Is the GELTAMO-IPI more accurate than NCCN-IPI or IPI in patients with DLBCL?

In January 2017, the British Journal of Haematology published the results of a study by Carlos Montalbán from the MD Anderson Cancer Centre Madrid, Spain, and colleagues which assessed the validity of the National Comprehensive Cancer Network International Prognostic Index (NCCN-IPI) compared with IPI. Furthermore, the authors assessed the accuracy of modifying the NCCN-IPI through adding primary extranodal status, intense treatment, or high β2-microglobulin level, to be called the Grupo Español de Linfomas/Trasplantes de Médula Ósea (GELTAMO)-IPI. IPI comparisons were made using the Hosmer-Lemeshow goodness-of-fit score.

Key Highlights:

  • 1848 DLBCL pts diagnosed Jan 1998­‑Jul 2014; pts were only excluded based on absence of data for each IPI
  • Primary endpoint was 5-year OS across all measures
  • 1777 pts had complete variable data for NCCN-IPI calculation resulting in:
    • Four distinct risk groups by 5-year OS
    • Failure to identify high-risk, poor-outcome pts (substantially < 50% 5-year OS) group in this data set
  • 1764 pts had complete data for IPI calculation resulting in:
    • Four distinct risk groups by 5-year OS
    • Failure to identify high-risk, poor outcome pts (< 50% 5-year OS) group in this data set
    • Two cohorts: 2/3 training cohort (1230 pts), 1/3 validation cohort (618 pts)
    • NCCN-IPI extranodal involvement, and primary extranodal presentation were found to not have an independent prognostic effect
    • High β2-microglobulin level found to have an independent prognostic effect and therefore was added to the NCCN-IPI to make the GELTAMO-IPI
    • Risk groups defined based on scores for each prognostic marker with total maximum score of 7:
      • Low Risk (LR) = 0 points (12% pts), Lower-Intermediate Risk (LIR) = 1-3 points (57% pts), Higher-Intermediate Risk (HIR) = 4 points (18% pts), and High Risk (HR) = more than 5 points (13% pts)
    • GELTAMO series analysis:
      • GELTAMO-IPI 5-year OS: LR = 93%, LIR = 79%, HIR = 66%, HR = 39% (all P < 0.001)
      • NCCN-IPI 5-year OS: LR = 93%, LIR = 83%, HIR = 67%, HR = 49%
      • IPI 5-year OS: LR = 88%, LIR = 77%, HIR = 68%, HR = 51%
      • GELTAMO-IPI was the only IPI to identify a high risk group with below 40% 5-year OS
    • Goodness-of-fit statistical analysis:
      • IPI = lack of fit (P = 0.016) vs. NCCN-IPI = good fit (P = 0.069)
      • NCCN-IPI = good discrimination of high-risk group (P = 0.037) vs. GELTAMO-IPI = better discrimination of high-risk group (P = 0.299)

In conclusion, the authors stated that their data validates the NCCN-IPI as being more accurate than the IPI, albeit differing from the original NCCN-IPI study. Furthermore, they conclude that the GELTAMO-IPI, which includes the β2-microglobulin level, results in a more discriminatory system, especially for the high-risk group where it was able to classify a high-risk group with 5-year OS of less than 50%. Further prospective validation of the GELTAMO-IPI will need to be published.


Montalbán C. et al. Validation of the NCCN-IPI for diffuse large B-cell lymphoma (DLBCL): the addition of β2-microglobulin yields a more accurate GELTAMO-IPI. British Journal of Haematology. 2017 Jan 20. DOI: 10.1111/bjh.14489. [Epub ahead of print]


The study included 1848 diffuse large B-cell lymphoma (DLBCL) patients treated with chemotherapy/rituximab. The aims were to validate the National Comprehensive Cancer Network International Prognostic Index (NCCN-IPI) and explore the effect of adding high Beta-2 microglobulin (β2M), primary extranodal presentation and intense treatment to the NCCN-IPI variables in order to develop an improved index. Comparing survival curves, NCCN-IPI discriminated better than IPI, separating four risk groups with 5-year overall survival rates of 93%, 83%, 67% and 49%, but failing to identify a true high-risk population. For the second aim the series was split into training and validation cohorts: in the former the multivariate model identified age, lactate dehydrogenase, Eastern Cooperative Oncology Group performance status, Stage III-IV, and β2M as independently significant, whereas the NCCN-IPI-selected extranodal sites, primary extranodal presentation and intense treatments were not. These results were confirmed in the validation cohort. The Grupo Español de Linfomas/Trasplante de Médula ósea (GELTAMO)-IPI developed here, with 7 points, significantly separated four risk groups (0, 1–3, 4 or ≥5 points) with 11%, 58%, 17% and 14% of patients, and 5-year overall survival rates of 93%, 79%, 66% and 39%, respectively. In the comparison GELTAMO IPI discriminated better than the NCCN-IPI. In conclusion, GELTAMO-IPI is more accurate than the NCCN-IPI and has statistical and practical advantages in that the better discrimination identifies an authentic high-risk group and is not influenced by primary extranodal presentation or treatments of different intensity.

Download this article:

You can now download this article in Adobe PDF® format.

Download as PDF