Utility of Kyoto Classification of Gastritis in subjects with a high-negative titer of anti-Helicobacter pylori antibody during a medical check-up

Abstract

Subjects with a high-negative titer (3–9.9 U/ml) of serum anti-Helicobacter pylori (H. pylori) antibody represent a heterogeneous group of currently H. pylori-infected, H. pylori-uninfected, and previously H. pylori-infected cases. We investigated the characteristics of subjects with a high-negative titer during a medical check-up and the utility of H. pylori infection score, the sum of scores of endoscopic findings based on the Kyoto Classification of Gastritis, for diagnosing H. pylori infection. Subjects with ¹³C-urea breath test-positive or H. pylori stool antigen test-positive were diagnosed as currently H. pylori-infected. Although around half of subjects with a high-negative titer were after eradication therapy (48.6%), currently H. pylori-infected were considerably confirmed (11.7%). H. pylori infection score showed a high value of area under the receiver operating characteristic curve [0.92; 95% confidence interval (CI), 0.84–1.00] with the most suitable cut-off value of 1.0 (sensitivity: 0.92; specificity: 0.90). Multivariate logistic regression analysis revealed that H. pylori infection score was an independent factor associated with increased prevalence of H. pylori infection (odds ratio, 9.53; 95% CI, 2.64–34.40; p<0.001). Currently H. pylori-infected subjects were considerably included among the subjects with a high-negative titer, and the Kyoto Classification of Gastritis was useful to predict current H. pylori infection.

Introduction

Helicobacter pylori (H. pylori) is a leading cause of chronic gastritis and peptic ulcers. It is considered that H. pylori-induced gastritis is the most important risk factor for gastric cancer.⁽¹⁾ As H. pylori causes approximately 90% of non-cardiac gastric cancers and there is evidence that H. pylori eradication therapy reduces by 30–40% the incidence of gastric cancer, the international agency for Research on Cancer working group recommends population-based screening and eradication therapy.⁽²⁾ In Japan, Kyoto global consensus report states that all of H. pylori-positive individuals should receive eradication therapy.⁽³⁾ An accurate, objective, and cost-effective screening method for diagnosing H. pylori infection is required in medical check-up.

Detection of serum anti-H. pylori antibody, produced during the immune response to H. pylori in the gastric mucosa, is a non-invasive test for diagnosing H. pylori infection. The antibody titer increases as the immune response becomes stronger and decreases after the successful eradication therapy or the natural disappearance of H. pylori. As the test is easy and is not affected by proton pump inhibitors that influence gut microbiota,⁽⁴⁾ it is widely used with serum pepsinogen concentration during a medical check-up in Japan. Previously, the cut-off value of the antibody (E plate ‘Eiken’ H. pylori antibody) used in clinical practice for diagnosing H. pylori infection was 10 U/ml. However, H. pylori-infected subjects were sometimes mixed among the subjects with anti-H. pylori antibody <10 U/ml. To increase the sensitivity and to identify subjects with a high-risk for gastric cancer in a medical check-up, the cut-off value has been changed to 3 U/ml since April 2017. Anti-H. pylori antibody titer between 3–9.9 U/ml is defined as “high-negative titer,” and subjects with a high-negative titer represent a heterogeneous group of currently H. pylori-infected, H. pylori-uninfected, and previously H. pylori-infected cases. They need to undergo esophagogastroduodenoscopy (EGD) and additional examination such as ¹³C-urea breath test (UBT) or H. pylori stool antigen test (HpSA) for diagnosing H. pylori infection in Japan. As UBT has high diagnostic accuracy,⁽⁵⁾ it is the best recommended non-invasive test in the Maastricht V/Florence consensus report.⁽⁶⁾ HpSA is also a non-invasive method that detects directly with high accuracy the H. pylori-derived antigen excreted through gastrointestinal tract,⁽⁷⁾ and has the advantage of reliably detecting H. pylori in the remnants of the stomach after distal gastrectomy.⁽⁸⁾

In evidence-based guidelines for gastric cancer screening 2014 in Japan, EGD is recommended for gastric cancer screening during a medical check-up because there is accumulating evidence that endoscopic screening effectively reduces mortality from gastric cancer.^(9–12) EGD is a useful tool to diagnose gastric cancer, and it can also evaluate the progress of atrophic gastritis and the status of H. pylori infection at the same time. Kimura-Takemoto Classification was introduced in 1969 to evaluate the progress of atrophic gastritis according to the location of the endoscopic atrophic border.⁽¹³⁾ The Kyoto Classification of Gastritis was proposed in 2014 to standardize the endoscopic diagnosis of H. pylori-infected gastritis. According to the Kyoto Classification of Gastritis, endoscopic findings of currently H. pylori-infected gastric mucosa are diffuse redness, mucosal swelling, enlarged tortuous fold, sticky mucus, and nodularity. These endoscopic findings will be useful to decide whether the gastric mucosa is infected with H. pylori or not.

In this study, we investigated the actual status and the utility of the Kyoto Classification of Gastritis for diagnosing H. pylori infection in subjects with a high-negative titer of anti-H. pylori antibody during a medical check-up.

Methods

Study design

This is a single-center, retrospective, observational study.

Study population

Between April 2017 to December 2019, a total of 38,789 subjects underwent a medical check-up in the clinic MedCity21. Anti-H. pylori IgG antibody was measured in 13,203 subjects and a high-negative titer of anti-H. pylori antibody was confirmed in 1,690 subjects (12.8%). This study was focused on 111 subjects with a high-negative titer of anti-H. pylori antibody who consulted our clinic and were subsequently examined by UBT or HpSA for diagnosing H. pylori infection.

The clinical characteristics [age, sex, body mass index (BMI), alcohol intake, and smoking habit] of these individuals were obtained from their medical records. Exclusion criteria were as follows: lacking data, unknown history of H. pylori eradication, and users of proton-pump inhibitor, potassium-competitive acid blocker, low-dose aspirin, nonsteroidal anti-inflammatory drugs, or steroid.

Examinations of H. pylori infection

We used the anti-H. pylori antibody for the serological diagnosis of H. pylori infection (E plate ‘Eiken’ H. pylori antibody; Eiken Chemical Co., Ltd., Tochigi, Japan). The anti-H. pylori antibody titer of 3–9.9 U/ml was defined as high-negative titer. Subjects with a high-negative titer underwent UBT using a 100 mg ¹³C-urea tablet (Otsuka Pharmaceutical Co., Ltd., Tokyo, Japan) or HpSA (Becton, Dickinson and Company, Bergen, NJ) for diagnosing H. pylori infection. A value of UBT ≥2.5‰ was defined as UBT-positive.

Endoscopic examination

Screening with EGD was performed for all subjects with a high-negative titer by several expert endoscopists. We used a 5.4 mm-diameter endoscope for EGD (GIF-XP290N; Olympus Medical Systems Co., Ltd., Tokyo, Japan).

We referred to Kimura-Takemoto Classification to evaluate the progress of atrophic gastritis.⁽¹³⁾ The classification scores of gastric atrophy were categorized into seven grades: Closed (C)-0, C-1, C-2, C-3, Open (O)-1, O-2, and O-3. The C-type indicates atrophic pattern with a margin between the non-atrophic fundic mucosa and atrophic mucosa located in the lesser curvature of the stomach. O-type indicates atrophic pattern whose margin does not cross the lesser curvature. We classified them into no atrophy (C-0 or C-1), minor atrophy (C-2 or C-3), and severe atrophy (O-1 to O-3).

Furthermore, we utilized the Kyoto Classification of Gastritis to determine the presence or absence of H. pylori infection. The existence of diffuse redness, spotty redness, mucosal swelling, enlarged tortuous fold, sticky mucus, or nodularity suggests currently H. pylori-infected gastric mucosa. Diffuse redness is uniform redness with continuous expansion observed in non-atrophic mucosa, and spotty redness is irregular dots-like redness that appears on the background of diffuse redness. We considered spotty redness as one of the findings of diffuse redness in this study. Mucosal swelling is recognized as soft thick mucosa, which may appear uneven in swollen gastric areas.^(14,15) Enlarged tortuous fold is a fold that is not flattened or is only partially flattened by insufflation with a thickness ≥5 mm and visually tortuous.⁽¹⁶⁾ Sticky mucus is cloudy mucus adhered to the mucosal surface prior to washing with water. Nodularity is characterized by uniform small granular elevations of the gastric mucosa like goosebumps.⁽¹⁷⁾ Nodularity usually flattens but white dots sometimes remain after eradication therapy of H. pylori. We considered the flattening of small granular elevations with or without white dots as negative in this study. Typical images of these endoscopic findings are shown in Fig. 1. The presence of these endoscopic findings, according to the Kyoto Classification of Gastritis, were independently evaluated by three raters with more than five years of experience of endoscopy who were blinded to eradication history and the results of anti-H. pylori antibody, UBT, and HpSA. It was judged to be present if more than two raters agreed on the presence of endoscopic findings.

Fig. 1.

Typical endoscopic images of currently H. pylori-infected gastric mucosa. (A) diffuse redness, (B) mucosal swelling, (C) enlarged tortuous fold, (D) sticky mucus, (E) nodularity.

Assessment of H. pylori infection

UBT-positive or HpSA-positive individuals were diagnosed as having current H. pylori infection. Individuals with a successful eradication history for H. pylori were diagnosed as previously H. pylori-infected after eradication therapy. UBT or HpSA was performed for individuals who have not received UBT or HpSA after eradication therapy, and UBT-positive or HpSA-positive individuals were judged as having current H. pylori infection.

We developed “H. pylori infection score,” that is the sum of scores of the following endoscopic findings based on the Kyoto Classification of Gastritis: diffuse redness (none 0, presence 1); mucosal swelling (none 0, presence 1); enlarged tortuous fold (none 0, presence 1); sticky mucus (none 0, presence 1); and nodularity (none 0, presence 1).

Outcome measurement

The primary aim of this study was to examine the actual status and the utility of H. pylori infection score based on the Kyoto Classification of Gastritis for diagnosing H. pylori infection in subjects with a high-negative titer of anti-H. pylori antibody during a medical check-up.

Ethical approval

The study protocol was approved by the ethics committee of Osaka City University Graduate School of Medicine (No. 3763) on May 26, 2017. We have disclosed the information about this study on the webpage of the institution and the subjects had the opportunity to opt out. The study was conducted in accordance with the Declaration of Helsinki ethical principles for medical research involving human subjects.

Statistical analysis

Data are expressed as median and interquartile range (IQR) for continuous variables and as numbers for categorical variables. Receiver operating characteristic (ROC) curves were constructed, and the threshold at which the sum of sensitivity and specificity became maximum was determined as optimal cut-off value for predicting H. pylori infection. The Kendall’s coefficient of concordance (Kendall’s W) was used to assess agreement among raters of endoscopic images. Clinical factors associated with current H. pylori infection were investigated using univariate and multivariate logistic regression analysis, and the odds ratio (OR) and 95% confidence intervals (CI) were estimated for each variable. The statistical calculations and analyses were performed using IBM SPSS Statistics ver. 26 (IBM Corporation, Armonk, NY). The ROC curve was constructed using R software, ver. 3.6.1 (The R Foundation for Statistical Computing, Vienna, Austria). P values <0.05 were considered statistically significant.

Results

H. pylori infection status of subjects with a high-negative titer of anti-H. pylori antibody

Among the 111 subjects with a high-negative titer of anti-H. pylori antibody who consulted our clinic for diagnosing H. pylori infection, 11.7% were currently H. pylori-infected, 48.6% were previously H. pylori-infected after eradication therapy, and the others were 39.6% (Table 1).

Table 1.

H. pylori infection status of subjects with a high-negative titer of anti-H. pylori antibody

H. pylori infection status	n (%)
Currently H. pylori-infected	13 (11.7%)
Previously H. pylori-infected after eradication therapy	54 (48.6%)
The others	44 (39.6%)

Baseline characteristics and endoscopic findings of subjects with a high-negative titer of anti-H. pylori antibody

The clinical characteristics of the enrolled subjects are shown in Table 2. The median age of subjects with a high-negative titer of anti-H. pylori antibody during a medical check-up was 60.0 (IQR, 51.0 to 68.5). The subjects who had been drinking alcohol more than 5 days a week were 28 (25.2%), and those who currently smoked were 16 (14.4%). The median value of the anti-H. pylori antibody was 4.0 (IQR, 3.0 to 6.0). UBT positive (≥2.5‰) was confirmed in ten subjects and HpSA was confirmed in three subjects.

Table 2.

Baseline characteristics of subjects

Variables	Value or n (%)
Number of cases	111
Age [years, median (IQR)]	60.0 (51.0, 68.5)
Sex
Male	53 (47.7%)
Female	58 (52.3%)
BMI [kg/m2, median (IQR)]	22.7 (20.0, 24.5)
Alcohol intake (5 days/week)
Yes	28 (25.2%)
No	83 (74.8%)
Smoking habit (current smoker)
Yes	16 (14.4%)
No	95 (85.6%)
Anti-H. pylori antibody [U/ml, median (IQR)]	4.0 (3.0, 6.0)
UBT
≥2.5‰	10
<2.5‰	88
HpSA
Positive	3
Negative	10
Atrophy (no atrophy/minor atrophy/severe atrophy)
All subjects	26/52/33 (23.4%/46.8%/29.7%)
Currently H. pylori-infected	0/6/7 (0%/46.2%/53.8%)
Previously H. pylori-infected after eradication therapy	1/36/17 (1.9%/66.7%/31.5%)
The others	25/10/9 (56.8%/22.7%/20.5%)
Diffuse redness 1/0	17/94 (15.3%/84.7%)
Mucosal swelling 1/0	5/106 (4.5%/95.5%)
Enlarged tortuous fold 1/0	1/110 (0.9%/99.1%)
Sticky mucus 1/0	8/103 (7.2%/92.8%)
Nodularity 1/0	2/109 (1.8%/98.2%)

IQR, interquartile range; BMI, body mass index; UBT, ¹³C-urea breath test; HpSA, H. pylori stool antigen test.

In endoscopic findings of Kimura-Takemoto Classification, no atrophy, minor atrophy, and severe atrophy were found in 26, 52, and 33 subjects, respectively. No atrophy was uncommon in currently H. pylori-infected and previously H. pylori-infected after eradication therapy. In the others, 25 subjects had no atrophy and 19 subjects had minor or severe atrophy. In endoscopic findings of the Kyoto Classification of Gastritis, diffuse redness was found in 17 subjects (15.3%), mucosal swelling was found in five subjects (4.5%), enlarged tortuous fold was found in one subject (0.9%), sticky mucus was found in eight subjects (7.2%), and nodularity was found in two subjects (1.8%).

Utility of endoscopic findings in predicting H. pylori infection

We examined the utility of endoscopic findings in predicting H. pylori infection in subjects with a high-negative titer of anti-H. pylori antibody. The Kendall’s W and sensitivity, specificity, and accuracy of diffuse redness, mucosal swelling, enlarged tortuous fold, sticky mucus, and nodularity are shown in Table 3. The Kendall’s W was highest in sticky mucus (0.94) and lowest in enlarged tortuous fold (0.73). The sensitivity was highest in diffuse redness (0.77) and the specificity was highest in enlarged tortuous fold (1.00). The accuracy was highest in diffuse redness and mucosal swelling (0.91).

Table 3.

Utility of endoscopic findings based on the Kyoto Classification of Gastritis in predicting H. pylori infection

Endoscopic findings	Kendall’s W	Sensitivity	Specificity	Accuracy
Diffuse redness	0.81	0.77	0.93	0.91
Mucosal swelling	0.75	0.31	0.99	0.91
Enlarged tortuous fold	0.73	0.08	1.00	0.89
Sticky mucus	0.94	0.39	0.97	0.90
Nodularity	0.90	0.08	0.99	0.88

Next, we investigated the utility of the H. pylori infection score, which is the summed score of diffuse redness, mucosal swelling, enlarged tortuous fold, sticky mucus, and nodularity, to predict H. pylori infection. The ROC curve is shown in Fig. 2. Area under the ROC curve (AUC) of H. pylori infection score was 0.92 (95% CI, 0.84–1.00). The optimal cut-off value, the threshold that the sum of sensitivity and specificity became maximum, of H. pylori infection score was 1.0 (sensitivity: 0.92; specificity: 0.90).

Fig. 2.

The ROC curve of H. pylori infection score in predicting H. pylori infection.

Clinical factors associated with H. pylori infection

We examined the clinical factors associated with H. pylori infection, including age, sex, BMI, alcohol intake, smoking habit, anti-H. pylori antibody, and H. pylori infection score. Univariate logistic regression analysis showed that alcohol intake (≥5 days/week) (OR, 4.28; 95% CI, 1.30–14.10; p = 0.017), anti-H. pylori antibody titer (OR, 1.73; 95% CI, 1.27–2.34; p<0.001), and H. pylori infection score (OR, 13.70; 95% CI, 3.85–48.60; p<0.001) were significantly associated with H. pylori infection. Multivariate logistic regression analysis including alcohol intake, anti-H. pylori antibody, and H. pylori infection score revealed that the H. pylori-infection score was an independent factor associated with the increased prevalence of H. pylori infection (OR, 9.53; 95% CI, 2.64–34.40; p<0.001) (Table 4).

Table 4.

Univariate and multivariate analyses of factors associated with H. pylori infection

Variables	Univariate			Multivariate
	OR (95% CI)	p value		OR (95% CI)	p value
Age	1.01 (0.97–1.07)	0.569
Sex
Male	0.93 (0.29–2.97)	0.903
Female	1.00
BMI (kg/m2)	0.98 (0.80–1.19)	0.805
Alcohol intake (≥5 days/week)
Yes	4.28 (1.30–14.10)	0.017		1.91 (0.32–11.30)	0.477
No	1.00			1.00
Smoking habit (current smoker)
Yes	1.96 (0.48–8.08)	0.351
No	1.00
Anti-H. pylori antibody (per 1 U/ml increase)	1.73 (1.27–2.34)	<0.001		1.39 (0.92–2.09)	0.119
H. pylori infection score (per 1 score increase)	13.70 (3.85–48.60)	<0.001		9.53 (2.64–34.40)	<0.001

BMI, body mass index; OR, odds ratio; CI, confidence intervals.

Discussion

In this study, we examined the actual status of subjects with a high-negative titer of anti-H. pylori antibody during a medical check-up. We found that currently H. pylori-infected subjects were considerably included (11.7%) among the subjects with a high-negative titer. As currently H. pylori-infected subjects have a potential risk of developing gastric cancer, they need to be treated with eradication therapy at the earlier stage of mucosal atrophy to prevent gastric cancer.⁽¹⁾ It was previously reported that H. pylori-infected was found in 11.7–16.9% in patients with a high-negative titer in clinical practice.^(18,19) Although the proportion of currently H. pylori-infected was similar to that of our study, this is the first survey on actual status of high-negative titer during a medical check-up. Around half of subjects with a high-negative titer of anti-H. pylori antibody (48.6%) were previously H. pylori-infected after eradication therapy. Previously H. pylori-infected subjects after eradication therapy should be carefully followed up by scheduled endoscopic surveillance. It was reported that the risk of developing gastric cancer was 0.3% per year after the eradication therapy and it increased according to the background mucosal atrophy.^(20,21) It is considered that the others include H. pylori-uninfected (naïve) and previously H. pylori-infected after natural disappearance, unexpected eradication of H. pylori due to antibiotics or spontaneous clearance of H. pylori after advanced atrophy. It is supposed that 19 subjects (17.1%) with minor or severe atrophy among the others are previously H. pylori-infected after natural disappearance and should be carefully followed up. Subjects with spontaneous clearance of H. pylori after advanced atrophy have the highest risk of developing gastric cancer and they should receive scheduled endoscopic surveillance. Altogether, it is considered that approximately 77.4% of subjects with a high-negative titer (currently H. pylori-infected: 11.7%; previously H. pylori-infected after eradication therapy: 48.6%; previously H. pylori-infected after natural disappearance: 17.1%) have risk of developing gastric cancer. Therefore, high-risk subjects will be overlooked if the previous cut-off value of anti-H. pylori antibody (10 U/ml) is applied to diagnose H. pylori infection during a medical check-up.

In the current strategy, subsequent additional test (UBT or HpSA) for diagnosing H. pylori infection is recommended in subjects with a high-negative titer after the screening EGD. However, we consider that macroscopic endoscopic images enable diagnosis of the H. pylori infection status. Toyoshima et al.⁽¹⁹⁾ have previously reported that the scores for the endoscopic findings related to the risk of gastric cancer according to the Kyoto Classification of Gastritis can predict H. pylori infection in H. pylori-infected and H. pylori-uninfected subjects with a high-negative titer. The above-mentioned endoscopic findings include atrophy, intestinal metaplasia, fold enlargement, nodularity, and diffuse redness. Although these endoscopic findings may be useful to predict H. pylori infection, they cannot distinguish current H. pylori infection from a previous H. pylori infection because atrophy and intestinal metaplasia remain even after the eradication of H. pylori. In this regard, we chose the following five endoscopic findings associated with H. pylori infection that disappear relatively quickly after the eradication therapy: diffuse redness, mucosal swelling, enlarged tortuous fold, sticky mucus, and nodularity. We examined the utility of these endoscopic findings to predict H. pylori infection in subjects with a high-negative titer. Diffuse redness was mostly found in 15.3% with high sensitivity, specificity, and accuracy. We developed H. pylori infection score, the sum of scores of these endoscopic findings, in this study, and the score can be applied even for previously H. pylori-infected subjects because the endoscopic findings disappear after the disappearance of H. pylori. The H. pylori infection score had high AUC (0.92) and high sensitivity and specificity with the cut-off value of 1.0. This means that if the H. pylori infection score is ≥1, that is, at least one endoscopic finding is confirmed among the endoscopic findings related to current H. pylori infection, the subject is likely to have current H. pylori infection. We proved that H. pylori infection score could be an independent indicator to predict H. pylori infection. Although Suki et al.⁽²²⁾ have reported that there is a positive association between H. pylori infection and increased BMI, this association was not observed in this study. H. pylori infection may be diagnosed during the procedure of screening EGD using the H. pylori infection score, and it may substitute additional examinations.

Nishizawa et al.⁽¹⁸⁾ revealed that false diagnosis using anti-H. pylori antibody was observed in 11.7% of a high-negative titer and combination of anti-H. pylori antibody and endoscopic finding based on the Kyoto Classification of Gastritis could provide a more accurate diagnosis of H. pylori. This means that we should not easily judge the subjects with a high-negative titer to be H. pylori-negative and endoscopic examination is useful to predict the H. pylori infection. Furthermore, it is ideal to know not only H. pylori infection status but risk of gastric cancer and mortality during a medical check-up. Graham et al.⁽²³⁾ have suggested that evaluation of the severity and extent of atrophy leads to the development of a cost-effective surveillance method for gastric cancer. In addition, Inoue et al.⁽²⁴⁾ have recently reported that subjects with a high-negative titer and moderate or severe atrophic gastritis increase the long-term risk of gastric cancer. From these findings, endoscopic diagnosis will be primarily important and we need to establish the best cost-effective surveillance method to diagnose H. pylori infection status and pick-up the high-risk subjects for gastric cancer during a medical check-up.

This study has several limitations. First, this is a small sample sized, retrospective study in a single institution; therefore, selection bias is unavoidable. Second, we used UBT or HpSA as the gold standard for diagnosing H. pylori infection. However, the combination of more than two types of examinations improves accuracy. Third, we did not assess autoimmune gastritis (type A gastritis). Although autoimmune gastritis is rare, it sometimes results in sticky mucus and urease-positive bacteria other than H. pylori gives a UBT-false positive result. We should be careful regarding this point to avoid an endless eradication therapy.⁽²⁵⁾ Fourth, Kendall’s W of mucosal swelling and enlarged tortuous fold was relatively low, and it is considered that these findings are difficult to be judged. To improve the accuracy of H. pylori infection score, it is considered that the education for the raters on the Kyoto Classification of Gastritis is necessary.

In conclusion, currently H. pylori-infected subjects were considerably included among the subjects with a high-negative titer of anti-H. pylori antibody during a medical check-up, and the Kyoto Classification of Gastritis was useful to predict current H. pylori infection.

Author Contributions

KO designed, analyzed data, and drafted the paper; TW supervised this study, and revised the paper; SK, YM, and AN evaluated the endoscopic findings based on the Kyoto Classification of Gastritis; YNadatani, SF, SH, FT, NK, KT, YNagami, TT, TK, SF, NK, and YF reviewed the manuscript and provided valid inputs on the study.

Abbreviations

AUC

area under the receiver operating characteristic curve

BMI

body mass index

C

closed

CI

confidence intervals

EGD

esophagogastroduodenoscopy

HpSA

H. pylori stool antigen test

IQR

interquartile range

O

open

OR

odds ratio

ROC

receiver operating characteristic

UBT

¹³C-urea breath test

Conflict of Interest

No potential conflicts of interest were disclosed.