Checkpoint Immunotherapy
Antibodies that release the brakes on T cells (CTLA-4, PD-1) give lasting remissions in melanoma and other cancers; Allison and Honjo won the 2018 Nobel Prize.
Open in the interactive tree →Tumours exploit inhibitory 'checkpoint' receptors on T cells, and antibodies that block them restart the immune attack. The FDA approved ipilimumab (CTLA-4) in March 2011 and pembrolizumab (PD-1) in September 2014, more PD-1 and PD-L1 antibodies followed, and James Allison and Tasuku Honjo received the 2018 Nobel Prize in Physiology or Medicine. In advanced melanoma, the 10-year follow-up of the CheckMate 067 trial (ESMO, September 2024) shows overall survival of 43% with nivolumab plus ipilimumab, 37% with nivolumab alone and 19% with ipilimumab alone.
As of October 2026
Pembrolizumab (Keytruda) was the world's best-selling drug in 2024 with US$29.5 billion in sales, nearly half of Merck's total. A subcutaneous version, Keytruda Qlex, launched in late September 2025 for 38 of the original's solid-tumour indications, and Merck expects 30-40% of patients to switch while the intravenous form loses patent protection later this decade. In CheckMate 067 the combination reached a median overall survival of 71.9 months, the longest in any phase 3 trial of an anti-PD-1 drug.
Open steps
- Personal mRNA vaccine plus PD-1 drug Medium AI leverageIn melanoma, adding a personal mRNA vaccine to pembrolizumab cut recurrence in phase 2 (5-year HR 0.51). A phase 3 reported a positive topline in 2026, without numbers.
- Who will respond? Medium AI leverageMany patients do not respond. AI scores beat PD-L1 and mutation burden on past trial data; proof from prospective trials that they improve treatment choices is still needed.
- How long to treat Low AI leveragePatients are often treated for two years. In DANTE, 80.2% who stopped at 12 months kept disease control versus 87.6% who continued; the trial closed early, underpowered.
- Next-generation PD-1 antibodies Medium AI leverageA PD-1 and VEGF bispecific beat pembrolizumab on survival in a phase 3 run in China (HR 0.73). Does the gain hold in other populations and cancers?
Where AI could help
Medium AI leverage. AI can help choose patients and targets, but only prospective trials can prove that a prediction improves outcomes.
- Predict responders from routine labs, pathology slides and tumour sequencing
- Find tumour neoantigens and new checkpoint targets
- Match combination regimens to tumour profiles
- Flag immune-related side effects early from health records
Shown so far
- In June 2024 NCI and Memorial Sloan Kettering published LORIS in Nature Cancer, a model using tumour mutation burden plus five routine clinical features that predicted response and survival in 2,881 patients across 18 solid tumour types; the authors say prospective studies are still needed. source
Prerequisites
- Antibodies and serum therapy1890Checkpoint drugs are antibodies, a medicine type that began with antibody sera
- Chemotherapy1942
- Genetic engineering1973
- Monoclonal antibodies1975Checkpoint inhibitors are monoclonal antibodies
Unlocks
- Curing canceropen