Primary CNS lymphoma

Rituximab + high-dose methotrexate + cytarabine ± thiotepa (MATRix) ×4, then high-dose thiotepa-based chemotherapy with autologous transplant (IELSG32, IELSG43).

High-dose methotrexate with temozolomide, procarbazine or rituximab (e.g. MT-R, R-MP), then maintenance (temozolomide, lenalidomide) or reduced-dose WBRT; ibrutinib-based induction in trials.

Re-induction with methotrexate if durable first remission; ibrutinib, lenalidomide-rituximab, high-dose chemotherapy/ASCT if not done, WBRT; CD19 CAR-T and PD-1 inhibitors in trials or off-label.

A diffuse large B-cell lymphoma confined to the brain, spinal cord, eyes or meninges. The drugs that cure it elsewhere do not reach it: cyclophosphamide, doxorubicin and vincristine cross the blood-brain barrier poorly, so R-CHOP produces responses in the body and not in the brain, and trials of it in this disease were abandoned decades ago. Surgery has no therapeutic role beyond biopsy, and debulking does not help. Corticosteroids shrink the tumour dramatically and dissolve the diagnostic material with it, so steroids are withheld until the biopsy is taken wherever the patient is stable enough to wait. Induction is built around high-dose methotrexate at 3 to 3.5 g per square metre or more, given with leucovorin rescue and urinary alkalinisation, every two to three weeks. MATRix adds cytarabine, thiotepa and rituximab: in the first randomisation of IELSG32 the complete remission rate was 49 per cent with MATRix against 23 per cent with methotrexate and cytarabine alone and 30 per cent with methotrexate, cytarabine and rituximab. Six per cent of patients died of toxicity, so it is for patients fit enough to take it. The contribution of rituximab itself is contested: HOVON 105 randomised it into a methotrexate-based regimen and did not show the benefit that IELSG32's arm B against arm A comparison suggested.

Induction alone relapses, so responders are consolidated. The two options are thiotepa-based high-dose chemotherapy with an autologous stem cell transplant, or whole-brain radiotherapy, and the choice is dominated by what each does to thinking. MATRix/IELSG43 randomised consolidation in the largest trial ever run in this disease: three-year progression-free survival was 78 per cent (95 per cent confidence interval 69 to 85) with transplant against 51 per cent (41 to 60) with non-myeloablative R-DeVIC consolidation, hazard ratio 0.43, with overall survival also improved; the cost was more toxicity, a mean of 14.6 against 9.3 adverse events per patient and five against two fatal serious adverse events. In the earlier IELSG32 trial, whole-brain radiotherapy and autologous transplant were similarly effective, and at seven years the difference was cognitive: patients who had whole-brain radiotherapy lost attention and executive function, while those who had transplant improved in those domains, in memory and in quality of life. Seven-year overall survival was 21, 37 and 56 per cent for the three induction arms, and 70 per cent for patients who received MATRix and went on to consolidation. So transplant is preferred for patients fit enough. Whole-brain radiotherapy is reserved for those who are not, usually at a reduced dose, and is increasingly deferred to relapse.

Age is not by itself a reason to withhold methotrexate: reduced-intensity methotrexate-based combinations, usually with rituximab and an alkylating agent such as procarbazine or temozolomide, are given to patients in their seventies and eighties with attention to renal function, and produce durable remissions in a minority. Whole-brain radiotherapy in older patients carries a high risk of progressive cognitive decline and is generally avoided as first consolidation. At relapse the options depend on the interval. Re-treatment with methotrexate is effective in patients who relapse late, and in the seven-year IELSG32 report it was the only salvage that clearly benefited anyone. Other options are high-dose cytarabine with thiotepa and transplant if not already given, whole-brain radiotherapy if not already given, ibrutinib, which crosses into the brain and has single-agent activity, lenalidomide with rituximab, temozolomide, and a clinical trial. Ocular involvement is treated with systemic therapy, often with intravitreal methotrexate or rituximab added.