Abstract
Abstract 3104
Diffuse large B-cell lymphoma (DLBCL) is the most common sub-type of non-Hodgkin's lymphoma.1 DLBCL is a heterogeneous, clinically aggressive disease, which has recently been sub-categorized based upon gene expression profiling. The prognosis of patients with DLBCL is presently assessed by the International Prognostic Index (IPI), which predicts estimated five-year survival based on clinical criteria.1,2
Over the past decade, a number of reports have established that the intra-tumoral content of the phospholipid-related metabolites phosphoethanolamine and phosphocholine (Etn-P and Cho-P) is increased in clinically aggressive malignant disease, and decreases when the malignancy responds to anticancer treatment. These data suggest that phospholipid metabolism is an intrinsic part of the disease process.3,4 Based on these observations, we hypothesized that intra-tumor levels of Etn-P and Cho-P may be a reliable predictive biomarker for therapeutic outcome in aggressive malignant diseases. To test this hypothesis, we measured noninvasively the tumor content of the sum of Etn-P plus Cho-P normalized by the tumor content of nucleoside triphosphates ([Etn-P+Cho-P]/NTP) in patients with DLBCL prior to initiating therapy. These data show that the pretreatment metabolic ratio (PMR) of [Etn-P+Cho-P]/NTP, is a potentially valuable biomarker of outcome, which identifies a subset of DLBCL patients likely to experience early failure to standard therapy, and for whom alternatives to therapy should be considered.
Under ethical review board approval, 27 previously untreated DLBCL patients prior to receiving standard doxorubicin-based therapy were studied noninvasively using 3D localized, 1H-decoupled, nuclear Overhauser-enhanced phosphorus MR spectroscopy at 1.5 T.
As only complete response (CR) is clinically meaningful in achieving durable remissions of DLBCL, we divided these patients by response at six months into CR and all the other responses (not complete response, NCR). In the 17 CR patients, the PMR was significantly lower than in the ten NCR patients (PMR mean ± 95% CI, 1.46 ± 0.21 vs. 2.47 ± 0.24, p < 3 × 10-6). The prediction of response using a Fisher test was significant for the PMR alone (p < 1 × 10-4; sensitivity of 1.00, specificity of 0.70) and improved further when combined with the IPI (p < 2 × 10-6; sensitivity of 1.0, specificity of 0.90). The progression-free survival (PFS) strongly correlated with the PMR alone (p < 4 × 10-8) and with the PMR and IPI as covariates (p < 1 × 10-7) by the Cox regression model. Using the Kaplan-Meier model, the PMR discriminated 64% of patients with PFS below 11 months (p < 1 × 10-7), while as covariate with the IPI it discriminated 82% of these patients (p < 2 × 10-7).
Our results show that the PMR contains information that predicts outcome to standard therapy in DLBCL patients. This prediction improves when the PMR is combined with the IPI. While the PME alone identifies 64% of those patients who will go on to show an extremely poor response with a progression-free survival below 11 months, the combination of the PME and IPI identify 82% of these patients. In conclusion, we have successfully demonstrated that the PMR can be an important determinant in predicting response to treatment in patients with DLBCL, especially when integrated with the IPI.
1) Cancer: principles and practice of oncology. 6th ed. Philadelphia, PA, Lippincott, Williams & Wilkins, 2001; 2) Smith MR, Non-Hodgkin's lymphoma. St. Louis, MO, Mosby, 1966; 3) Arias-Mendoza F, Brown TR, Dis. Markers, 2003;19:49-68; 4)Griffiths JR, et al, Lancet 1983;1:1435-6
No relevant conflicts of interest to declare.
Author notes
Asterisk with author names denotes non-ASH members.
