We undertook this 10 year retrospective study to evaluate the clinicopathologic and molecular features of GI lymphomas at a major US tertiary care medical center to determine the features of GI lymphomas in North America in comparison to other non-North American lymphoma cohorts. No previous extensive US study describing the primary and secondary GI lymphomas presenting in a Northern American population has been published that utilizes the current World Health Organization lymphoma classification. We show that there are distinct differences and similarities in our cases compared to other population based studies. Although many types of GI lymphomas were diagnosed, most of the cases presenting to our facility were referral cases initially diagnosed as lymphoma at an outside facility which were later referred to our facility for additional treatment. Thus, many of these patients had been treated with many different therapeutic regimens, so treatment regimens varied greatly among our cases and were too heterogenous to correlate treatments with clinical outcome in our analysis.
Several series from Europe, Canada-North America, and Asia have reported on primary GI lymphomas by lymphoma type and anatomic site
[2–5, 7, 9, 28–30]. Several trends emerge when reviewing the data collectively. First, primary gastric lymphomas tended to be more common than primary intestinal lymphomas in the European and Asian studies. Intestinal cases accounted for only 14%, 19%, 21%, 10%, 30%, 34%, 50% and 43% of cases in the Greek, Japanese, German, Austrian, Serbian, Turkish, Canadian-North American and Danish studies, respectively
[2–5, 7, 9, 28–30]. Gastric lymphomas accounted for >50% of total cases in most of these series. Our study stands in contrast to these, in that intestinal cases out-number gastric cases, accounting for 54% of total primary GI lymphomas. This difference noted in our study could be due to lower prevalence rates of HP infection in the US resulting in fewer gastric MALTs diagnoses, earlier treatment for HP infection resulting in fewer cases progressing to gastric MALTs, or that US patients with intestinal involvement have more severe symptoms such as intestinal obstruction and/or easier health care access resulting in more frequent visits to our tertiary care surgery facilities than patients with only solitary gastric involvement. The second trend to emerge is that high grade gastric lymphomas (i.e. gastric DLBCLs) tend to occur in approximately equal proportion to low grade gastric lymphomas (i.e. gastric MALTs). In this regard, our data are consistent with the European and Asian studies. Third, MALTs are much more common in the stomach than other sites, while DLBCLs tend to affect all segments of the GI tract. Our series also shows this trend to hold true.
Our study is also consistent with a recent study of primary GI FL from France by Damaj et al., which showed that primary FL of the GI tract tends to be a disease of the small intestine
. Similar findings were reported in a recent large US study which revealed small intestinal FLs made up 63% of cases while gastric FLs accounted for only 3% of primary gastrointestinal FLs
. Similarly, 14 of the 25 (56%) of primary FLs in our series were found in the small intestine. No other low grade B lymphoma type in our series showed such a strong propensity for small intestinal involvement. In contrast to the study by Damaj et al., which showed a female predominance, the small intestinal FLs in our series were equally distributed among men and women.
In examining the stage of presentation in our gastric MALTs and gastric DBLCLs, we also noted that gastric DLBCLs tended to have higher stage disease at presentation, while gastric MALTs often showed lower stage disease at presentation. Previous studies have shown that stage of lymphoma is an important prognostic predictor of clinical outcome, and that the lower grade gastric MALTs tend to have low stage while higher grade DLBCLs tend to have higher stage at presentation
[3, 4, 28, 33]. Thus, our study reveals that this association also occurs in the lymphomas from a Central North American population.
Environmental factors, especially Helicobacter pylori (HP) infection, have been noted to play a prominent role in the development of gastric lymphomas
. In 2 Italian studies, HP has been reported to occur in 88% of low grade gastric MALTs identified by histologic examination, while it occurred at lower frequency (52-63%) in high grade gastric lymphomas
[28, 35]. Interestingly, both Italian studies revealed a statistically significant association between HP infection and low grade gastric lymphoma (p < 0.0001)
[28, 35]. Similarly, a serologic study performed at a New York center showed HP seroprevalence up to 67% in gastric MALTs, and increased seropositive rates were associated with increased age and country of birth outside the US or Canada (p = 0.0001)
. Interestingly, HP infection in our study was seen in only 24% of the gastric MALTs and 9% of the gastric DLBCLs by positive serology or histology, and we also noted higher frequency of HP infection in the low grade gastric MALTs compared to high grade gastric DLBCLs. A 2008 Canadian-North American study also showed HP infection in 20% of gastrointestinal non-Hodgkin lymphomas, of which 44% were in MALTs and 13% were in DLBCLs
. The less frequent HP infection rates in our cases could relate to our patient population having a higher percentage of patients born in the US or Canada, differences in the ethnicities of the Central-Midwestern compared to the New York population, or lower prevalence of HP infection in the Midwest. Alternatively, the absence of serologic testing in all our cases, which may be more sensitive than histology in detecting HP infections, could also contribute to a lower HP prevalence rate in our gastric MALT patients.
Because of the high prevalence of genetic defects in p16/CDNK2A in non-Hodgkin B cell lymphomas, we also evaluated P16 expression in our gastric MALTs and DLBCLs. Previous studies have shown that P16 is underexpressed more frequently in gastric MALT than in other lymphoma types
[20, 21]. This has been shown to primarily relate to p16 promoter hypermethylation in non-transformed MALTs
[20, 36, 37]. As such, p16 hypermethylation, and the resulting loss of P16 protein, is a high-frequency event in gastric MALTs occurring in up to 75% to 79% of cases in some studies
[20–22, 38, 39]. One Korean study by Min et al. also showed increased frequency of p16 hypermethylation in gastric DLBCLs (N = 11, 72.7%) compared to gastric MALTs (N = 24, 41.7%) and proposed that malignant transformation of gastric MALTs was attributed in part to p16 hypermethylation
. Our results confirm the prior findings that p16 is frequently inactivated in gastric MALTs, since P16 expression was absent in 11 of 13 (85%) gastric MALTs, and expression was scored as 0 (i.e. less 5% of tumor cells positive) in these cases. In addition, P16 was expressed in only 3 of the 6 (50%) gastric DLBCLs. Most p16 hypermethylation has been previously seen in HP dependent cases. This fact may explain the weak P16 expression seen in one HP + gastritis case we tested, but does not provide a mechanism for P16 loss in the our HP- gastritis cases and HP- gastric MALTs. Although the overall numbers of cases analyzed for P16 expression is low, our study showed no statistically significant difference was observed in P16 expression between gastric DLBCL and gastric MALT (p = 0.1078). Our result differs from the Korean study by Min et al. that showed hypermethylation of p16 (which can consequently downregulate P16 expression) to be more frequently associated with gastric DLBCLs compared to low grade gastric MALTs
. Since the number cases in our study is only half as many as studied by Min et al., other studies would be needed to determine if this association between p16 hypermethylation (P16 expression loss) in high grade gastric DLBCLs holds true in other populations.
P53 over-expression has been shown to correlate with mutations in TP53 . This may relate to decreased rate of degradation of the mutant protein, leading to elevated cytoplasmic accumulation. P53 immunohistochemistry may be useful in evaluating primary GI lymphomas, because P53 accumulation has been shown to be more common in high grade lymphomas than low grade lymphomas
. Moller et al. showed that P53 protein overexpression, defined as ≥ 20% of tumor cells positive for P53, was 80-90% sensitive and 100% specific in predicting TP53 mutations in DLBCLs
. Moller et al. also reported that p53 overexpression was an independent predictor of poor outcome in both T and B non-Hodgkin lymphomas (n = 199), and that P53 predicted poorer overall survival in indolent and aggressive non-Hodgkin lymphomas in addition to being associated with treatment failure and relapse-free survival. Using the same scoring criteria, high grade gastric DLBCLs were more significantly associated with P53 overexpression than low grade gastric MALTs in our series (p = 0.008). Thus our data corroborates the prior studies which have suggested mutations in TP53 may relate to progression from low grade to high grade lymphoma, as well as increased risk of relapse
Increased Ki-67 index has also been shown to correlate with increased grade of lymphoma
. Our results were consistent with these previous observations, since all of our tested gastric DLBCLs had high Ki-67 index ≥80%, while the majority (85%) of the gastric MALTs had Ki-67 ≤20%. Statistical analysis on our lymphomas also showed higher Ki-67 levels were significantly associated with high grade lymphoma, specifically gastric DLBCL (p = 0.00042). Of note, the two MALTs with the highest Ki-67 index (50%) both showed strong P53 expression and an increased admixture of large B cells. This suggests they are not typical low grade MALTs, and likely are intermediate or transitional cases between a low grade and a high grade gastric B cell lymphoma. Alternatively, since these two gastric MALTs were diagnosed from small endoscopic biopsies, undersampling of a DLBCL on a small biopsy could also explain the P53 overexpression in these two outlier cases. Identification of a high proliferative index (Ki-67 >80%) on small gastric biopsies is also a helpful feature in diagnosing DLBCL, since all low grade gastric MALTs lacking evidence of large cell transformation (or admixed large cells) tend to show lower Ki-67 levels (≤20%). Assessment of the proliferative index may be a useful feature to study in small gastric biopsies with significant crushed features where morphologic evaluation is limited, and optimal grading and subtyping of the B cell lymphomas can be challenging due to the suboptimal quality of the biopsy.
The API2-MALT1 translocation has been shown to be relatively frequent in gastric MALTs, and has been shown to predict both lack of resolution with H. pylori eradication
 and resistance to further genetic damage
[44, 45]. Perhaps unsurprisingly, this translocation has been reported to be a rare event in gastric DLBCLs
[20, 45–47], indicating a gastric MALT with this translocation rarely progress to gastric DLBCL. Our series confirm these previous findings, as none of our gastric DLBCLs showed the API2-MALT1 translocation, and none of the gastric MALTs with this translocation had concomitant HP infection. In addition, this translocation was quite frequent (38%) in the gastric MALTs in our series, similar to the 17%-40% reported in other studies
[17, 18, 30, 48, 49].
Although IGH translocations such as t(1;14)(p22;q32);BCL10-IGH are reported in 3% of gastric MALTs and 5-10% of intestinal MALTs
[17, 18], we did not detect frequent IGH translocations in this series, as our tested gastric MALTs were all negative. Only one duodenal MALT tested showed the IGH translocation. Unfortunately, due to the absence of karyotype analysis, we could not further determine the partner gene for this case. Nevertheless, our data corroborates the prior data that an IGH translocation (such as the IGH-MALT1
IGH-FOXP1), which is not the same as a clonal IGH gene rearrangement detected by polymerase chain reaction analysis, is an infrequent genetic alteration in the pathogenesis of gastric MALTs.
In summary, our study highlights the differences in distribution and subtype of GI lymphomas at a large North American medical center. Secondary GI lymphomas were less frequent than primary GI lymphomas, and there were clear differences in frequency and subtype between primary and secondary cases. We confirm that certain genetic alterations occur frequently in GI lymphomas, such as loss of P16 protein expression and API2-MALT1 translocations in gastric MALTs, and P53 overexpression in gastric DLBCLs. However, we also found differences in the subtype, molecular, and clinical features in our Central-Midwestern North American patient population compared to other non-North American series.