¹⁸F-FDG uptake in the stomach on screening PET/CT: value for predicting Helicobacter pylori infection and chronic atrophic gastritis

Medical records of subjects who underwent ¹⁸F-FDG PET/CT for cancer screening between April 2005 and November 2015 were retrospectively investigated. Among them, 88 subjects underwent gastrointestinal fiberscopy within 6 months of the PET/CT study. The reasons for gastrointestinal fiberscopy were increased uptake of ¹⁸F-FDG in the stomach (55 subjects), previous history of peptic ulcer (5 subjects), familial history of peptic ulcer or gastric cancer (4 subjects), or request by the examinee (32 subjects). None of these 88 subjects had a previous history of gastric cancer or MALT lymphoma. The presence or absence of H. pylori infection, as well as the diagnosis of chronic atrophic gastritis, were determined in the medical records in all 88 subjects who underwent gastrointestinal fiberscopy. Thus, ¹⁸F-FDG PET/CT images were evaluated in these 88 subjects. This retrospective study was approved by the institutional review board of Mie University Hospital (study no. 2989) and was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice. Informed consent was waived for this retrospective study.

All subjects fasted for at least 6 h before PET/CT acquisitions. Prior to ¹⁸F-FDG injection, blood glucose levels were determined from capillary blood samples and were confirmed to be less than 150 mg/dl in all subjects. A 3.7-MBq/kg dose of ¹⁸F-FDG was injected intravenously in one arm. PET/CT was performed by using an Aquiduo PCA-700B scanner (Toshiba, Nasu, Japan) or Discovery PET/CT 690 scanner (GE, Milwaukee, WI). Images from the skull to the mid-thigh were acquired approximately 60 min after ¹⁸F-FDG injection, by employing 3-dimentional acquisitions in 7-9 bed positions with 2-min acquisition in each position. Subjects were placed supine with the arms alongside the body or lifted up to the skull and were allowed to breathe normally during PET acquisitions. CT images acquired in approximately ten seconds during a natural breath-holding were used for attenuation correlation and generation of fusion images. Attenuation-corrected PET images with co-registered CT data were reviewed.

¹⁸F-FDG uptake in the stomach was measured semi-quantitatively by placing volumes of interest (VOIs) at the fornix, corpus and antrum of the stomach as well as in the liver by consensus of two observers. The VOIs were 3D spheres and the size of VOIs were 5 mm in diameter for the stomach, and 30 mm in diameter for the liver. The VOI for the stomach was carefully placed in the gastric wall by monitoring both PET-CT fusion images and PET images. VOI for the liver was placed to avoid the region just blow the diaphragm for preventing the motion blurring artifact. For each VOI, maximal SUV (SUVmax) and mean SUV (SUVmean) were recorded (Fig. 1). ROC analysis was performed for several different SUV indicators. Maximum SUVmax and mean SUVmax were the maximum and mean values of the SUVmax measured at the fornix, corpus and antrum. Maximum SUVmean and mean SUVmean were the maximum and mean values of the SUVmean measured at the fornix, corpus and antrum. In addition, maximum SUVmax / SUVmean liver, mean SUVmax /SUVmean liver, maximum SUVmean / SUVmean liver and mean SUVmean / SUVmean liver were determined to evaluate the diagnostic performance of these indicators in predicting H pylori infection and chronic atrophic gastritis.

SPSS version 22.0 software (IBM Japan, Tokyo, Japan) was used for statistical analyses. We determined whether statistically significance difference was observed in SUVs of the stomach between those with and without H. pylori, and those with and without chronic atrophic gastritis. The sensitivity and specificity of SUV indicators in predicting H. pylori infection and chronic atrophic gastritis were calculated by using an optimal cut-off point on the ROC curve that has the minimum distance to the upper left corner (where sensitivity = 1 and specificity = 1). The statistically significance was evaluated by Mann-Whitney U-test or Wilcoxon rank sum test. All analysis were 2-sided, a P-value of less than 0.05 was considered statistically significant.

Characteristics of the subjects including laboratory diagnosis by gastrointestinal fiberscopy are shown in Table 1. Diagnosis of H. pylori infection was made by a rapid urease test, a stool antigen test and an information of previous medical institution or prevention center. Three subjects who had chronic atrophic gastritis without H. pylori infection on medical records in previous medical institutions, were turned out to be H. pylori positive by further investigation in our hospital.

Table 2 summarizes the SUVmax and SUVmean of 18 F-FDG uptake at the fornix, corpus and antrum, as well as the maximum and the mean values of SUVmax and SUVmean at 3 regions in the stomach in associated with H. pylori infection. Table 3 summarizes these SUV indicators in associated with chronic atrophic gastritis. All of these SUV indicators in the stomach were significantly higher in patients with H. pylori infection than in those without H. pylori (P < 0.001). In addition, all of these SUV indicators were significantly higher in patients with chronic atrophic gastritis than in those without chronic atrophic gastritis (P < 0.001). It was also noted that the 18 F-FDG uptake of the fornix in the stomach was significantly higher than those in corpus and antrum, independent of the presence of H. pylori infection and chronic atrophic gastritis.

Figure 2 shows ROC curves for SUV indicators in predicting H. pylori infection and chronic atrophic gastritis. In Table 4, the area under ROC curves, the optimal cut-off values, the sensitivities and specificities in predicting H. pylori infection and chronic atrophic gastritis are presented. All of SUV indicators demonstrated good diagnostic performance for the prediction of H. pylori infection and chronic atrophic gastritis. Among these 4 SUV indicators, mean SUVmean exhibited the highest area under ROC curves for predicting H. pylori infection (0.807, 95%CI 0.715 – 0.898) and for chronic gastritis (0.784, 95 % CI 0.684 – 0.884). As shown in Table 5, normalization of these SUV indicators in the stomach by the liver SUV did not improve the area under ROC curves for the diagnosis of H. pylori infection. For predicting chronic atrophic gastritis, normalization of SUV indicators by the liver SUV slightly improve the area under ROC curves, with the highest area under ROC curve of 0.793 (95 % CI 0.686 – 0.900) by mean SUVmax. However, the amount of improvement with normalization by the liver SUV was quite limited.

Dot plots of mean SUVmean in subjects with and without H. pylori infection and in those with and without chronic gastritis were shown on Fig. 3. The sensitivity and specificity of mean SUVmean were 86.5 % and 70.6 % for H. pylori infection (optimal cut-off value of 2.66), and 75.0 % and 78.6 % for chronic gastritis (optimal cut-off value of 2.57), respectively.

Among the 88 subjects, seven neoplasms were found on gastrointestinal fiberscopy, including four early gastric cancers, two gastric adenomas and a MALT lymphoma. In four gastric cancer and two gastric adenomas, no focal increase in ¹⁸F-FDG uptake corresponding to tumors was observed, while H. pylori was positive in these cases. In a patient with MALT lymphoma, the antibody test of H. pylori was negative and increased focal ¹⁸F-FDG uptake at gastric corpus was detected which corresponded to MALT lymphoma proven by gastrointestinal fiberscopy. The infection of H. pylori was also demonstrated by histologic specimen taken by fiberscopy.

Pattern of accumulation of ¹⁸F-FDG in the stomach and its associated with endoscopic findings of the gastric mucosa and H. pylori infection were previously investigated by Takahashi et al [11] by using a visual assessment of ¹⁸F-FDG PET image. They classified ¹⁸F-FDG uptake in the stomach into three groups (A: localized accumulation in the fornix, B: diffuse accumulation throughout the entire stomach, C: no accumulation). They found that H. pylori infections were more frequent in Groups A and B than in Group C, concluding that accumulation of ¹⁸F-FDG in the stomach suggests a high probability of inflammatory changes to the gastric mucosa, forming a background for the development of cancer or malignant lymphoma. In our current study, we used a more objective approach by measuring SUVs of ¹⁸F-FDG in the fornix, corpus and antrum. Consistent with previous report [8, 11], we found that ¹⁸F-FDG uptake of the fornix was significantly higher than corpus and antrum. In addition, SUV of ¹⁸F-FDG in the fornix was significantly higher than those in corpus and antrum, not only in the subjects with H. pylori infection and chronic atrophic gastritis, but also in those without H. pylori infection or chronic atrophic gastritis, suggesting that high ¹⁸F-FDG uptake in an oral side of the stomach is physiological. We also noticed that H. pylori infection and chronic atrophic gastritis are associated with elevated SUVs in all gastric regions including the fornix, corpus and antrum. This indicates that H. pylori infection causes increased ¹⁸F-FDG uptake reflecting active inflammation throughout the entire stomach.

For the assessment of patients with advanced gastric cancer, ¹⁸F-FDG PET/CT has been shown to be useful in detecting nodal metastasis and distant metastasis, and in predicting prognosis [16‐20]. However, ¹⁸F-FDG PET/CT is not useful for screening gastric cancers [5, 11, 21, 22]. Shoda et al. studies 2861 asymptomatic subjects and found that the sensitivity of ¹⁸F-FDG PET for gastric cancer was as low as 10 % [3]. Consequently the use of gastrointestinal fiberscopy is considered more appropriate in screening gastric cancer.

Our results demonstrated that semi-quantitative assessment of ¹⁸F-FDG uptake with SUV has high diagnostic accuracy in predicting H. pylori infection and chronic atrophic gastritis. As previously mentioned, H. pylori infection and subsequent chronic atrophic gastritis lead to increased risk of gastric cancer formation. Inflammatory change in the gastric mucosa caused by H. pylori forms a background for the development of gastric cancer or malignant lymphoma. In a Japanese cohort study, the population attributable fraction (PAF) of H. pylori infection for gastric cancer incidence (i.e. the fraction of gastric cancer incident cases that is attributable to H. pylori infection) was estimated to be 84 % [3]. Despite the declined prevalence of H. pylori infection for the past 30 years, gastric cancer is the second most frequent cause of cancer death in both males and females in Japan, and the most frequent cancer in males and the second most frequent cancer in females [23]. Therefore, gastrointestinal fiberscopy should be strongly recommended for subjects with increased ¹⁸F-FDG uptake in the stomach. According to the results in this study, high area under ROC of 0.807 and high sensitivity of 86.5 % can be achieved when mean SUVmean values of > 2.66 was used as a threshold.

Several limitations must be acknowledged in this study. First, this is a single-center study with a limited number of subjects, and there is a selection bias for subjects who underwent gastrointestinal fiberscopy. Second, the degree of chronic atrophic gastritis was not evaluated in current study, because the laboratory diagnosis by gastrointestinal fiberscopy was qualitative and operator-dependent. Third, CT images were acquired during natural breath-holding while PET images were obtained during free-breathing. This may result in misregistration artifact and alteration in SUV. Forth, SUV of ¹⁸F-FDG uptake in the stomach was not compared with the gastrointestinal fiberscopy findings in detail. Further investigation by prospective multi-center study using both PET-CT and gastrointestinal fiberscopy is necessary to determine the value of ¹⁸F-FDG PET in early detection and prevention of gastric cancer.