Diffuse large B-cell lymphoma

R-CHOP × 4 with PET-guided omission of radiation (FLYER, S1001): 4 cycles if interim PET negative; involved-site radiotherapy if PET positive.

R-CHOP × 6 (or Pola-R-CHP); consider 4 cycles plus 2 rituximab in young low-risk patients (FLYER).

Pola-R-CHP × 6 (POLARIX) or R-CHOP × 6; tafasitamab + lenalidomide + R-CHOP (frontMIND) pending approval for IPI 3-5; DA-EPOCH-R for double-hit lymphoma; CNS prophylaxis for high CNS-IPI (contested).

R-mini-CHOP; epcoritamab-based regimens in trials for the elderly (EPCORE NHL-2 cohorts); tafasitamab-lenalidomide where transplant is never an option.

CD19 CAR-T (axi-cel or liso-cel) preferred over salvage chemotherapy and transplant (ZUMA-7, TRANSFORM); bridging therapy while manufacturing; bispecific ± chemotherapy if CAR-T unavailable.

Salvage chemotherapy (R-ICE, R-DHAP, R-GemOx) → high-dose therapy and autologous transplant if chemosensitive; CAR-T if not.

CD20×CD3 bispecific (glofitamab, epcoritamab) or mosunetuzumab-polatuzumab (SUNMO); pola-BR; tafasitamab-lenalidomide; loncastuximab tesirine.

CAR-T if not yet given; bispecific after CAR-T (active in CD19-negative relapse if CD20 retained); loncastuximab; zilovertamab vedotin (trial); allogeneic transplant in selected fit patients; clinical trials.

An excision or core biopsy reported to the current WHO classification, with immunohistochemistry for CD20, CD10, BCL6, MUM1, BCL2, MYC, Ki-67 and, where MYC is expressed, fluorescence in situ hybridisation for MYC, BCL2 and BCL6 rearrangements, because a high-grade B-cell lymphoma with MYC and BCL2 rearrangements is treated differently from diffuse large B-cell lymphoma. Staging is by FDG-PET-CT reported by the Lugano classification, with a bone marrow biopsy only where PET leaves a question. Bloods include LDH, which is an IPI factor, and hepatitis B surface antigen and core antibody, hepatitis C and HIV, because all three change treatment. Cardiac function is assessed before doxorubicin. Fertility preservation is offered before the first cycle, not after it. The IPI (age over 60, stage III or IV, more than one extranodal site, performance status 2 or worse, raised LDH) sets the risk group, and the CNS-IPI adds kidney or adrenal involvement to estimate the risk of relapse in the brain.

Four cycles of R-CHOP with two extra doses of rituximab, rather than six cycles, for patients aged 18 to 60 with stage I or II disease, normal LDH, performance status 0 to 1 and no mass of 7.5 cm or more. FLYER randomised 588 such patients and found three-year progression-free survival of 96 per cent with the four-cycle arm, non-inferior to six cycles, and with fewer adverse events recorded in the four-cycle group (294 haematological and 1,036 non-haematological events against 426 and 1,280). Radiotherapy is not given routinely; an involved-site 30 to 40 Gy field is added where the end-of-treatment PET is positive, or for a bulky or skeletal site. The practical point is that a young person with truly limited, low-risk disease is now finished in about three months.

Six cycles of R-CHOP every 21 days. This has been the reference regimen since rituximab was added to CHOP in 2002, and every attempt to improve on it in this risk group has failed: DA-EPOCH-R was no better and more toxic in Alliance/CALGB 50303 (491 patients, progression-free survival hazard ratio 0.93, febrile neutropenia 35.0 against 17.7 per cent), and obinutuzumab in place of rituximab was no better in GOYA (1,418 patients, hazard ratio 0.92). POLARIX, which established polatuzumab vedotin in place of vincristine, enrolled only patients with IPI 2 to 5, so there is no randomised evidence for pola-R-CHP in IPI 0 to 1 disease. Cure is the aim and is achieved in the large majority.

Six cycles of polatuzumab vedotin with rituximab, cyclophosphamide, doxorubicin and prednisone (pola-R-CHP), or six cycles of R-CHOP. POLARIX randomised 879 previously untreated patients aged 18 to 80 with IPI 2 to 5: two-year progression-free survival 76.7 against 70.2 per cent (hazard ratio 0.73) and five-year progression-free survival 64.9 against 59.1 per cent (hazard ratio 0.77). Overall survival has not separated: the five-year figures are 82.3 against 79.5 per cent, hazard ratio 0.85, not significant. So polatuzumab prevents some relapses without yet being shown to prevent deaths, and the prespecified subgroup analyses suggested the benefit sat with activated B-cell subtype and with IPI 3 to 5 rather than with germinal centre or IPI 2 disease. That is the whole argument, and it is a reasonable one to have out loud with the patient: an extra drug, more peripheral neuropathy, fewer relapses, no proven survival gain. The newer option is tafasitamab and lenalidomide added to R-CHOP for IPI 3 to 5. frontMIND randomised 899 such patients and reported two-year progression-free survival 71.1 against 62.9 per cent (hazard ratio 0.75). DA-EPOCH-R is used instead where the disease is a high-grade B-cell lymphoma with MYC and BCL2 rearrangements, or is primary mediastinal, testicular, or leukaemic.

Dose-adjusted EPOCH-R rather than R-CHOP, with central nervous system-directed treatment, on the basis of consistent retrospective series rather than a randomised trial; there has never been one, because the entity is uncommon and was only separated out in 2016. The diagnosis requires fluorescence in situ hybridisation, so it is missed wherever FISH is not done, which is the argument for testing every case that expresses MYC by immunohistochemistry. Dual expression of MYC and BCL2 protein without a rearrangement (double-expressor) carries a worse outlook but is not itself a reason to leave R-CHOP. MYC with BCL6 rearrangement is now classified separately and behaves less badly than MYC with BCL2. Relapsed disease follows the large B-cell lymphoma pathway: CD19 CAR-T, then bispecific antibodies.

Attenuated R-CHOP (R-mini-CHOP), typically at 50 per cent of the cyclophosphamide, doxorubicin and vincristine doses, is standard for patients over about 80 and for frail patients of any age, and is given with the intention to cure rather than to palliate. A comprehensive geriatric assessment before treatment predicts who can take full-dose therapy better than age does. A pre-phase of prednisolone with or without one dose of vincristine for a week before cycle 1 improves performance status and reduces early deaths. G-CSF is given from the first cycle. Where an anthracycline cannot be given at all, options include substituting etoposide or liposomal doxorubicin, or an anthracycline-free regimen; all are less effective and should be a considered choice rather than a default. Trials of mosunetuzumab consolidation after pola-R-mini-CHP in older patients with detectable circulating tumour DNA are open.

Relapse in the brain or spinal fluid occurs in roughly 2 to 5 per cent of patients overall and in about 10 per cent of those with a high CNS-IPI. The traditional response, intrathecal methotrexate with each cycle or two to four doses of systemic high-dose methotrexate, has not been shown to reduce it. In 1,162 adults across 21 United States academic centres who all received single-route prophylaxis, central nervous system relapse occurred in 5.7 per cent, with no difference between intrathecal (5.4 per cent) and systemic high-dose methotrexate (6.8 per cent), and the observed rate matched the rate predicted from CNS-IPI alone. There has never been a randomised trial. Practice has changed accordingly: intrathecal prophylaxis is largely abandoned in the United Kingdom, and systemic high-dose methotrexate is offered selectively, to testicular involvement, to high CNS-IPI and to high-grade B-cell lymphoma, and is increasingly framed as a choice. The interventions that do change outcome are an adequate systemic regimen and prompt investigation of any new neurological symptom.

CD19 CAR-T is the second-line standard. ZUMA-7 randomised 359 patients with large B-cell lymphoma refractory to, or relapsing within twelve months of, first-line therapy to axicabtagene ciloleucel or to salvage chemotherapy with autologous transplant in responders: median event-free survival 8.3 against 2.0 months (hazard ratio 0.40) and four-year overall survival 54.6 against 46.0 per cent (hazard ratio 0.73). TRANSFORM randomised 184 transplant-eligible patients to lisocabtagene maraleucel or the same standard of care: complete response 74 against 43 per cent, progression-free survival hazard ratio 0.400, with grade 3 cytokine release syndrome in 1 per cent and grade 3 neurological events in 4 per cent. BELINDA, with tisagenlecleucel and a longer manufacturing interval, was negative (event-free survival hazard ratio 1.07), which is why product and pathway speed are treated as part of the treatment rather than a detail of it. In practice: refer at the first suspicion of relapse, confirm with biopsy, collect cells early, and bridge with steroids, radiotherapy, polatuzumab-based chemotherapy or a bispecific antibody while the product is made. Where CAR-T is not available or the patient is not fit for it, salvage chemoimmunotherapy with autologous transplant, or a bispecific antibody, are the alternatives.

Two to three cycles of platinum-based salvage immunochemotherapy (R-ICE, R-DHAP or R-GDP), then, if the disease responds, high-dose therapy with BEAM conditioning and an autologous stem cell transplant. The randomised comparisons between the salvage regimens found no difference in response or survival, so the choice is made on toxicity: R-DHAP is harder on the kidneys and hearing, R-ICE on the marrow, R-GDP is the gentlest and can be given as an outpatient. Patients whose disease does not respond to salvage should be moved to CAR-T rather than given a second salvage regimen. Stem cells must be collected before the marrow is exhausted, which is a reason to avoid bendamustine in anyone who may need a transplant.

Several options, none of them curative on current evidence, and the choice turns on what the person wants from treatment. Fixed-duration bispecific antibodies. Glofitamab for twelve cycles then stop, after a dose of obinutuzumab to blunt cytokine release. With gemcitabine and oxaliplatin, STARGLO reported median overall survival 25.5 against 12.9 months for rituximab-GemOx (hazard ratio 0.62). This is one of the places where England is ahead of the United States: NICE TA1113 allows glofitamab with gemcitabine and oxaliplatin for relapsed or refractory diffuse large B-cell lymphoma not otherwise specified after one line of treatment in adults who are not eligible for an autologous transplant, while glofitamab's only American indication remains the accelerated approval of 15 June 2023 for disease after two or more lines. Continuous bispecific antibodies. Epcoritamab, given subcutaneously; EPCORE DLBCL-1 randomised 552 transplant-ineligible patients against investigator's choice of R-GemOx or bendamustine-rituximab and improved progression-free survival (hazard ratio 0.74) without improving overall survival (hazard ratio 0.96). Combinations. Mosunetuzumab with polatuzumab vedotin in SUNMO gave progression-free survival 11.5 against 3.8 months for R-GemOx (hazard ratio 0.41), overall response 70 against 40 per cent and complete response 51 against 24 per cent. Polatuzumab with rituximab, gemcitabine and oxaliplatin in POLARGO improved overall survival, 19.5 against 12.5 months (hazard ratio 0.6), in 255 transplant-ineligible patients. Antibody-drug conjugates and immunomodulatory pairs. Loncastuximab tesirine alone (LOTIS-2: response 48 per cent, complete response 24 per cent). Tafasitamab with lenalidomide (L-MIND: response 60 per cent, complete response 43 per cent), which suits someone who wants an outpatient oral-and-infusion regimen. Polatuzumab with bendamustine and rituximab, which should be avoided in anyone who may still go to CAR-T because bendamustine damages the T cells. Brentuximab vedotin with lenalidomide and a rituximab product, approved in the United States on 11 February 2025 on the ECHELON-3 trial, for people after two or more lines who are not eligible for an autologous transplant or for CAR-T.

If CAR-T has not been given and the patient is fit, it is given now: ZUMA-1 reported an objective response of 82 per cent and complete response 54 per cent in refractory large B-cell lymphoma, with 42 per cent still in response at a median 15.4 months. After CAR-T, a CD20 bispecific antibody is the usual next step and retains activity where CD19 has been lost, provided CD20 is still expressed; a repeat biopsy is therefore worth doing rather than assuming. Other options are loncastuximab tesirine, selinexor, zilovertamab vedotin in a trial, an allogeneic transplant in a small number of fit younger patients with chemosensitive disease, and re-treatment with a drug that previously worked after a long interval. A clinical trial is a reasonable first choice at this point rather than a last resort. Where the aim changes from control to comfort, palliative radiotherapy to a symptomatic site works quickly and early palliative care alongside oncology is recommended.

Interim PET is not used to change treatment in diffuse large B-cell lymphoma outside a trial, because no randomised trial has shown that switching on an interim scan helps. End-of-treatment PET-CT, reported on the five-point Deauville scale, decides whether treatment is complete: scores 1 to 3 are a complete metabolic response, 4 to 5 need biopsy of the residual site before any further treatment, because inflammation and brown fat both light up. Afterwards, follow-up is clinical: history, examination and bloods every three months for two years, then less often. Routine surveillance CT in a person without symptoms detects few relapses that the patient would not have brought forward, and both ESMO and NCCN advise against it. Most relapses occur in the first two years and most are announced by a symptom. Late effects of doxorubicin and of any radiotherapy are watched for over decades.