GC refers to malignant tumors derived from gastric epithelial cells, and vast majority are adenocarcinomatous, including tubular adenocarcinoma, signet ring cell carcinoma, papillary adenocarcinoma, mucinous adenocarcinoma, poorly differentiated adenocarcinoma, undifferentiated carcinoma, and other subtypes. Lymphoma is a malignant tumor of the blood system that occurs in the extranodal sites. The most common extranodal site of non-Hodgkin lymphoma (NHL) is the stomach, which represents 30% − 40% of all extranodal lymphomas and 55% − 65% of all GL [5]. PGL is a rare tumor, with an incidence equals to 4% − 20% that of NHL and approximately 5% of primary gastric neoplasms [6]. PGL refers to a malignancy of lymphoid tissue that originates in the submucosal layer of the stomach wall, and vast majority of these are NHL. Forty percent of PGL are low-grade lymphomas, including MALT lymphoma, and about 60% of PGL are highly malignant lymphomas. The most common pathological type is DLBCL. PET/CT is a new imaging technique that combines systemic functional imaging and anatomical imaging. It has been widely used in diagnosing, staging and evaluation of therapeutic effects.
PGL originates from the lymphoid tissues of lamina propria and submucosa of the gastric mucosa, and the tumor cells infiltrate into the periphery of the mucosa. GC originates from the mucosal epithelial cells, and invades the mucosal and submucosal layers. CT images of GC and PGL showed different degrees of gastric stenosis, thickening of stomach wall, and the extent of gastric wall thickening showed association with the depth of invasion of the tumor and the range of involvement of the stomach wall, but the extent and depth of gastric lesions of PGL are more obvious than GC. In the present study, a significant positive correlation was observed between THKmax and SUVmax of the gastric wall lesions of GC and GL. The results indicated that the THKmax was statistically larger and the SUVmax was significantly higher in patients with GL compared to those with GC. Considering the positive correlation between THKmax of gastric wall lesions and SUVmax, comparative ROC curves were used to obtain the cutoff thresholds for SUVmax or SUVmax/THKmax for differential diagnosis between GL and GC. The optimal cutoff values of SUVmax and SUVmax/THKmax were 16.13 and 0.41, respectively. The sensitivity of SUVmax was higher than that of the SUVmax/THKmax (91.7% vs 48.5%), but the specificity of SUVmax/THKmax was higher than that of the SUVmax (89.4% vs 57.8%). The AUC of SUVmax and SUVmax/THKmax was 0.747 and 0.760, respectively, but no statistical deference was observed. In Li XF et al [7] and Fu Let al [8] studies, the SUVmax/THKmax is considered as a more reliable indicator for identifying GC and GL when compared to SUVmax alone. Further studies with larger number of patients are warranted.
SUVmax is a semi-quantitative indicator of the normalized concentration of radioactivity in a lesion, and this reflects the degree of tumor malignancy and the level of metabolism of the cells. It is one of the important indicators for diagnosing malignant tumors by PET/CT. A correlation between 18F-FDG uptake and histological classification of lymphoma was observed, and this is because the rapidly proliferating lymphoma cells possess a high metabolic rate and aggressive subtypes of NHL takes up high levels of 18F-FDG [9]. High grade PGL is actually always a B-cell phenotype and is related to invasive clinical manifestations, but the role of PET / CT in malt lymphoma is still controversial. MALT lymphoma is often reported as PET negative due to its inert behavior. Previous studies [10, 11] have reported that SUVmax in high-grade GL is higher than that in low-grade GL, and this study confirmed this finding. Although low-grade GL lesions were difficult to identify, especially in MALT lymphoma, MALT types had varied uptakes in this study. In another study [12–14], it was reported that low-grade NHL showed 18F-FDG uptake, but the intensity is lower than high-grade NHL, and the PET results were related to histological findings. After diagnosing PGL, the staging is considered essential for treatment schedule and prognosis assessment. It is also of importance to exclude systemic lymphoma secondary to the stomach. Few years ago, staging for GL remained controversial due to various staging systems in use. After working on staging in lymphoma by an international workgroup, a modified Lugano staging system has now been regarded as the standard in patients with PGL [15]. In the present study, the SUVmax of lymphoma patients with Lugano stage I was significantly lower than that of II1 + II2 + IV, suggesting a positive correlation between SUVmax and lymphoma staging.
At present, for PGL, PET/CT is the standard initial imaging study of DLBCL histology, but it is not recommended for MALT PGL cases. Plasma cell differentiation has been recognized as an important factor affecting the detection rate of MALT by FDG PET in recent years. In contrast, another study revealed that plasma cell differentiation has no effect on the detection of MALT. The differences in the results of this study can be explained by different stages of the study, and with the same pathological subtype. There are many PET or PET/CT studies of gastric MALT lymphoma, in which the stomach is the most common affected organ [16]. Nevertheless, the results were not completely consistent. Enomoto et al [16] have reported 5 patients with gastric MALT lymphoma, and none of them showed abnormal tracer accumulation. Perry et al [17] and Radan et al [18] have reported that only 38.9% and 71% of patients with gastric MALT lymphoma have gastric FDG affinity, respectively. According to the studies of Ambrosini et al [19] and Song et al [20], all cases with gastric MALT demonstrated pathological FDG uptake, but compared with invasive gastric NHL, the FDG uptake of MALT lymphoma was much lower and correlated with treatment response. These discrepancies are due to the presence of heterogeneous cellular population [21], the shape of the lesions [16] and the physiological change or inflammatory process stimulating the uptake of this lymphoma type [17]. In the present study, the incidence of gastric FDG uptake was 97.05% (66/68) in patients with GL. Of the 68 GL patients, two patients with negative gastric tracer accumulation presented with MALT lymphoma, both of which were MALT, and Lugano stage I (Fig. 6A). This suggested that patients with gastric MALT might have a false negative PET/CT imaging. For patients with relevant clinical symptoms and negative PET/CT results, further diagnosis should be combined with other findings. Furthermore, two patients with MALT have significantly higher SUVmax than other patients in the present study. Because inert tumors are closely related to Helicobacter pylori infection, gastric MALT lymphoma may not only coexist with DLBCL, but also transform into DLBCL during follow-up [20, 22] (Fig. 6B). DLBCL exhibited greater FDG accumulation than other types of lymphoma [16]. Therefore, for patients with gastric MALT lymphoma with high intragastric uptake level, the possibility that biopsy samples do not include large cell parts should be considered [17]. Compared with other imaging methods, even compared with endoscopic biopsy, 18F-FDG PET/CT can minimize the possibility of DLBCL being misdiagnosed as malt lymphoma and monitor the transformation from malt lymphoma to DLBCL, because it has the advantage of overall evaluation of gastric metabolism and structural status [16, 17].
GC is increasingly recognized as a heterogeneous disease [23–25]. Nevertheless, current clinical practice does not take this heterogeneity into account, GC is considered to be a single type of malignancy, using a one size fits all approach. Although some previous studies have shown that SUVmax in PET/CT may be significantly different due to different histological subtypes, most studies still evaluated the prognostic impact of SUVmax by treating GC as a single disease entity rather than classifying it as other histological subtypes. Not surprisingly, this led to inconsistencies between studies.
There are major controversies in the diagnosis of gastric mucinous adenocarcinoma with PET/CT in recent years. Several previous studies [26, 27] have indicated that mucinous and signet ring cell adenocarcinoma and poorly-differentiated adenocarcinoma commonly exhibit significantly lower metabolic rates of FDG uptake, and this is due to the lack of expression of Glut1 transporter on the surface of gastric mucosal adenocarcinoma. The degree of tumor differentiation is an important factor affecting the measurability of primary tumors. Well differentiated or moderately differentiated adenocarcinoma is easier to measure on FDG PET/CT than poorly differentiated adenocarcinoma or signet ring cell type. The result of our study showed inconsistencies with the previous studies that reported the accumulation of more FDG in well or moderately differentiated GC than poorly differentiated tumors [28, 29]. In the present study, no significant differences were observed in the SUVmax between gastric mucinous adenocarcinoma and non-mucinous adenocarcinoma, as well as between well or moderately differentiated GC and poorly differentiated GC.
Among all the 118 patients with GC in the present study, 8 had FDG uptake equal to or even lower than that of the liver. Seven of the 8 patients were non-mucinous adenocarcinomas with moderate to poor differentiation, and only 1 had low differentiation mucinous adenocarcinoma. In mucinous adenocarcinoma, the SUVmax is significantly higher than in other 5 cases, and these results were inconsistent with that of the previous studies [26–29]. Mochiki et al [30] have indicated that the SUVmax of primary GC showed significant correlation with the degree of tumor invasion. The present study found that the SUVmax of T3 + 4 GC was higher than that of T1 + 2, showing a statistically significant difference, which was consistent with the above study. These findings suggested that SUVmax was associated with T stage, independent of pathological type and degree of differentiation, and was inconsistent with that of the previous studies. This inconsistency might be due to different stages of these gastric lesions as detected by 18F-FDG PET/CT, or the lesion might be associated with inflammatory lesions.
The distribution of FDG is different in the stomach wall due to varied growth patterns of GC and GL. In this study, PET/CT of GL was divided into three types according to the degree of gastric infiltration and FDG distribution. GC originates from gastric mucosa and grows on the mucosal surface, forming a mass that protrudes into the stomach cavity. Additionally, for the 109 GC patients with FDG uptake in the current study, the incidence of type I, II, and III lesions was 11.01%, 56.88% and 32.11%, respectively. GL originates from the submucosal layer, infiltrates along the long axis of the stomach wall under the mucosa, and invades inside and outside of the cavity. The extent and thickness of the lesion are more obvious than that of GC. In 66 patients with GL, type I (36.36%) and type II (34.84%) lesions accounted for the majority, while type III lesions represented focal involvement of the stomach, accounting for only a few lesions (28.78%). When comparing the incidence rate between GL patients and GC patients, there was statistical difference between the two groups in type I and II lesions, but no difference was observed in III type lesions. These data suggest that type I lesions in 18F-FDG PET/CT mode are more common in GL patients, while type II lesions are more common in GC patients. This suggests that PET/CT pattern of GL acts as a reliable indicator for identifying GC and GL.
Because of the junction of the fundus gland and the location of the pyloric gland in the antrum of the stomach, the concentration of gastric acid and enzyme remained higher, the antrum is more susceptible than other parts, and long-term stimulation can lead to malignancy. Additionally, the submucosal tissue of the antrum is relatively tight, less active, and more susceptible to mechanical damage. In the present study, with regard to the localization of the lesions in the stomach, the cardia was shown to be less involved and the antrum was shown to be more involved in GL and GC patients, and the fundus and body were less involved in GC patients than in GL patients. Moreover, the incidence of gastric involvement in GL patients was higher than that in GC patients. These results suggest that GL is more likely to infiltrate the gastric wall, while GC is more localized.
In the previous study [31], GL more likely invaded the retroperitoneal sub-renal lymph nodes and splenomegaly. In the present study, a total of 6 patients with GL had splenomegaly, and none of the patients with GC had splenomegaly. Nineteen patients with GL involved sub-lymph nodes, and only 7 cases with GC. Splenomegaly or sub-renal lymph node enlargement might provide additional clues in diagnosing GL.