negative in 41 out of 47 UBT-negative cases (87.2% concordance). While the 13C-UBT and 14C-UBT each had a high sensitivity level of 95.1% and 96.9%, respectively, their specificity was low, at 62.2% and 54.7%, respectively. By using new optimal cutoff values and including an indeterminate range (3–10.3% DOB for 13C-UBT and 87–237 DPM for 14C-UBT), the specificity values can be improved to 76.7% and 76.9% for the 13C- and 14C-UBT, respectively. Conclusions The establishment of an indeterminate range is recommended to allow for repeated testing to confirm H. pylori infection, and thereby avoiding unnecessary antibiotic treatment. Trial registration: Chinese Clinical Trial Registry, ChiCTR2000041570. Registered 29 December 2020- Retrospectively registered, http://www.chictr.org.cn/edit.aspx?pid=66416&htm=4 Supplementary Information The online version contains supplementary material available at 10.1186/s13099-021-00435-3.


Introduction
Helicobacter pylori infection is common in China, with an overall estimated prevalence of 55.8% [1]. It is an important gastric pathogen that can lead to several gastroduodenal disorders including chronic gastritis, gastric atrophy and peptic ulcer disease, and less commonly, to gastric adenocarcinoma and mucosa associated lymphoid tissue (MALT) lymphoma [2,3].
Helicobacter pylori is able to convert urea into carbon dioxide and ammonia via its urease enzyme, where the ammonia is used to neutralize the acid for its survival in the stomach [4]. Based on this feature of H. pylori, the urea breath test (UBT), a non-invasive H. pylori infection diagnostic method was developed. This requires a patient to swallow a capsule containing a dose of urea labeled with carbon-13 ( 13 C) or carbon-14 ( 14 C) isotope. If the patient is an H. pylori carrier, the labeled urea will be hydrolyzed by the bacterial urease enzyme within the stomach, resulting in the release of labeled carbon dioxide which is then absorbed into the bloodstream and expelled from the lungs in a few minutes. Hence the amount of labeled carbon dioxide within a patient's breath sample can be measured to determine current H. pylori infection status [5,6].
Due to its accuracy, simplicity and non-invasive nature, the UBT has been the preferred method of many medical professionals for testing H. pylori infection in their patients. Both the 13 C-UBT and the 14 C-UBT are widely used. The former utilizes the stable 13 C isotope of carbon while the latter uses the radioactive 14 C carbon isotope. Nevertheless, it is important to mention that both are naturally occurring isotopes and the radiation exposure from the 14 C-UBT is even lower than that from background radiation [7]. In fact, the 14 C-UBT has been approved by the Food and Drug Administration (FDA) of the United States for its usage in everyone, including children and pregnant women [8].
While UBT is useful in detecting H. pylori infection, we noticed that several UBT results were contradictory to the outcomes determined via histopathology examination, prompting us to reconsider the diagnostic accuracy of the commercial UBT kits used for screening H. pylori infection in our hospital. In this study, we recruited 484 individuals who underwent endoscopic examination at Shenzhen Kuichong People's Hospital, among which 217 and 267 were tested for H. pylori infection using the 13 C-UBT and 14 C-UBT, respectively. By comparing the outcomes to that of histopathology examination of gastric biopsies, which is the gold standard for diagnosing H. pylori infection, we assessed the diagnostic performance of both UBT kits. Additionally, as these commercial kits available for use at our hospital provide only a cutoff value for H. pylori positivity, resulting in high positive rates, we therefore sought to introduce an "indeterminate zone". Should a UBT value fall within the indeterminate range, we would like to recommend a repeat UBT or the use of another H. pylori diagnostic method to confirm the presence or absence of an infection.

Overview of entire study
The schematic flow of experimental program was shown in Fig. 1.

Development of research study
During the clinical practice, we noticed that several UBT results were contradictory to the outcomes determined via histopathology examination, prompting us to reassess the diagnostic performance of both 13 C-UBT and 14 C-UBT commercial kits used for screening H. pylori infection in our hospital.

Defining the criteria of study
The following exclusion criteria on our study subjects were applied: (i) previous treatment of H. pylori infection, (ii) received proton pump inhibitor, H 2 receptor antagonist, expectorant, or antibiotic treatment within the last four weeks prior to endoscopy, (iii) history of gastric surgery, (iv) diagnosed with gastrointestinal cancer, (v) had severe heart, lung, liver, kidney or blood system disorder, (vi) aged below 18 or above 70 years old and (vii) pregnancy.

Ethics approval
This study was approved by the research ethics committee of Shenzhen Kuichong People's Hospital (Reference no. 201609) and registered at www. chictr. org. cn (Reference no. ChiCTR2000041570). Each patient was given a Fig. 1 The overview of entire study detailed introduction to the purpose and process of the research by a gastroenterologist. Written and informed consents were obtained from all patients prior to their participation in the study.

Study population
From January 2017 to November 2018, following the application of the exclusion criteria, 484 patients (18-70 years of age) who visited Shenzhen Kuichong People's Hospital (26 Kuixin N Road, Dapeng New District, Shenzhen City, China) for endoscopic check-up agreed to participate in this study.
Prior to the endoscopy session, the patients received either a 13 C-UBT or a 14 C-UBT at the discretion of a gastroenterologist via the alternating assignment method. The general health information of patients was collected by clinical nurses. During endoscopy, two gastric biopsy specimens (one each from antrum and corpus) were collected for histopathology examination. The histology examination was performed by Da'an Clinical Laboratory, a third-party pathology service provider. The UBTs were conducted by clinical technicians who had no knowledge of this study.

Sample size determination
The samples size required for this study was estimated based on a 95% confidence interval using the following formula [9]: Sample size (n1) based on sensitivity = Sample size (n2) based on specificity = where Z, the normal distribution value, was set to 1.96 as corresponding with the 95% confidence interval, and d, the maximum acceptable width of the 95% confidence interval, was set to 10%. Based on a previous study, the UBT was shown to achieve a sensitivity value of 96% and a specificity value of 93% [10]. The prevalence rate (Prev) of H. pylori in Shenzhen, China was 35.85% [11]. Based on the criteria above, this study required at least 41 H. pylori-positive patients (n1) and 39 H. pylori-negative patients (n2), yielding a minimum total sample size of 80 participants for each UBT test.

Urea breath test 13C-UBT
The 13 C-UBT (Beijing Boran Pharmaceutical Co. Ltd., China) was performed according to the manufacturer's instructions. Briefly, an initial baseline breath sample was collected from each patient after fasting for at least four hours prior to ingesting a capsule containing 50 mg 13 C isotope labeled urea with 80-100 mL of water. After 30 min of sitting, exhaled breath was again collected. The 13 CO 2 content within the initial and 30-min expiratory air bags were analyzed using an HG-IRIS13C infrared spectrometer (Beijing Richen-Force Science & Technology Co., China). Following 30 min of administration, a delta over baseline (DOB) value of 4% or above was regarded as a positive indicator of H. pylori infection.

C-UBT
The 14 C-UBT (Zhonghe Headway Bio-Sci & Tech Co. Ltd., China) was performed according to the manufacturer's instruction. Briefly, patients who fasted for at least four hours were requested to ingest a gelatin capsule containing 0.75 µCi of 14 C isotope with 20 mL of water. After 25 min, each patient was then asked to exhale continuously into a bottle until the purple-colored CO 2 capturing liquid within turned colorless. The scintillation fluid was subsequently added, and the homogenized solution was measured for 14 CO 2 quantity. A reading with more than 100 disintegrations per minute (DPM) was classified as H. pylori positive.

Histopathology
Two gastric biopsy specimens (one each from antrum and corpus) were sent to Da'an Clinical Laboratory (Guangzhou, China) for histopathology examination. The histopathologists were unaware of the UBT results. The presence of Helicobacter-like organism was confirmed with routine hematoxylin and eosin (HE) staining. Giemsa staining was further performed if HE could not confirm the presence of H. pylori clearly, and in those patients who had chronic active gastritis, but no H. pylori found in HE stains.

Statistical analysis
The sensitivity and specificity values of each UBT method were reported according to manufacturer's recommended cut-off value. To evaluate the diagnostic capacity of each UBT method, the receiver operating characteristic (ROC) curve was generated by plotting the true-positive rate on the y axis against the true-negative rate on the x axis [12]. Our ROC analysis was performed using the R package pROC (version 1.16.2) [13]. The area under the curve (AUC) was calculated to quantify the overall accuracy of each UBT method to diagnose H. pylori infection outcomes. The optimal cutoff UBT value that generates the highest true positive rate together with the lowest false positive rate, was determined by using maximum Youden index method, where Youden index = sensitivity + specificity − 1 [14]. For the comparison of categorical variables, the Fisher's exact test was used. The level of statistical significance was considered at p < 0.05.
We also compared the discordance of H. pylori infection status as determined by each UBT assay to histopathology in three different age groups (Table 2). Interestingly, in patients aged 18-30 years, there was a significantly higher discordance between the 13 C-UBT and histopathology outcomes as compared to the 14 C-UBT counterparts (29.4% versus 8.3%, P = 0.032). On the other hand, in patients aged above 50 years, the discordance was significantly greater in the 14 C-UBT group than those who were tested by the 13 C-UBT (28.4% versus 12.2%, P = 0.045). No significant difference was observed in the 31-50 years patient group.
To improve the sensitivity of each assay, we thought that it was necessary to establish an additional cutoff value as a negative indicator of H. pylori infection, following which the values situated between the upper and lower cutoffs would be classified as indeterminate results and therefore require repeated testing. Again, ROC analysis was performed on each UBT assay and this time, only with values below the optimal cutoff as previously determined. As shown in Fig. 3, the new cutoff value was 3% DOB for 13 C-UBT, and 87 DPM for 14 C-UBT, which in turn implies that UBT values less than 3% DOB or 87 DPM were very likely to be H. pylori negative. Taking both upper and lower cutoff values for each assay into consideration, we recommend that for 13 C-UBT readings   (Table 4). Further, the diagnostic accuracy of the 13 C-and 14 C-UBT in this population can be boosted from initially 83.9% (CI 78.4%-88.2%) and 85% (CI

Discussion
The urea breath test is widely accepted as an accurate non-invasive method for diagnosing H. pylori infection.
In the present study, we assessed the diagnostic performance of the 13 C-and 14 C-UBT commercial kits used in our hospital by comparing each UBT outcome against that of histological examination, which was considered the "gold standard" reference method for determination of H. pylori infection. Importantly, based on histology, the 14 C-UBT was significantly more accurate than the 13 C-UBT in determining H. pylori infection status in patients aged 18-30 years, whereas, in patients older than 50 years of age, the 13 C-UBT method was more accurate than the 14 C-UBT. The 13 C-UBT measures the ratio of labeled CO 2 to human respiratory CO 2 in the breath. Hence its outcome can be affected by one's gender, age, urea hydrolysis rate and CO 2 production rate [15]. Therefore, in the event where there were many false-positive 13 C-UBT results among the young patients, it is possible that these individuals have a relatively low basal CO 2 production rate and/or a high urea hydrolysis rate, releasing breath with a proportionally higher quantity of labeled CO 2 and thus, generating a false-positive DOB value. In the situation where 14 C-UBT generated substantially more false-positive results than 13 C-UBT in older patients, some of these patients might have hypochlorhydria, a condition where there is a low-level production of gastric acid and which is commonly associated with aging, leading to the growth of urease-producing non-H. pylori bacteria originating either from the oral cavity or the intestine and thus, a UBT-positive outcome [16][17][18].
Also, it is also possible that the four-hour fasting time (prior to ingesting a capsule containing 50 mg 13 C isotope-labeled urea) of this current study is insufficient to empty the stomach in some of these individuals. This situation is most likely to generate a less-acidic gastric environment, which would be rather permissive for the growth of other bacteria with urease activity that could eventually induce a false-positive UBT reaction. Therefore, a longer fasting period, potentially overnight when possible, should be considered the preferred option before testing. Additionally, attention to detail when performing the tests could improve the accuracy. As an example, cleaning teeth and mouth immediately prior to the test might decrease gastric contamination from swallowed oropharyngeal (urease positive) bacteria. Ensuring the patient was sitting quietly prior to the test would lower the amount of endogenous CO 2 resulting in a slightly higher breath enrichment of the isotope.
Depending on populations and the doses of 13 C-urea or 14 C-urea, no one-size-fits-all UBT cutoff value can be used to define whether an individual is H. pyloripositive or -negative [19][20][21][22]. In our study, to overcome the low specificity of each UBT kit, two optimal cutoff points, indicating UBT-positive and -negative, respectively, along with an indeterminate zone to address UBT readings that are inconclusive, were established. The intermediate zone, defined as ranging from 3% to 10.3% DOB for 13 C-UBT, and from 87-237 DPM for 14 C-UBT, contained at least half of the false-positive test results in this study. By using new optimal cutoff values and including an indeterminate range, the false positive rates can be greatly reduced. More importantly, we suggest that

Limitations of study
We concede that there are limitations in this study which was performed in a busy clinical setting rather than in a formal research environment. Despite with the new cutoffs and the establishment of an indeterminate range, the specificity of each UBT kit was only improved to approximately 77%, which is still considerably lower than reported (10). The lower specificity of each UBT kit used in this study was probably due to the misdiagnosis of H. pylori infection when histology alone was used as the reference method, as its accuracy depends on the skills of the operator, the size and number of biopsies taken and whether or not the biopsy site contained H. pylori or missed it by chance.
To address this issue, a larger cohort study with a more even distribution of different age categories, as well as the ability to test every study participant using both 13 C-and 14 C-UBT, and the concordant use of two or three methods as references should be further performed. For example, rapid urease test, bacterial culture, histology and even PCR, would create a truer "gold standard" and allow for better comparison of the diagnostic accuracy of both tests and the validation of our suggested indeterminate zone. At the same time, observation of how the tests were administered in a busy clinical setting could increase the value of the test even further.

Conclusions
Via comparing the UBT outcomes to that of histopathology examination, we demonstrated that both 13 C-and 14 C-UBT kits used in this study have high sensitivity but low specificity. Based on ROC analysis and the maximum Youden index method, new optimal cutoff values were identified and used to establish an indeterminate range (3-10.3% DOB for 13 C-UBT and 87-237 DPM for 14 C-UBT), improving the specificity from 62.2% to 76.7% and 54.7% to 76.9% for the 13 C-and 14 C-UBT, respectively. We strongly suggest that for any patient who had an indeterminate UBT result, a repeated UBT or other diagnostic test should be performed to confirm H. pylori infection, avoiding misdiagnosis and unnecessary antibiotic treatment. For future studies, a larger cohort study with two or three methods as references should be further performed to validate our suggested indeterminate zone.