Use of Molecular Immune Signatures for Frontline Treatment Selection in Patients with Advanced Stage Follicular Lymphoma
Authors
Tony Z. Zhuang, MD¹, Lei Feng, MS²*, Dai Chihara, MD, PhD³, Chijioke Nze, MD, MPH³, Amy A. Ayers, MPH³, Mansoor Noorani, MD³, James D. Ayres⁴, Daniela E. Duenas, MD⁵, Kumudha Balakrishnan, MPhil, MSc, PhD³, Ekaterina Nesterenko, MD, PhD⁶, Mark Meerson, MSc⁶, Ksenia Zornikova, PhD⁷, Krystle Nomie, PhD⁷, Nikita Kotlov, PhD⁷, Nathan Fowler, MD⁶, Maria Alma Rodriguez, MD³, Sattva S. Neelapu, MD³, Christopher R. Flowers, MD, MS³ and Paolo Strati, MD³
- Division of Cancer Medicine, The University of Texas MD Anderson Cancer Center, Houston, TX
- Department of Biostatistics, The University of Texas MD Anderson Cancer Center, Houston, TX
- Department of Lymphoma and Myeloma, The University of Texas MD Anderson Cancer Center, Houston, TX
- Department of Hematopathology, The University of Texas MD Anderson Cancer Center, Houston, TX
- Department of Translational Molecular Pathology, The University of Texas MD Anderson Cancer Center, Houston, TX
- BostonGene, Corp., Waltham, MA
- BostonGene Corp., Waltham, MA
Abstract
Introduction.
Chemoimmunotherapy (CIT) with bendamustine and rituximab (BR) or rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (RCHOP) represent the standard frontline treatment for patients with advanced and high tumor burden follicular lymphoma (FL). The combination of lenalidomide and rituximab (R2) has shown to be non-inferior to frontline CIT. While clinical prognostic scores such as FLIPI, FLIPI-2 and PRIMA-PI have been developed in CIT-treated patients, predictive biological tools for frontline treatment selection are currently lacking for FL. Lymphoma molecular subtyping, identified through transcriptomic characterization of the lymphoma microenvironment (LME), and precise reconstruction of its composition, through deconvolution, have been shown to predict outcomes in patients with large B-cell lymphoma (Kotlov et al., Cancer Discov, 2021). It remains unclear, however, whether molecular immune signatures could also be utilized for treatment selection in FL patients.
Methods.
Pre-treatment tissue biopsies were collected from patients with advanced and high tumor burden FL treated with frontline BR, RCHOP, or R2 and with available long-term follow-up clinical data. Baseline clinical and laboratory characteristics were retrospectively used to calculate FLIPI, FLIPI-,2 and PRIMA-PI scores. Response and survival were assessed according to 2014 Lugano criteria. Bulk RNA sequencing was generated for LME subtyping, cell deconvolution using the Kassandra algorithm (Zaitsev et al., Cancer Cell, 2022), B-cell associated gene signatures (BAGS) analysis, and B-cell receptor (BCR) repertoire analysis. Mann-Whitney test was used for group comparison. The log rank test was used in survival analysis.
Results.
Thirty-five patients were included in the analysis: 12 treated with frontline BR, 12 with frontline RCHOP, and 11 with frontline R2. Twenty-three (66%) had high FLIPI scores, 14 (40%) had high FLIPI-2 scores, and 7 (20%) had high PRIMA-PI scores. Patients with high FLIPI score (3‒5) displayed a significant decrease in BCR diversity as compared to those with low-intermediate FLIPI score (0‒2), shown by smaller Shannon and higher Simpson indexes. In addition, patients with higher FLIPI scores (evaluated as a continuous variable) tended to have higher T-regulatory (T-reg) and T-reg traffic signature scores. No significant associations were observed between LME subtypes, BAGS, cell deconvolution, or BCR repertoire and FLIPI-2 or PRIMA-PI scores.
Median follow-up for the whole cohort was 110 months (95% confidence interval [CI], 81-120 months) and median progression-free survival (PFS) was 78 months (95% CI, 65-105 months). No significant difference in median PFS was observed when comparing CIT-treated patients (BR or RCHOP) to R2-treated patients (78 vs 64 months; p=0.5). However, when analyzing PFS within the 4 BAGS subgroups (light-zone like, dark zone-like, normal-like, and plasma cell-like), patients with normal-like BAGS experienced a significantly longer median PFS when treated with BR (not reached [NR]) compared to RCHOP (143 months) or R2 (31 months)(p=0.05), and PFS with CIT compared to R2 (NR vs 31 months, p=0.03). In addition, when utilizing a T-cell signature score, including CD3D, CD3G, TRBC1, TRAT1, TRAC, CD3E, CD28, TRBC2, ITK, TBX21, and categorizing it as high or low according to the cohort median, patients receiving frontline R2 with high T-cell signature scores to experienced significantly longer median PFS as compared to those with low T-cell signature scores (NR vs 21 months, p=0.04). No association between treatment modality and PFS based on TME subtypes was observed.
Discussion.
While FLIPI scores are associated with the molecular features of FL, including BCR diversity and T-reg signatures, clinical prognostic scores have limited biological significance. Pre-treatment molecular immune signatures can facilitate frontline treatment selection, optimal outcomes being observed when utilizing CIT in patients with normal-like BAGS and R2 in patients with high T-cell signature scores. Larger datasets and implementation in clinical trials are needed to validate these findings.
Chemoimmunotherapy (CIT) with bendamustine and rituximab (BR) or rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (RCHOP) represent the standard frontline treatment for patients with advanced and high tumor burden follicular lymphoma (FL). The combination of lenalidomide and rituximab (R2) has shown to be non-inferior to frontline CIT. While clinical prognostic scores such as FLIPI, FLIPI-2 and PRIMA-PI have been developed in CIT-treated patients, predictive biological tools for frontline treatment selection are currently lacking for FL. Lymphoma molecular subtyping, identified through transcriptomic characterization of the lymphoma microenvironment (LME), and precise reconstruction of its composition, through deconvolution, have been shown to predict outcomes in patients with large B-cell lymphoma (Kotlov et al., Cancer Discov, 2021). It remains unclear, however, whether molecular immune signatures could also be utilized for treatment selection in FL patients.
Methods.
Pre-treatment tissue biopsies were collected from patients with advanced and high tumor burden FL treated with frontline BR, RCHOP, or R2 and with available long-term follow-up clinical data. Baseline clinical and laboratory characteristics were retrospectively used to calculate FLIPI, FLIPI-,2 and PRIMA-PI scores. Response and survival were assessed according to 2014 Lugano criteria. Bulk RNA sequencing was generated for LME subtyping, cell deconvolution using the Kassandra algorithm (Zaitsev et al., Cancer Cell, 2022), B-cell associated gene signatures (BAGS) analysis, and B-cell receptor (BCR) repertoire analysis. Mann-Whitney test was used for group comparison. The log rank test was used in survival analysis.
Results.
Thirty-five patients were included in the analysis: 12 treated with frontline BR, 12 with frontline RCHOP, and 11 with frontline R2. Twenty-three (66%) had high FLIPI scores, 14 (40%) had high FLIPI-2 scores, and 7 (20%) had high PRIMA-PI scores. Patients with high FLIPI score (3‒5) displayed a significant decrease in BCR diversity as compared to those with low-intermediate FLIPI score (0‒2), shown by smaller Shannon and higher Simpson indexes. In addition, patients with higher FLIPI scores (evaluated as a continuous variable) tended to have higher T-regulatory (T-reg) and T-reg traffic signature scores. No significant associations were observed between LME subtypes, BAGS, cell deconvolution, or BCR repertoire and FLIPI-2 or PRIMA-PI scores.
Median follow-up for the whole cohort was 110 months (95% confidence interval [CI], 81-120 months) and median progression-free survival (PFS) was 78 months (95% CI, 65-105 months). No significant difference in median PFS was observed when comparing CIT-treated patients (BR or RCHOP) to R2-treated patients (78 vs 64 months; p=0.5). However, when analyzing PFS within the 4 BAGS subgroups (light-zone like, dark zone-like, normal-like, and plasma cell-like), patients with normal-like BAGS experienced a significantly longer median PFS when treated with BR (not reached [NR]) compared to RCHOP (143 months) or R2 (31 months)(p=0.05), and PFS with CIT compared to R2 (NR vs 31 months, p=0.03). In addition, when utilizing a T-cell signature score, including CD3D, CD3G, TRBC1, TRAT1, TRAC, CD3E, CD28, TRBC2, ITK, TBX21, and categorizing it as high or low according to the cohort median, patients receiving frontline R2 with high T-cell signature scores to experienced significantly longer median PFS as compared to those with low T-cell signature scores (NR vs 21 months, p=0.04). No association between treatment modality and PFS based on TME subtypes was observed.
Discussion.
While FLIPI scores are associated with the molecular features of FL, including BCR diversity and T-reg signatures, clinical prognostic scores have limited biological significance. Pre-treatment molecular immune signatures can facilitate frontline treatment selection, optimal outcomes being observed when utilizing CIT in patients with normal-like BAGS and R2 in patients with high T-cell signature scores. Larger datasets and implementation in clinical trials are needed to validate these findings.
Latest publications
