How deeply a cancer responds to checkpoint inhibition may tell oncologists more about the risk of progression than the cancer it came from, say researchers.
Achieving a complete rather than partial response to PD-1/PD-L1 blockade appears to be the dominant predictor of durable disease control across melanoma, renal cell carcinoma, and non-small cell lung cancer, according to a large real-world study.
The nationwide Danish cohort of 2127 responders found progression-free survival trajectories were strikingly similar across the three cancers once patients were grouped according to whether they had achieved a complete response (CR) or partial response (PR).
The findings suggest response depth may provide substantially more prognostic information after successful immunotherapy than tumour type or many of the other clinical characteristics traditionally used to assess prognosis.
“Depth of response (complete vs partial) was the strongest independent factor associated with progression risk within each tumour cohort,” the researchers reported.
“These findings suggest that, within the tumour types studied, response durability is primarily associated with response depth rather than other known clinical features, supporting prospective evaluation of response-guided follow-up strategies and interventions designed to deepen responses.”
Published in the European Journal of Cancer, the population-based study included adults with metastatic or unresectable melanoma, RCC, or NSCLC who achieved an objective response to PD-1/PD-L1 monotherapy or, where applicable, dual checkpoint blockade. Patients treated with combinations involving chemotherapy or targeted therapies were excluded.
Of the 2127 responders, 1199 had melanoma, 261 had RCC, and 667 had NSCLC. Complete responses were considerably more common in melanoma, occurring in 48% of responders, compared with 30% of RCC responders and just 11% of those with NSCLC.
But once patients had achieved a CR, their subsequent progression risk looked surprisingly similar regardless of which of the three cancers they had.
However, the difference between complete and partial responders was substantial.
Among melanoma responders, five-year PFS was 74% for patients with a CR compared with just 19% for those with a PR. Five-year overall survival was 87% versus 44%, respectively, while disease-specific survival was 93% versus 53%.
After adjustment, partial response was associated with almost seven times the risk of progression or death compared with complete response (HR 6.76, 95% CI 5.41-8.43). The corresponding hazard ratios were 6.86 for overall mortality and 10.62 for disease-specific mortality.
The pattern was much the same in RCC. Five-year PFS reached 68% among complete responders compared with 24% among partial responders, while five-year OS was 95% versus 52%.
Partial response was again the dominant adverse prognostic factor, carrying a more than fourfold increased risk of progression or death compared with CR (HR 4.69, 95% CI 2.89-7.62).
And in NSCLC, five-year PFS was 67% following CR but only 19% following PR. Five-year OS was 75% and 40%, respectively.
Partial responders had a 4.67-fold greater risk of progression or death than complete responders and a 3.59-fold greater risk of death. Older age and baseline CNS metastases also independently predicted poorer outcomes, but their effect sizes were considerably smaller.
The researchers then compared the different types of tumours in a head-to-head fashion.
Before propensity-score matching, PFS curves among complete responders and among partial responders were already similar across melanoma, RCC, and NSCLC.
After patients were matched for age, sex, ECOG performance status, treatment line, and response depth, the five-year PFS curves were “virtually superimposable”.
Overall survival was also similar between matched melanoma and NSCLC patients. RCC patients showed some later survival advantage over those with NSCLC, which the researchers suggested could reflect slower disease after progression, better access to salvage therapies, or differences in underlying health rather than durability of the original checkpoint inhibitor response.
Traditional baseline prognostic markers added relatively little once response depth was known. In melanoma, for example, age, sex, LDH, BRAF status, ECOG performance status, stage, CNS metastases, treatment line, and checkpoint inhibitor regimen showed no independent association with PFS, OS, or disease-specific survival in the main analyses.
The researchers said the findings could have implications for how oncologists think about surveillance and treatment after a patient has responded to immunotherapy.
“Together, these findings indicate that, within the tumour types studied, response depth is the dominant stratifier of progression risk among responders, while tumour origin and routinely available baseline characteristics contribute limited additional prognostic information,” they wrote.
“This interpretation is consistent with a recent meta-analysis of phase III trials across 11 tumour types in which lower CR rates correlated with higher observed rates of acquired resistance.”
The researchers stressed that the observational study could not establish how frequently patients should be monitored or how long treatment should continue.
They said the pronounced difference in durability between complete and partial responses provided a rationale for prospective trials of response-guided follow-up and strategies designed to deepen partial responses.
Related
They said the type of checkpoint regimen did not appear to add prognostic information once response depth had been achieved. Dual PD-1 plus CTLA-4 blockade in melanoma and RCC increased the likelihood of achieving CR, but was not associated with more durable PFS after response depth was taken into account.
That observation favoured investigating consolidation or other response-deepening strategies rather than simply escalating checkpoint inhibition in patients left with residual disease, the researchers said.
However, there were important limitations to the study, the researchers said. Response was investigator-assessed using routine RECIST imaging without central review, and imaging practices varied between cancers.
Some patients classified as having a PR may also have had a metabolic complete response, while the researchers could not distinguish a near-complete response, such as 80% or greater tumour regression, from less profound partial responses.
Response depth was also based on the best response achieved during follow-up rather than response status at a fixed landmark, creating the potential for time-to-response and immortal-time biases.
An 18-month landmark analysis in melanoma produced similar findings, but the researchers said residual bias could not be excluded. Nor was it clear whether the finding applied beyond these three relatively checkpoint-sensitive cancers.
Despite these limitations, the researchers said that with more than 2000 well-characterised responders and up to five years of follow-up, the study provided a strikingly simple message – that after PD-1/PD-L1 blockade worked, and how completely it worked may matter more than where the cancer started.
“These findings support prospective evaluation of response-guided strategies after objective response, including follow-up approaches and response-deepening interventions for patients with PR, and inform the interpretation of response depth as a clinically meaningful endpoint in immunotherapy research,” the researchers concluded.
