Relapsed and Refractory DLBCL in 2026: More Options Than Ever, and Three Questions That Actually Matter

Introduction: Progress That Actually Matters — and How to Use It

Even with PolaRCHP as the new frontline standard, roughly 35% of DLBCL patients will still experience primary induction failure or relapse. The five-year PFS in POLARIX was 64.9% for PolaRCHP versus 59.1% for R-CHOP.  While this is meaningful progress, it still leaves a substantial population with relapsed or refractory disease who will need additional therapies.

The good news is that the options for those patients in 2026 are better than they have ever been. In the entire history of randomized trials in relapsed/refractory DLBCL, only five trials have shown an overall survival benefit: PARMA in 1995, ZUMA-7 and TRANSFORM (establishing CAR-T in second line), STARGLO in 2024, and ECHELON-3 in 2025. Four of those were published in just the last four years. The field is finally moving.  In addition there are many other trials expected to read out in the next 1-2 years. 

The challenge is not finding something to offer. It is selecting the right option for the right patient. This requires thinking in categories.

---

Thinking in Buckets: Patient Stratification

The most useful organizing framework for relapsed/refractory DLBCL puts patients into three groups based on two dimensions: what happened in first line, and what are they eligible for now.

Bucket 1: Patients who are eligible for either transplant or CAR-T-eligible. These are patients who had primary induction failure or relapsed within twelve months of completing first-line therapy, as defined in the ZUMA-7 and TRANSFORM randomized trials. For primary induction failure (never achieving a CR with first-line therapy), going directly to CAR-T is the standard approach. For early relapse, the data favor CAR-T, although the story is more nuanced than a simple guideline.

Bucket 2: CAR-T-eligible but not transplant-eligible. These patients, often older or with comorbidities that preclude high-dose chemotherapy and stem cell transplant, could still potentially be CAR-T candidates.  Liso-cel has been specifically studied in this population, and the data support its use in patients who would have been excluded from the registrational trials.

Bucket 3: Neither transplant nor CAR-T eligible — or already past CAR-T. This is the population where the recent approvals and trials matter most. Multiple regimens are now available, some with OS benefit, and choosing among them requires a careful look at the individual patient.

---

The Early Relapse Question: Two Shots on Goal

The randomized data from ZUMA-7 and TRANSFORM established overall survival benefits for CD19 CAR-T versus standard second-line chemotherapy followed by autologous transplant. At median follow-up of 47 months, ZUMA-7 showed a three-year OS of 56% versus 48% in favor of axi-cel. TRANSFORM showed a trend toward OS benefit with liso-cel.

These are real, important data, and allowed us to extend CAR-T to many patients who previously would not have qualified.  But they deserve careful interpretation. In both trials, 74-75% of enrolled patients were primary induction failures. In addition, the SOC arm was structurally disadvantaged: half the patients receiving salvage chemotherapy never responded (as expected) and immediately experienced an event, and did not proceed to transplant.  Meanwhile, almost all of the CAR-T arm patients received their assigned treatment. A head-to-head trial in which half of the control arm never actually gets the assigned treatment, cannot be expected to show the control arm winning.

Again, for primary induction failure patients there is no question that they should proceed to CAR-T.  However, a question worth asking is: what happens to the patients in the SOC arm who did respond to chemotherapy and made it to autologous transplant? The data, from both ZUMA-7 supplemental analyses and from personal communication with the TRANSFORM investigators, suggest that chemosensitive patients who proceeded to auto-transplant had outcomes that were not dramatically inferior to CAR-T. In ZUMA-7, the 95% confidence interval for CAR-T's advantage in that subgroup was "consistent with the possibility of no effect."

This matters because decades of research have shown that auto-transplant can be curative in approximately 40 to 45% of chemosensitive early-relapse patients, including those who relapsed within twelve months. And if they relapse after transplant, CAR-T remains a completely viable option. The reverse is not reliably true: auto-transplant after CAR-T failure is not a standard pathway.

The "two shots on goal" argument: for a patient who relapses around 9-11 months, has responded to salvage chemotherapy, has a negative PET scan, and is fit for transplant — is it always better to go directly to CAR-T and forgo the chance to cure them with auto-HCT? This is not a settled question. In practice, many of these patients arrive having already received second-line chemotherapy in the community, with a negative PET, and the decision needs to be made individually. CAR-T after auto is well established; auto after CAR-T is much harder to execute.  These type of patients are not common but it is worth considering auto-HCT in some cases in order to have two good options at cure beyond first-line therapy.

---

Options Beyond CAR-T: What Has OS Benefit

For patients who are not candidates for transplant or CAR-T, or who have already had CAR-T, the following options now have the strongest data:

Glofitamab plus Gem-Ox (STARGLO): The three-year follow-up from STARGLO showed a two-year PFS of 45% versus 20% and two-year OS of 50% versus 30% in favor of Glofit-Gem-Ox. For second-line patients specifically, two-year PFS was 50% versus 25% and two-year OS was 55% versus 35%. Excluding PARMA (auto-HCT), ZUMA-7 (CAR-T) and TRANSFORM (CARiT), this is one of only two randomized R/R DLBCL trials with an OS benefit (alongside ECHELON-3). Only 8% of patients in STARGLO had prior CAR-T, so its performance in that context is less established. FDA did not approve this regimen primarily because of enrollment heavily skewed toward Asian patients; however, Glofit-Gem-Ox is still widely used in the US based on the strength of the STARGLO data.

Epcoritamab (EPCORE DLBCL-1): The phase 3 data comparing epcoritamab monotherapy versus R-GemOx or BR were announced in January 2026 showing improved PFS, CR rate, and duration of response, though OS benefit had not yet been seen at the time of the announcement. Full presentation is expected at an upcoming 2026 meeting. The phase 2 single-agent data showed 82% ORR, 62% CR, median PFS of 18 months, and 74% of CR patients still in remission at 18 months, with the advantage of subcutaneous administration and outpatient delivery.

Mosunetuzumab plus polatuzumab (Mosun-Pola): in the randomized SUNMO trial, Mosun-Pola showed superior PFS compared to R-Gem-oxali (12 month PFS 48% vs 18%).  The OS data is not mature yet.  Whether this regimen will show reduced activity in patients previously exposed to polatuzumab (such as those who received Pola-RCHP as first-line therapy) is not yet known.

Glofitamab plus polatuzumab (Glofi-Pola): In 129 patients, including 44 with high-grade B-cell lymphoma and 21% with prior CAR-T, the combination produced 78% ORR, 60% CR, median PFS of 12 months, and median OS of 34 months. CRS was 43% but nearly all grade 1 to 2. While this was not a randomized trial, these results are excellent, especially as a chemo-free option.  Similar to Mosun-Pola, it is not clear if the results will be as favorable in those with prior polatuzumab exposure.   

Loncastuximab (Lonca):  The LOTIS-2 trial tested single-agent Lonca in R/R DLBCL patients, the majority of whom had three or more prior lines of therapy.  The ORR was 48% with a CR rate of 25%.  While the median PFS overall was only 4.9 months, for patients achieving CR, many of these CRs were durable at 2 years and beyond.  This is among the simplest regimens to deliver (outpatient, every three weeks, no step-up dosing, no risk for CRS) and it retains activity in heavily pretreated patients. The LOTIS-5 randomized trial comparing loncastuximab-rituximab versus R-Gem-Ox is ongoing, with results expected to be presented sometime later this year.

Brentuximab vedotin plus R2 (ECHELON-3): In 230 patients with a mean of 2-3 prior lines of therapy, including 30% with prior CAR-T and over 50% refractory to first-line therapy, BvR² versus R² produced 64% versus 42% ORR, 40% versus 19% CR, and a median PFS of 4.2 versus 2.6 months, with a median OS of 13.8 versus 8.5 months (p=0.009). Importantly, only 37% of patients were CD30-positive, and the clinical benefit was present regardless of CD30 expression. 

Epcoritamab plus lenalidomide: in the next 1-2 years we expect readouts on several randomized phase 3 trials in R/R DLBCL including the EPCOR DLBCL-4 trial which is comparing epcoritamab + lenalidomide vs investigator’s choice of chemo-immunotherapy.

\

---

How to Choose Among These Options

Based on the currently available data reviewed above, the bispecific-based regimens (Glofit-Gem-Ox, Glofit-Pola, epcoritamab) have the best efficacy data. The tradeoffs are logistical: step-up dosing requires monitoring visits, and some patients prefer not to come in frequently or are not candidates for inpatient observation during step-up.

Practical decision points:

• If proven OS benefit is the priority: Glofit-Gem-Ox or BvR2

• If the patient is elderly and has a relapse after a long first remission, tafasitamab-lenalidomide is a well tolerated option 

• If simplicity (IV infusion once every 3 weeks, no step up dosing, etc), loncastuximab is a good option

• If the patient has significant neuropathy: avoid brentuximab-containing regimens

• If there is any downstream possibility of CAR-T or bispecific antibody therapy: avoid bendamustine, which depletes T cells

• If CD20 expression is lost: responses to CD20-directed bispecifics are inferior; CD19 or CD19/CD20 targeting may be preferable

• If CD19 is lost, or the patient has progressed after CD19 CAR-T: this actually does not seem to predict lack of response to loncastuximab or tafasitamab

• If CD30 expression is absent: this did not predict lack of benefit in ECHELON-3; in fact the majority of the patients in the trial were CD30-negative by IHC

---

For Patients

If your DLBCL has returned or did not respond to initial treatment, the treatment options available in 2026 are substantially better than they were even a few years ago. Several important questions will guide which approach is best for you: How quickly did your lymphoma return? Did it respond to initial treatment? What is your overall health and fitness level?

For patients whose disease returned within twelve months or never responded to initial treatment, a type of cellular therapy called CAR-T, in which your own immune cells are reprogrammed to attack the lymphoma, is often the recommended approach. For patients who are not candidates for CAR-T, or who have already received it, several newer options are available, including drugs called bispecific antibodies that redirect your immune system to the cancer cells, antibody-drug conjugates that deliver chemotherapy directly to cancer cells, and other targeted agents. Ask your oncologist specifically which of these options might apply to your situation, and ask about clinical trials.

---

Key Takeaways

• 35 to 40% of DLBCL patients experience primary induction failure or relapse; we now have five randomized R/R DLBCL trials with OS benefit, three of them in the last two years.

• Patient stratification: early relapse or primary induction failure (bucket 1) → prioritize CAR-T; CAR-T eligible but transplant-ineligible (bucket 2) → CAR-T directly; neither transplant eligible nor CAR-T eligible, or post-CAR-T (bucket 3) → bispecific antibody-based or ADC regimens.

• For chemosensitive late-relapsed patients (first remission >12 months), and also potentially for some early-relapse (<12 months) patients, the "two shots on goal" argument for auto-HCT followed by CAR-T at relapse remains valid; CAR-T after auto is well-established.

• STARGLO (Glofit-Gem-Ox): 2-yr OS 50% vs 30%; OS benefit confirmed; preferred for patients where survival benefit is the priority.

• ECHELON-3 (BvR²): OS benefit (13.8 vs 8.5 months); CD30 expression not required for benefit; relevant for heavily pretreated patients.

• Epcoritamab: phase 3 data announced January 2026, improved PFS and CR vs R-GemOx; outpatient subcutaneous delivery; full presentation pending.

• Glofit-Pola: 78% ORR, 60% CR, active in post-CAR-T patients (21% had prior CAR-T); chemo-free.

• Loncastuximab: simplest to deliver, durable CRs, active in heavily pretreated patients

• avoid bendamustine in patients who may proceed to CAR-T 

---

About the Author

Timothy S. Fenske, MD, is a hematologist-oncologist at the Sarah Cannon Transplant and Cell Therapy Program at Methodist Hospital in San Antonio, Texas. He specializes in advanced therapies for patients with hematologic malignancies such as cellular (CAR-T) therapy, hematopoietic stem cell transplantation, with a particular focus on management of patients with aggressive lymphomas such as DLBCL.

---

References

1. STARGLO trial (3-year update). Glofitamab-Gem-Ox vs R-Gem-Ox in R/R DLBCL. Lancet, 2025; ASH 2025 update.

2. ECHELON-3. Brentuximab vedotin plus R2 in R/R DLBCL. Journal of Clinical Oncology, 2025.

3. EPCORE DLBCL-1 (phase 3 topline results). Epcoritamab vs R-GemOx or BR. Announced January 2026.

4. Hutchings M, et al. Glofitamab plus polatuzumab in R/R DLBCL. Journal of Clinical Oncology, 2025.

5. Caimi P, et al. LOTIS-2: loncastuximab in R/R DLBCL. Haematologica, 2023.

6. Westin J, et al. ZUMA-7: axi-cel vs standard care in 2L DLBCL (OS update). New England Journal of Medicine, 2023.

7. Kamdar M, et al. TRANSFORM: liso-cel vs standard care in 2L DLBCL. Presented at ASCO 2024.

8. Fenske TS. "Treatment of Relapsed and Refractory Aggressive B-Cell NHLs." Presented at the 2026 Best of Hematology, Austin, TX, May 2026.