Patients with natural killer T (NK/T) -cell lymphomas have poor survival outcome, and for this condition there is no optimal therapy. The purpose of this study was to design a prognostic model specifically for extranodal NK/T-cell lymphoma, which can identify high-risk patients who need more aggressive therapy.

This multicenter retrospective study was comprised of 262 patients who were diagnosed with NK/T-cell lymphoma.

After a median follow-up duration of 51.2 months, 5-year overall survival rate in 262 patients was 49.5%. Prognostic factors for survival were “B” symptoms (P = .0003; relative risk, 2.202; 95% CI, 1.446 to 3.353), stage (P = .0006; relative risk, 2.366; 95% CI, 1.462 to 3.828), lactate dehydrogenase (LDH) level (P = .0005; relative risk, 2.278; 95% CI, 1.442 to 3.598), and regional lymph nodes (P = .0044; relative risk, 1.546; 95% CI, 1.009 to 2.367). Of 262 patients, 219 had complete information on four parameters. We identified four different risk groups: group 1, no adverse factor; group 2, one factor; group 3, two factors; and group 4, three or four factors. The new model showed a superior prognostic discrimination as compared with the International Prognostic Index (IPI). Notably, the distribution of patients was balanced when a new model was adopted (group 1, 27%; group 2, 31%; group 3, 20%; group 4, 22%), whereas 81% of patients were categorized as low or low-intermediate risks using IPI.

The newly proposed model for extranodal NK/T-cell lymphoma demonstrated a more balanced distribution of patients into four groups with better prognostic discrimination as compared with the IPI.