Though post-treatment surrogate endpoints, such as disease progression within 24 months (POD24; defined as relapse or progression of FL within 24 months of first-line treatment initiation), can predict overall survival (OS) in patients with follicular lymphoma (FL), these endpoints are of little clinical value, since they cannot guide upfront treatment decisions. In an analysis of predictive risk models for FL, researchers found that m7-FLIPI, a pre-treatment risk model combining clinical and molecular information prospectively identifies the smallest subgroup of FL patients with the highest risk for first-line treatment failure and early death – including patients who did not fulfill POD24 criteria. The study was published in Blood.
“Strategies to guide risk-adapted treatment approaches in FL are needed to avoid overtreatment of low-risk patients and to prioritize alternative over standard treatment regimens in high-risk patients,” the researchers, led by Vindi Jurinovic, MD, of the University Hospital of the Ludwig-Maximilians-University Munich in Germany, explained.
Dr. Jurinovic analyzed the following risk models:
- the FL International Prognostic Index (FLIPI; a risk-stratification tool for FL patients)
- the m7-FLIPI (which includes the mutation status of seven genes: EZH2, ARID1A, MEF2B, EP300, FOXO1, CREBBP, and CARD11)
- the Eastern Cooperative Oncology Group performance status at the time of treatment initiation (ECOG-PS)
Researchers used clinical and molecular data from two independent cohorts of patients with symptomatic, advanced stage or bulky FL who were in need of first-line immunochemotherapy, but who were considered ineligible for curative radiotherapy:
- German Low-Grade Lymphoma Study Group (GLSG) trial (n=151)
- British Columbia Cancer Agency (BCCA; n=107)
In the randomized GLSG study, patients with advanced, bulky disease received R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone) and interferon maintenance therapy. The median patient age was 57 years (range = 27-77 years), and 51 percent (n=77) had high-risk disease, according to FLIPI score. After a median follow-up of 7.7 years, the five-year failure-free survival (FFS) was 66 percent and the five-year OS was 83 percent.
The BCCA group included patients from the population-based BCCA registry who received R-CVP (rituximab, cyclophosphamide, vincristine, and prednisone), followed by rituximab maintenance therapy by intention to treat in 87 percent of patients (n=93). The median patient age was 62 years (range = 37-83 years), and 50 percent (n=53) had high-risk disease, according to FLIPI score. After a median follow-up of 6.7 years, the five-year FFS and OS rates were 58 percent and 74 percent, respectively.
Dr. Jurinovic and colleagues calculated two risk models that specifically predict POD24 using data from the GLSG cohort: In the first model, coefficients for high-risk FLIPI and ECOG-PS >1 were not penalized (forcing these variables into the model). In the second model, all coefficients were penalized.
The BCCA cohort was used as an independent validation cohort.
Nineteen patients (13%) from GLSG and five (5%) from BCCA were not evaluable for POD24. Among the remaining patients, POD24 occurred in 17 percent of the GLSG group (n=23) and 23 percent of the BCCA group (n=23).
OS differed significantly between patients with and without POD24, with six patients in the GLSG group and four patients in the BCCA group alive at five years. The five-year OS rates for GLSG patients with and without POD24 were 41 percent versus 91 percent (hazard ratio [HR] = 9.72; 95% CI 4.51-20.96; p<0.001) and 26 percent versus 86 percent for the BCCA cohort (HR=11.93; 95% CI 5.31-26.76; p<0.001).
These results confirmed the predictive utility of POD24. “[POD24] very closely reflects either the aggressiveness of the disease and/or treatment-specific resistance,” the authors noted. “POD24 will be useful immediately in clinical practice to select high-risk patients for experimental salvage treatments.”
First, the researchers evaluated FLIPI risk in patients with and without POD24, finding that those with high-risk FLIPI and no POD24 did not have inferior FFS compared with those with low-risk FLIPI and no POD24. “FLIPI, which uses only clinical factors and classifies 51 percent of GLSG and 50 percent of BCCA patients as high-risk, overestimates the number of patients with poor outcome,” they reported.
Next, using the m7-FLIPI model, the researchers found that 28 percent of GLSG patients (n=43) and 22 percent of BCCA patients (n=24) were classified as high-risk. These patients were significantly more likely to develop POD24 (OR=5.82 in the GLSG group; p=0.00031 and OR=4.76 in the BCCA group; p=0.0052).
For those with POD24, the five-year OS was 65 percent in the GLSG group and 90 percent in the BCCA group, compared with 42 percent and 84 percent, respectively, for those without POD24 (HR=3.4 in the GLSG group and 4.9 in the BCCA group; p<0.0001 for both).
Compared with FLIPI, the m7-FLIPI model had greater specificity for predicting POD24 (56% vs. 79% in the GLSG group and 58% vs. 86% in the BCCA group). However, 21 percent of GLSG patients and 14 percent of BCCA patients who did not experience POD24 were assigned as high-risk patients per m7-FLIPI measures; six percent and 12 percent of patients, respectively, were classified as low-risk m7-FLIPI but developed POD24. In both cohorts, though, high-risk m7-FLIPI was associated with a shorter FFS and OS among patients without POD24. There also was a subset of patients with POD24 who were not detected by any of the risk models, which presents the need for improvements and integration of additional biomarkers, they added.
“The m7-FLIPI has the highest accuracy and positive predictive value for POD24 among all pre-treatment risk models,” Dr. Jurinovic and colleagues concluded. “Also, high-risk m7-FLIPI is associated with inferior outcome in patients who do not fail treatment within 24 months, a subset currently missed by the POD24 classifier.”
“The m7-FLIPI establishes solid grounds for upfront patient stratification by actual risk, however several challenges still need to be addressed before it can be applied in clinical trials and practice,” the authors cautioned, including standardization of molecular technologies and analysis pipelines to ensure widely reproducible results.
The study’s results are limited because it analyzed a specific patient population that might not be reflective of patients seen in real-world clinical practice. The authors recommend a larger cohort and longer follow-up to further understand these results.
Jurinovic V, Kridel R, Staiger AM, et al. Prospective clinicogenetic risk models to predict early progression of follicular lymphoma after first-line immunochemotherapy. Blood. 2016. [Epub ahead of print]