A New Prognostic Index In Diffuse Large B-Cell Lymphoma Using Serum Albumin: A Pilot Study Evaluating The Albumin Adjusted-International Prognostic Index (A-IPI)

Low serum albumin (SA) has been identified as a prognostic marker in multiple hematological malignancies including diffuse large B-cell lymphoma (DLBCL). Low SA may be an indicator of worsening disease biology, increased inflammation, or increased co-morbidities. Changing demographics and improved care of the elderly may abrogate some of the risk attributable to age. SA, however, may not be influenced by such changes. Currently the Revised International Prognostic Index (R-IPI), the currently best accepted model for prognosis in DLBCL, does not account for albumin. We therefore explored the use of albumin as a prognostic variable in conjunction with the covariates included in the R-IPI.

Patients with DLBCL treated between 2007-2010 with RCHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone) were retrospectively identified using the Moffitt Total Cancer Care platform. Age, ECOG performance status, LDH, extrandoal sites of disease, stage, and albumin were collected and analyzed using Random Forest Modeling. Both age and extranodal involvement failed to retain statistical significance and were dropped from the model, leading to the generation of a new algorithm, the Albumin Adjusted IPI (A-IPI). This score assigned a point for SA<3.7g/dL, LDH >the upper limit of normal, ECOG performance status >=2, and Ann Arbor stage III-IV. Progression free and overall survival was compared by the risk groups using Kaplan-Meier curves along with the log rank test. Statistical analysis was done using R statistical software.

A total of 124 patients were identified. Median age was 58 years with 62% male. 46% were over the age of 60 at diagnosis and 63% were Ann Arbor stage 3 or 4. The A-IPI score identified three groups of patients (Very good: A-IPI=0, Good: A-IPI=1-2, Poor: A-IPI=3-4), similar to the R-IPI. PFS and OS at 4 years were compared using the A-IPI and R-IPI (Table 1).

In comparison to the R-IPI, the A-IPI appears to better define poor risk patients while retaining the ability to discriminate well between low (“Very Good”) and intermediate (“Good”) risk groups. Confirmation in a larger and prospective cohort is indicated, however, these preliminary data suggest SA may be a better surrogate for co-morbid status, pro-inflammatory states, and worse disease biology than age in patients with DLBCL.

No relevant conflicts of interest to declare.