Gastric Diffuse Large B-Cell Lymphoma: Analyses of Prognostic Factors and Value of Pretreatment FDG-PET Scan.

Up to 40% of patients with DLBCL present with extranodal involvement of lymphoma, among which stomach is a common site. Patients with gastric DLBCL are generally treated with chemotherapy with or without radiotherapy similar to other DLBCL, but the disease may carry specific biological features. We herein present our experience with gastric DLBCL, and attempted to identify potential prognostic factors. Furthermore, the value of pretreatment PET scan was analyzed in this population, since CT scan is generally not sensitive in detecting gastrointestinal lesions.

We reviewed patients with gastric DLBCL treated with CHOP or R-CHOP based therapy in Aichi Cancer Center Hospital from 1995 to 2008. Multiple characteristics were evaluated for their prognostic values for complete response rate (CRR), overall survival (OS) and progression free survival (PFS). We also calculated the sensitivity of PET scan in detection of gastric involvement of lymphoma.

Seventy-five patients with gastric DLBCL were identified and analyzed. The median age of patients was 66 (range 21-87). Thirty-five patients (47%) had Lugano stage I or II1, defined here as limited disease. International prognostic index (IPI) was low in 39 patients (52%), low-intermediate in 17 (23%), high-intermediate in 7 (9%) and high in 12 (16%). The CRR after initial treatment was 83%, and 5-year overall survival (OS) rate was 72% (median follow up duration was 32months). In 35 patients with limited disease, 25 patients (71%) underwent three courses of CHOP or R-CHOP followed by involved field radiation. Eight patients (23%) underwent gastrectomy (curative intent [n=6], control of bleeding [n=1] or obstruction [n=1],) followed by CHOP or R-CHOP. Two of those with limited disease (avoided radiation due to large primary lesion [n=2]) and all with advanced stage (Lugano II2 or IV; n=40) were treated with 6 to 8 cycles of CHOP with or without rituximab. Multivariate analysis with logistic regression model for CRR revealed that rituximab (odds ratio [OR] 0.1, p=0.007), Hb>12.0g/dl (OR 0.11, p=0.004) were independently associated with higher CRR. Multivariate analysis with Cox proportional hazard model for OS revealed that treatment without rituximab (hazard ratio [HR] 3.27, p=0.027), Hb<12.0g/dl (HR 4.06, p=0.006), and serum albumin level < lower limit of normal range (HR 3.43, p=0.023) were independently associated with shorter OS. An analysis for PFS revealed that treatment without rituximab (HR 2.73, p=0.03), advanced stage (HR 3.36, p=0.006), Hb<12.0g/dl (HR 3.61, p=0.004), and serum albumin level < lower limit of normal range (HR 2.96, p=0.021) were independently associated with shorter PFS.

In the initial work up of patients with newly diagnosed DLBCL in general, upper gastrointestinal endoscopy has been a part of staging studies in our institution. PET scan was introduced as a part of pretreatment staging studies in 2003. Gastric lymphoma was diagnosed by gastroscopy performed for screening purposes or digestive symptoms (i.e. DLBCL was diagnosed by biopsy of gastric tumor, group A) in 52 patients (69%), by gastroscopy performed as a part of staging evaluations for proven lymphoma by biopsies of other site in 17 patients (23%, group B) and the diagnostic process was not well documented in 6 (8%). Twenty-two patients in group A underwent pretreatment PET scans, among which five did not show any abnormalities (sensitivity 77%). All these 5 had small stage I disease originally detected by screening gastroscopy. Eleven patients in group B underwent pretreatment PET scan, which all showed gastric lesions (sensitivity: 100%).

We showed the survival benefit of rituximab in gastric DLBCL. Furthermore, hemoglobin and albumin levels were determined to be potential prognostic factors in these patients. While PET scan seems reasonably sensitive to detect gastric involvement of lymphoma as a part of staging evaluation, small stage I gastric DLBCL can be missed by PET scan. Larger scale studies are needed to validate the prognostic factors identified here, and more data is needed to determine the value of pretreatment PET scan in this patient group.

No relevant conflicts of interest to declare.