FLIPI-2Follicular Lymphoma Prognostic Index
Demographics & Serology
Tumor Burden & Staging
Awaiting Clinical Inputs
The FLIPI-2 is the updated clinical standard for risk-stratifying patients with follicular lymphoma.
Risk Tiers
- 0 Points: Low Risk
- 1-2 Points: Intermediate
- >=3 Points: High Risk
Guidelines & Evidence
Verified
Last Review: 2026-07-17
When to Use
When to Use
Initial prognostic assessment of patients with newly diagnosed, untreated follicular lymphoma (FL)
Risk stratification in the immunochemotherapy era — FLIPI2 was specifically developed and validated in rituximab-treated populations
To guide rational selection of patients who may benefit from more intensified treatment regimens
To identify low-risk patients within the FLIPI low-risk group who may still have meaningfully different PFS outcomes
Applicable across patients receiving first-line R-chemotherapy (R-CHOP, R-CVP, fludarabine-based regimens)
Can be used alongside PRIMA-PI for patients receiving first-line immunochemotherapy; the two indices are complementary
Patient Population
Derived from 832 patients with complete data in the prospective international F2 study (Federico et al., 2009), registered across 69 European and American institutions between January 2003 and May 2005. Validated externally in an independent Italian cohort of 231 patients (Federico et al., 2009) and in a further independent series of 280 patients by Arcaini et al. (2010). Covers FL grade 1–3a; not validated for grade 3b or concurrent transformation to DLBCL.
Special Population: Elderly Patients (≥ 60 years)
Because age ≥ 60 years is itself a FLIPI2 risk factor, all elderly patients (≥ 60) are automatically excluded from the low-risk FLIPI2 group and can only be classified as intermediate or high risk. Wang et al. (Heliyon 2025) developed the Age-adjusted FLIPI2 (A-FLIPI2), which re-sets the age threshold to ≥ 70 years, restoring three-tier stratification in patients aged ≥ 60. A-FLIPI2 demonstrated superior AUC for death prediction (0.793 vs 0.704 for FLIPI2) and disease progression (0.678 vs 0.620) in 128 elderly FL patients, with external validation in a further 67 cases.
When Not to Rely on FLIPI2 Alone
FL grade 3b — behaves more like aggressive lymphoma; use DLBCL-specific indices (IPI)
Concurrent transformation to diffuse large B-cell lymphoma at diagnosis — not captured by FLIPI2
Watch-and-wait policy — PFS events are not comparable to treated patients; FLIPI2 was built on treated populations only
Do not use FLIPI2 to defer treatment in a high-risk patient if clinical judgement indicates need for immediate therapy
Elderly patients (≥ 60): consider A-FLIPI2 for more meaningful three-tier stratification in this cohort
How it Works
The Five Scoring Variables
FLIPI2 is calculated by summing the number of adverse factors present. Each factor is weighted equally — one point each — for a maximum score of 5. The five variables were selected by bootstrap resampling procedures with backward elimination from 11 candidate variables in a multivariate Cox proportional hazard model on 832 patients with complete data (event/variable ratio 24:1). The final five factors were validated as robust independent predictors of progression-free survival (PFS).
Variable Definitions
01
β2-microglobulin (B2M) > upper limit of normal (ULN): The dominant predictor in the bootstrap procedure — the highest inclusion frequency of any variable. B2M is a component of MHC class I molecules shed from lymphoma cells proportionally to tumour proliferation and burden. Its independent prognostic value had been recognised since the 1980s in myeloma and CLL, and it had shown high univariate significance in the original FLIPI dataset but was excluded due to missing data. FLIPI2 formally incorporates it for the first time in FL-specific scoring.
02
Longest diameter of the largest involved node (LoDLIN) > 6 cm: A surrogate for tumour burden (TB). TB was calculated as nodal volume (formula 0.5236 × ØMax³ × 0.6) but was laborious to derive. LoDLIN strongly correlates with log(TB) with R² = 0.933. A TB of 50 mL — the inflection point identified by Cox spline analysis — corresponds to a LoDLIN of approximately 6 cm at the upper 95% CI limit. In practice, 97% of patients were classified identically by LoDLIN and TB (κ = 0.954). LoDLIN replaces the number of nodal sites used in FLIPI.
03
Bone marrow involvement (BMI): Established by bone marrow biopsy. Reflects disseminated disease and impaired haematopoietic reserve. FLIPI2 includes BMI as a binary variable (present/absent). The prognostic relevance of BMI in indolent lymphoma was well established prior to the F2 study; approximately 40% of the F2 population had BMI at baseline.
04
Haemoglobin (Hb) < 120 g/L (< 12 g/dL): Reflects bone marrow infiltration, systemic disease burden, and impaired physiological reserve. Together with age, haemoglobin was the only FLIPI variable retained in the FLIPI2 model, underscoring its consistent prognostic importance across eras.
05
Age > 60 years: Captures the interaction of biological age with disease biology and treatment tolerance. Multivariate hazard ratio 1.41 (SE 0.18, P = 0.008) in the F2 dataset. Older patients received less intensive regimens and had inferior clinical outcomes across both FLIPI and FLIPI2 development cohorts.
Risk Group Classification
01
Low risk (score 0): 20% of F2 population. 3-year PFS 91%; 5-year PFS 79.5%; 3-year OS 99%; 5-year OS ~98%.
02
Intermediate risk (score 1–2): 53% of F2 population. 3-year PFS 69%; 5-year PFS 51.2%; 3-year OS 96%; 5-year OS ~88%.
03
High risk (score 3–5): 27% of F2 population. 3-year PFS 51%; 5-year PFS 18.8%; 3-year OS 84%; 5-year OS ~77%.
Comparison with FLIPI
FLIPI uses age > 60, Ann Arbor stage III–IV, Hb < 120 g/L, number of nodal sites > 4, and serum LDH > ULN. The critical differences are: (1) B2M replaces LDH — B2M carries higher independent prognostic weight and was excluded from FLIPI only due to missing data; (2) LoDLIN replaces number of nodal sites — simpler to measure, equivalent discriminatory power; (3) BMI is newly added. Harrell C statistics consistently favour FLIPI2 over FLIPI, including in the independent validation by Arcaini et al. (2010), who found FLIPI2 superior and noted it could further stratify patients who would appear low-risk under FLIPI.
Clinical Pearls
Advantages Over FLIPI in the Rituximab Era
FLIPI was built retrospectively from archived data collected before rituximab became standard. Its component LDH and nodal site count performed less well in rituximab-treated patients. FLIPI2 was prospectively constructed during the rituximab era and explicitly validated in rituximab-containing regimens: in the F2 dataset, 3-year PFS by FLIPI2 in rituximab-treated patients was 89%, 73%, and 57% for low, intermediate, and high risk respectively (P < 0.001). This confirms the index is not rendered obsolete by immunochemotherapy.
Stratification Within FLIPI Low-Risk Patients
A clinically important finding from Arcaini et al. (2010): among 139 patients classified as FLIPI low-risk, FLIPI2 identified three statistically distinct PFS subgroups (P = 0.009 by Gehan's Wilcoxon test). This is of direct practical relevance — a patient in the FLIPI low-risk category with elevated B2M may have a substantially shorter PFS than their FLIPI score implies. FLIPI2 provides a more granular prognosis in this group.
Limitations
B2M is not always available in resource-limited settings or as part of routine workup — this was the primary reason it was excluded from FLIPI originally
The score uses OS as a secondary endpoint only; PFS is the principal endpoint. In the F2 study, the 5-year OS was 89% rather than the assumed 70%, forcing a switch from OS to PFS mid-study
Performance in very elderly patients (≥ 70) is limited — age is a dichotomous variable at 60 years, so patients aged 65 and 85 receive the same score. A-FLIPI2 addresses this by using age ≥ 70 as the threshold
PRIMA-PI (β2-microglobulin + bone marrow involvement only) is simpler and performs comparably for first-line immunochemotherapy patients, though it does not achieve three-tier stratification in elderly populations
Not validated for watch-and-wait patients; 115 WW patients were excluded from F2 model development due to incomparable PFS event definitions
Biological factors (tumour microenvironment, macrophage content, molecular signatures) are not captured but are under active investigation and may complement FLIPI2 in future models
A-FLIPI2 for Elderly Patients
Wang et al. (Heliyon 2025) demonstrated that in 128 FL patients aged ≥ 60, FLIPI and PRIMA-PI had poor discrimination. FLIPI classified 71.9% as high-risk — an over-classification that reduces clinical utility. Standard FLIPI2 stratified patients into only two groups (intermediate and high) because age ≥ 60 is universally present. A-FLIPI2 modifies the age threshold to ≥ 70 years, preserving the remaining four FLIPI2 variables, and successfully restores three-tier stratification. In the training cohort: 3-year PFS 77%/55%/26% and 5-year OS 91%/72%/38% for low/intermediate/high risk respectively. Validated in an external cohort of 67 patients with comparable stratification.
Next Steps
Interpreting the Result
01
Low risk (score 0): Excellent prognosis. 5-year PFS approximately 79%, 5-year OS approximately 98%. Standard first-line immunochemotherapy is appropriate. No indication for escalated intensity based on score alone.
02
Intermediate risk (score 1–2): Comprises the majority of FL patients (53%). 5-year PFS approximately 51%, 5-year OS approximately 88%. Standard immunochemotherapy with R-maintenance is appropriate. Consider enrolment in clinical trials of novel agents. Monitor closely for early progression (POD24).
03
High risk (score 3–5): 5-year PFS approximately 19%, 5-year OS approximately 77%. Higher probability of requiring early re-treatment. These patients may benefit from intensified regimens or novel agents in clinical trial settings. POD24 risk is significantly elevated and warrants close surveillance.
POD24 and Post-FLIPI2 Risk
Progression of disease within 24 months of diagnosis (POD24) is the strongest predictor of subsequent OS in FL, regardless of FLIPI2 score. High-risk FLIPI2 patients have substantially elevated POD24 rates. However, POD24 is a dynamic post-treatment endpoint — FLIPI2 is a baseline pre-treatment tool. Integrating both is essential for ongoing risk assessment and treatment planning.
Complementary Tools
FLIPI (Follicular Lymphoma International Prognostic Index)
PRIMA-PI
A-FLIPI2 (for patients aged ≥ 60)
IPI (for transformation to DLBCL)
The Evidence
Original Derivation — The F2 Study
Follicular Lymphoma International Prognostic Index 2: A New Prognostic Index for Follicular Lymphoma Developed by the International Follicular Lymphoma Prognostic Factor Project.
Federico M et al. • Journal of Clinical Oncology.. 2009;27(27):4555–4562. Prospective, multinational. n = 942 assessable; n = 832 for model development. 69 European and American institutions. January 2003–May 2005. Five-variable model (B2M, LoDLIN, BMI, Hb, age) selected by bootstrap resampling. 3-year PFS: 91% / 69% / 51% (LR/IR/HR). External validation in n = 231 (5-year PFS: 76% / 46% / 29%). Corrected C-Harrell 0.648.
View SourceIndependent Validation — Arcaini et al.
Validation of follicular lymphoma international prognostic index 2 (FLIPI2) score in an independent series of follicular lymphoma patients.
Arcaini L et al. • British Journal of Haematology.. 2010;149(3):455–457. n = 498 consecutive FL patients (1980–2008); n = 280 with complete FLIPI2 data; n = 190 treated with rituximab. FLIPI2 stratified into three risk groups with significantly different PFS (P = 0.001) and OS (P = 0.009). FLIPI2 superior to FLIPI by Harrell C statistics. FLIPI2 also stratified 139 FLIPI low-risk patients into three meaningful subgroups. Median follow-up 4.7 years.
View SourceA-FLIPI2 — Elderly Population
The age-adjusted international prognostic index 2 (A-FLIPI2) for elderly patients with follicular lymphoma.
Wang J et al. • Heliyon.. 2025;11(4):e42497. Retrospective analysis. Training cohort n = 128 (Tianjin Medical University Cancer Institute & Hospital, 2002–2020); external validation n = 67 (Fujian Medical University Cancer Hospital). A-FLIPI2 (age threshold ≥ 70 instead of ≥ 60) showed AUC 0.793 for death and 0.678 for disease progression — superior to FLIPI (0.637/0.662), FLIPI2 (0.704/0.620), and PRIMA-PI (0.672/0.586). DeLong's test confirmed statistical significance for most comparisons.
View SourceClick to Read
Origins & History
From FLIPI to FLIPI2
The original FLIPI (Solal-Céligny et al., Blood 2004) was built retrospectively from archived data across multiple international institutions. Despite wide clinical adoption, it had key structural limitations: it predated the rituximab era, relied on retrospectively available variables, excluded B2M due to high missingness, and used OS — an increasingly unsuitable endpoint in an indolent disease with long natural history — as its target outcome. By 2003, the International Follicular Lymphoma Prognostic Factor Project launched the F2 study to address all four limitations simultaneously.
Score Development Timeline
01
2004: FLIPI published (Solal-Céligny et al., Blood). Retrospective, n = 1,795 from international archives. Five variables: age, stage, Hb, nodal sites, LDH. Widely adopted but pre-rituximab.
02
2003–2005: F2 study accrual — 1,093 patients registered prospectively across 69 institutions. Rituximab-containing therapy used in 59% of treated patients.
03
04
2009 (J Clin Oncol): FLIPI2 published. Five new variables (B2M, LoDLIN, BMI, Hb, age). Bootstrap-selected model. External validation in 231 patients. Three risk groups with significantly different 3-year PFS (91% / 69% / 51%).
05
06
2018: PRIMA-PI published (Bachy et al., Blood) — simplified 2-variable index (B2M + BMI) for immunochemotherapy patients.
07
2025 (Heliyon): Wang et al. develop A-FLIPI2, adapting FLIPI2 for elderly FL patients by adjusting age threshold from ≥ 60 to ≥ 70 years. External validation confirms superior risk stratification in this underserved population.
The Prospective Advantage
The shift from retrospective (FLIPI) to prospective (FLIPI2) design was methodologically decisive. It allowed the inclusion of B2M — unavailable retrospectively — and of LoDLIN, which could not be standardised across pre-existing databases. It also enabled the use of an electronic case report system (EasyStage) for prospective nodal mapping across all sites, and allowed replacement of OS with PFS as the principal endpoint once the 5-year OS of the study population (89%) proved far better than anticipated.
Last Comprehensive Review: 2026-07-17