Long-Term Outcomes in Cancer Immunotherapy: What the Data Shows

When pembrolizumab was first approved, the clinical question was straightforward: does it work better than what came before? A decade later, with long-term follow-up data now available from multiple large trials, a more nuanced and remarkable picture has emerged - one that centres on the durability of response that immunotherapy can produce in a meaningful subset of patients.

The "tail" on the survival curve

In oncology, survival data is often presented as Kaplan-Meier curves - graphs that show what proportion of patients in a trial are alive at each time point. With chemotherapy, these curves typically follow a downward trajectory that eventually reaches zero: all patients, eventually, die of their disease or treatment complications. With pembrolizumab in responsive patients, something different can happen: the curve flattens at a non-zero level, suggesting a population of long-term survivors whose disease has been durably controlled.

This "tail" on the survival curve is a distinctive feature of checkpoint inhibitor therapy. It does not occur in all patients - or even in a majority - but it occurs in enough patients, in enough tumour types, that it has fundamentally changed the prognosis conversation for eligible patients.

Five-year data from key trials

Long-term follow-up has now been published for several pivotal KEYNOTE trials:

• In advanced melanoma (KEYNOTE-001), approximately 34% of all patients were alive at five years. Among treatment-naive patients, the rate was around 41%. A proportion of these patients had received a defined course of treatment and remained in remission without ongoing therapy.
• In NSCLC with high PD-L1 expression (KEYNOTE-024), around 31% of pembrolizumab patients were alive at five years versus 16% with chemotherapy. This is a disease where five-year survival was rare before immunotherapy.
• In head and neck cancer (KEYNOTE-048), three-year overall survival with pembrolizumab monotherapy in the CPS 20+ population was approximately 22%, compared to 14% with the previous standard of care.

Remission after stopping treatment

One of the most clinically significant findings in long-term immunotherapy data is that some patients maintain remission after their treatment course ends. Current guidance typically involves treating for up to two years in responding patients - after which treatment is stopped and monitoring continues. The data suggests that a meaningful proportion of these patients remain in durable remission beyond that point, without further treatment.

This is qualitatively different from the pattern seen with chemotherapy, where disease control almost invariably depends on continued treatment. It suggests that, in some patients, pembrolizumab effectively re-trains the immune system to sustain an anti-tumour response independently.

What this means in practice

Long-term outcomes data does not predict individual results. What it does is describe the range of what is achievable - and for eligible patients, the upper end of that range includes outcomes that were not possible before checkpoint inhibitor therapy existed. An oncologist can help interpret what the data means in the context of your specific cancer, biomarkers, and clinical situation.