Is NCCN-IPI a Better Prognostic Tool for DLBCL Treated With R-CHOP?

Compared with the widely used International Prognostic Index (IPI) and revised IPI (R-IPI), the most recent National Comprehensive Cancer Network IPI (NCCN-IPI) better predicts survival of patients with diffuse large B-cell lymphoma (DLBCL) who are treated with R-CHOP (rituximab, cyclophosphamide, doxorubicin hydrochloride, vincristine sulfate, and prednisone), according to a study published in Blood.

Despite these findings, the authors of the study advocate for the continued use of the IPI, as this scoring system represents an acceptable prognostic tool to guide treatment decisions for most patients and allows comparison with numerous clinical trials that have used the IPI.

The investigators, led by biostatistician Amy Ruppert, PhD, from The Ohio State University, analyzed patient data from 7 multicenter randomized clinical trials, including:

  • LNH03-2B: ACVBP (doxorubicin, cyclophosphamide, vindesine, bleomycin, and prednisone) plus rituximab versus R-CHOP in patients with DLBCL and age-adjusted IPI
  • LNH03-6B: R-CHOP-14 (biweekly) vs. R-CHOP-21 (triweekly) and darbepoetin alpha in patients with DLBCL between ages 60-80
  • LNH-985: R-CHOP vs. CHOP in elderly patients with DLBCL
  • MegaCHOEP: conventional chemotherapy vs. high-dose chemotherapy with rituximab
  • MinT: combination chemotherapy with or without rituximab in non-Hodgkin lymphoma
  • RICOVER-60: CHOP-14 with or without rituximab in elderly patients with aggressive CD20+ B-cell lymphomas
  • University College London trial: R-CHOP in newly diagnosed DLBCL non-Hodgkin lymphoma

The analysis included a total of 2,124 patients with newly diagnosed DLBCL from whom data were available to calculate the IPI, R-IPI, and NCCN-IPI scoring systems. All participants received R-CHOP or variant as induction therapy and were treated between 1998 to 2009.

The median age of the pooled cohort was 63 years (range = 18-83), with slightly more than one-half of patients (56%) aged >60 years. Approximately 84% of patients had an Eastern Cooperative Oncology Group (ECOG) performance status score of ≤1, and 62% of patients had Ann Arbor stage III/IV disease. In 22% of patients, extranodal involvement was documented in >1 site. Around 59% of patients had elevated lactate dehydrogenase (LDH), with 10% demonstrating LDH measures >3 times the upper limit of normal.

Using the IPI scoring system, 34% of patients were considered low risk, 23% were low-intermediate risk, 23% were high-intermediate risk, and 20% of patients were considered high risk. The updated R-IPI scoring system categorized 9% of patients as very good risk, 48% as good risk, and 43% as poor risk. The NCCN-IPI scoring system classified 13% of patients as low risk, 41% as low-intermediate risk, 36% as high-intermediate risk, and 10% as high risk. (See TABLE.)

The authors advocate for the continued use of the IPI, as it represents an acceptable tool to guide treatment decisions for most patients and allows comparison with numerous clinical trials that have used the IPI.

During a median follow-up period of 4.9 years and a maximum of 10.7 years, a total of 559 deaths were reported. The median OS has not been reached, but the 5-year OS estimate for the pooled cohort was 73% (95% CI 71-75). Each of the risk scoring systems had significantly different estimated OS values (p<0.0001 for each).

The ranges in the estimated 5-year OS rates for each scoring system were:

  • IPI: 54% to 88%
  • R-IPI: 61% to 93%
  • NCCN-IPI: 49% to 92%

The NCCN-IPI distinguished a subgroup of patients with favorable long-term survival and improved on the R-IPI by identifying a less heterogenous group of high-risk patients.

According to models stratified by type of induction therapy, the NCCN-IPI provided the best discrimination between patients with poor and favorable OS compared with the IPI and R-IPI (c-indexes: 0.632 vs. 0.626 and 0.590, respectively).

The 5-year estimated PFS was 65% (95% CI 63-67). The NCCN-IPI also best discriminated outcomes versus the IPI and R-IPI. In addition, the NCCN-IPI featured the highest c-index as well as the lowest Akaike information criterion scores for PFS compared with the other two risk scores.

Limitations of the risk scoring systems in this study include their inability to identify patients at risk of very poor survival. The authors mention that these three scoring systems were developed before high-risk DLBCL was reclassified based on the presence of MYC and BCL2 or BCL6 rearrangements. “It is unclear how the ability to identify poorer performing patients might be impacted if these patients were not included as DLBCL,” the authors wrote. “This study could not properly address this question since information regarding which of these patients carry a poor prognosis, namely those with double hit disease in which MYC is translocated to an immunoglobulin partner, has been specified only recently.”

In addition, the authors suggest that even better prognostic scores are needed, particularly scores that can risk-stratify patients and determine those patients at very high risk of relapse and death who are most in need of novel treatments. Scoring systems that integrate molecular tests and other tumor features may aid in identifying such patients.


Ruppert AS, Dixon JG, Salles GA, et al. International prognostic indices in diffuse large B-cell lymphoma (DLBCL): a comparison of IPI, R-IPI and NCCN-IPI. Blood. 2020 Mar 13. [Epub ahead of print]