Senior-IPI: An Easily Applicable and Meaningful Prognostic Index for First-Line LBCL Patients Older Than 80 Years Treated with Age-Adapted Immunochemotherapy

Rationale

The number of people > 80 years old increased threefold between 1980 and 2010 (Wan He et al, Int Pop Reports 2016) and Large B Cell Lymphoma (LBCL) is expected to become more prevalent in this aging population. Nevertheless, clinical trials include very few patients > 80 years old (Beygi et al, Leuk Lymphoma 2018).

Efforts have been made to elaborate specific prognostic indexes for older LBCL patients but included patients with variable ages (≥ 65 years old), heterogeneous treatments (from palliative care to full-dose immunochemotherapy) and geriatric scales too complex to be widely used in clinical practice (Advani et al, Br J Haematology 2010; Merli et al, JCO 2021; Isaksen et al, Haematologica 2023). To date, there is no specific prognostic index dedicated to LBCL > 80 years old treated in first-line with gold-standard antiCD20 + miniCHOP (cyclophosphamide, doxorubicin, vincristine, and prednisone) regimen.

We defined the SENIOR-IPI, an internationally applicable, straightforward prognostic index for first-line LBLCL patients > 80 years old treated with antiCD20-miniCHOP.

Methods

The french LYmphoma Study Association (LYSA) has organized three clinical trials in first-line LBCL patients > 80 years old: LNH 03-7B using Rituximab-miniCHOP (Peyrade et al Lancet Oncol 2011), LNH09-7B using Obinutuzumab-miniCHOP (Peyrade et al Lancet Haematol 2017) and SENIOR evaluating R-miniCHOP versus R-miniCHOP + lenalidomide (Oberic et al, JCO 2021). Pooled together, 396 LBCL patients > 80 years old with homogeneous first-line treatment having completed at least one treatment cycle were included in the training set.

Prognostic value of candidate predictors for SENIOR-IPI were fitted and evaluated by Cox proportional hazard models on overall survival (OS) and progression free survival (PFS). In the training set, analyses were stratified by protocol and treatment arm to take heterogeneity of treatment arms and recruitment periods into account. Variable selection was performed on 100 datasets where missing data were imputed by multiple imputation chain algorithm.

C-index was considered as the primary criterion of comparison between candidate prognostic models. Restricted mean survival times (RMST) to 2 or 5 years were also evaluated by the pseudo-observation approach.

The external validation set included 140 patients > 80 years old with available data for SENIOR-IPI score from the prospective real-world REALYSA cohort (NCT03869619).

Results

In the training cohort, NCCN-IPI performed better than IPI and aaIPI for OS (C-index 0.621 versus 0.6) and PFS (C-index 0.604 versus 0.59). The superiority of NCCN-IPI was linked to the LDH > 3N threshold and not the number of extranodal sites, and an intermediate model including aaIPI+LDH>3N proved superior to the other indexes. Next, including continuous age value had the best additional prognostic value compared to aaIPI+LDH>3N. Finally, among candidate predictors with prognostic value in univariate analysis, only albumin < 35g/L was identified as having additional prognostic value (C-index 0.693, p value 2.61x10^-3).

Thus, SENIOR-IPI was defined as the sum of: number of years over 80 years old + number of aaIPI predictors + LDH>3N + albumin<35g/L, with weighting coefficients of 6 based on the Cox model coefficients. Overall, the index varied between 0 and 40, suggesting patient segmentation in 4 classes to facilitate reproducibility: 0-<10; 19.4%, 10-<20; 40.4%, 20-<30; 31.8%, and ≥30; 8.3%.

In the training set, SENIOR-IPI performed better than IPI, aaIPI and NCCN-IPI for OS (C-index 0.645 versus 0.6, 0.6 and 0.621 respectively) and for PFS (C-index 0.637 versus 0.59, 0.59 and 0.604 respectively). This was confirmed in the validation set, showing better performance of SENIOR-IPI than IPI, aaIPI and NCCN-IPI for OS (C-index 0.685 versus 0.593, 0.636 and 0.599 respectively) and for PFS (C-index 0.667 versus 0.618, 0.655 and 0.621 respectively).

RMST to 2 and 5 years showed that 10 additional SENIOR-IPI points led to 3.45 months and 10.5 months of survival lost respectively in the training set and 3.6 months and 11.85 months of survival lost respectively in the validation set.

Conclusions

SENIOR-IPI is an easily applicable prognostic index for LBCL patients > 80 years old fit for age-adapted immunochemotherapy, which has been validated in an external cohort. SENIOR-IPI better stratified prognosis of these patients as compared to IPI, aaIPI and NCCN-IPI scores.