Transformation of follicular lymphoma (FL) to a large B-cell lymphoma refers to the development of a large-cell lymphoma arising from FL and may be identified either concurrently or after a prior diagnosis of FL. The diagnosis is usually determined by pathology review, but clinically aggressive features, tempo of progression, and B symptoms suggest the diagnosis if biopsy is not feasible or indeterminate. The leading theory of transformation stems from the presence of common progenitor cells (CPC) that serve as a reservoir from which subclones may emerge [1].
The t(14;18) BCL-2 translocation is a key driver of cell survival, as it is in its indolent counterpart, but transformation lacks a single genetic event. Rather, it stems from a diverse spectrum of changes involving cell cycle, DNA damage repair, and immune evasion, among others [2].
Previously untreated transformed follicular lymphoma (tFL) has similar outcomes to de novo (dnDLBCL) with first-line chemoimmunotherapy. Those treated with chemotherapy for FL and those with early relapse have among the worst outcomes [3]. It is likely that aggressive disease biology and selection of chemotherapy-resistant clones contribute to this.
These challenges are precisely why signals toward improved outcomes of tFL with T-cell-engaging therapy are welcomed.
In reviewing prospective and retrospective studies of T-cell-redirected therapies in tFL, interpretation of results is made challenging by several factors unique to the indolent transformed subset. These include results derived from exploratory sub-analyses of small populations which seldom reach statistical significance, lack of separation of tFL from other indolent transformed subsets in data reporting, and the potential for misclassification of histology without central review. Despite these limitations, it is valuable to look at recurring trends while being cautious not to overinterpret.
Representation of transformed indolent lymphoma in prospective CAR-T trials ranges from 11% to 22%; some trials report on tFL specifically, and others have grouped transformed indolent lymphoma subsets together. Table 1 summarizes trial results for CAR-T in transformed indolent lymphoma with details provided for tFL, where data are available. For CD19 CAR-T axicabtagene ciloleucel (axi-cel) with CD28 co-stimulatory domain, 5-year follow-up of ZUMA-1 in (third line) 3L+ showed a numerical difference in median event-free survival (EFS) in tFL compared to dnDLBCL (29.7 months (95% CI [2.3 months-not reached (NR)]) in tFL vs.
4.8 months (95% CI [3.0–7.3 months]) in dnDLBCL), though not statistically significant. In this study, tFL represented 16% of the cohort [4]. ZUMA-7 in second line (2L) included 11% of patients with tFL and showed similar outcomes for tFL vs. dnDLBCL, as reported in an early analysis [5].
The TRANSCEND study evaluated 4-1BB CD19 CAR-T lisocabtagene maraleucel (liso-cel) in 3L+ with 22% of its cohort consisting of tFL. Complete response (CR) rates in tFL vs. dnDLBCL were (63% (95% CI [49.3–75.6%]) vs. 49% (95% CI [40.0–57.7%]).
However, median progression-free survival (mPFS) was improved for tFL vs. DLBCL (NR (95% CI [11.8 months-NR]) vs. 3 mo (95% CI [2.8–6.3 months)]) and there was a trend toward improved median overall survival (mOS) (NR (95% CI [21.1 months-NR]) vs.
12.7 months (95% CI [9–21.1 months])) [6].
Retrospective data mirror trial data and provide a more compelling argument for improved outcomes to CAR-T in tFL vs. dnDLBCL [10,11,12,13]. Table 2 summarizes relevant retrospective data. Cortese et al. noted superior 3 yr PFS (54% vs.
39%, p < 0.01) and OS (59% vs. 39%, p < 0.01) when comparing tFL to dnDLBCL. Patients with tFL and dnDLBCL possessed similar high-risk features such as prior lines of therapy, bulky disease, stage, etc.
Some of these high-risk subgroups (DHL and relapsed/refractory (R/R) disease) fared better when originating from FL versus de novo disease [10]. In another study, Fabbri et al. grouped tFL and transformed marginal zone lymphoma (tMZL) when compared to dnDLBCL; tFL accounted for the majority of transformed diseases (90%). In this study, an improved 2 yr PFS (44% vs.
31%, p < 0.01) and OS (61% vs. 48%, p < 0.01) were reported for indolent transformed NHL (itNHL) vs. dnDLBCL. itNHL patients were more heavily pretreated and had higher incidence of exposure to prior autologous stem cell transplants (ASCTs) compared to dnDLBCL. When compared to tMZL directly, tFL had similar PFS and OS [11].
No comparative studies reported significant differences in cytokine release syndrome (CRS), immune effector-associated neurotoxicity syndrome (ICANS), or infections.
It is unclear why tFL may potentially respond better to CAR-T compared to dnDLBCL. On comparing the tumor microenvironment (TME) in FL pre- and post-transformation, T follicular helper cells are decreased, CD8+ cells are increased, and there is a higher expression of immunosuppressive markers such as PD-1 and LAG-3, coupled with impairment in effective crosstalk between the malignant B-cells and CD4/CD8 T-cells [14,15,16]. However, tFL tissue exhibits tumor infiltrating T, NK, macrophage, and dendritic cells that suggest there is at least active cell trafficking of these immune subsets into the tissue.
Exploratory analyses in JULIET for 3L+ (tisa-cel) showed that while high CD3+ tumor infiltrate had better outcomes to CAR-T, high LAG3+CD3+ presence was detrimental [17]. How this relates to tFL histology is not clear. It is possible that CAR-T may be more effectively trafficked directly into the tumor tissue with the ability to overcome the local suppressive cytokine milieu; this hypothesis has yet to be confirmed but warrants evaluation.
An exploratory analysis of TRANSCEND NHL 001 identified an “FL-like” gene-expression signature derived from genes differentially expressed between FL and DLBCL. A higher FL-like expression score was associated with longer PFS after liso-cel. The signature was not specific for histologically transformed FL and was elevated in some dnDLBCL samples but does underscore the importance of ongoing efforts to identify biologic predictors of response and survival with CAR-T in patients including those with tFL [18].
Although CAR-T has curative potential in both tFL and dnDLBCL, more is needed to improve outcomes. Augmenting CAR-T responses is of interest in tFL. Our group reported encouraging results with CAR-T and adjunctive Bruton tyrosine kinase inhibitor (BTKi) zanubrutinib in DLBCL [19].
Other trials have reported success, and a recent review summarizes the available evidence for combining BTK inhibitors with CAR T-cell therapy [20]. Another approach is to consider CAR-T in frontline to exploit healthier T-cells without exposure to prior chemotherapy. The ZUMA-23 trial is underway to evaluate CAR-T in high-risk 1L LBCL including tFL but limited to those without prior anthracycline therapy.
Bi-and tri-specific CAR-T trials are also ongoing and include tFL with the expectation of yielding promising results in aggressive B-cell lymphoma collectively.
By contrast to CAR-T trials, trials evaluating bispecific antibodies (BsAbs) in aggressive B-cell lymphoma typically included fewer numbers of patients with tFL, as noted in Table 1. In the frontline setting, most BsAb trials underway have excluded this subset. In the R/R setting, granular data detailing outcomes in tFL subsets are more limited.
Thus far, only two CD3/CD20 BsAbs, namely epcoritamab and glofitamab, have been approved in the United States for the treatment of DLBCL including transformed indolent lymphoma. The epcoritamab monotherapy trial (≥3L) sub-analysis showed that similar CR rates for all grouped transformed indolent lymphomas vs. dnDLBCL were 48% (95% CI [32–64%]) vs. 38% (95% CI [28–49%]) [7].
Glofitamab monotherapy (≥3L) results were reported for tFL specifically and included a significant number of patients treated with prior CAR-T. In the glofitamab study, CR rates for tFL vs. dnDLBCL were 41% (95% CI [18–67%]) vs. 36% (95% CI [26% vs.
48%]) [8].
In ≥2L, both these BsAb have been evaluated in combination with gemcitabine-oxaliplatin (GemOx) for patients who are ineligible for autologous stem cell transplantation. However, only the epcoritamab-GemOx trial included patients with transformed indolent lymphoma and reported a CR rate of 75% (95% CI [53–90%]) vs. 56% (95% CI [44–67%]) in transformed lymphoma vs. dnDLBCL [9].
Thus, it is uncertain whether the addition of chemotherapy, which is commonly used outside of the limitation of frailty, is beneficial in the tFL population.
When selecting between BsAb and CAR-T for tFL, relevant considerations include eligibility for CAR-T, time toxicity in cases of planned indefinite therapy, access to specialized cellular-therapy centers, and the maturity of available follow-up data. As the follow-up for CAR-T is longer and shows a steady plateau for relapse, it is the authors’ view that until more definitive and long-term data are presented for bispecifics, patients equally eligible should be preferentially offered CAR-T. As BsAb and CAR-T move into the frontline for DLBCL, data will invariably be extrapolated to tFL, which will impact sequencing.
Retrospective data published by our group for DLBCL do not show a significant difference in outcomes based on sequencing of CAR-T and BsAbs but more work is needed to explore patient frailty influencing choice, T-cell exhaustion, and CAR-T-cell fitness post-BsAb [21].
An underexplored question of relevance to tFL in these reviewed studies is how bendamustine exposure impacts CAR-T and BsAb outcomes. As prior bendamustine use is a surrogate marker for prior lines of therapy and/or more aggressive disease in need of treatment, this topic is difficult to study retrospectively. DeCicco et al. reported inferior outcomes in FL treated with bendamustine within 1 year of CAR-T (2 yr OS 40% vs.
68%, p = 0.02) [22]. Other studies have noted a similar trend toward inferior outcomes to CAR-T or BsAb with recent bendamustine exposure in DLBCL [23]. Iacoboni et al. reported on bendamustine vs. non-bendamustine-exposed R/R DLBCL patients, noting significantly lower ALC at apheresis and significantly reduced PFS and OS that maintained the trend but lost significance in a matched analysis [24].
The impact of bendamustine surrounding T-cell-directed therapy continues to be debated and future studies assessing tFL outcomes should highlight this variable.
Transformed indolent lymphomas are heterogeneous in biology yet typically grouped together as a single entity or often excluded in clinical trials evaluating T-cell-redirecting therapy. Though repeated trends of improved outcomes are noted with CAR-T in trials and real-world data for tFL, robust numbers of these patients in prospective studies are lacking. While it does not eliminate unfavorable outcomes, CAR-T remains active in high-risk subgroups in tFL.
Data are even more limited for BsAb in this subset. As ongoing trials evolve and new trials are designed, it will be important to enrich this group to generate more relevant data.