Why this matters
Prophylactic cranial irradiation has been part of small-cell lung cancer management for decades because older randomized trials showed fewer brain metastases and, in some settings, improved survival.
But those studies largely preceded routine brain MRI surveillance. Modern imaging raises a different question: can frequent MRI detect brain metastases early enough to preserve survival while avoiding the cognitive and treatment-related burden of PCI?
SWOG S1827/MAVERICK is the first large randomized phase III trial to address that question across both limited-stage and extensive-stage SCLC in the MRI era.
The primary result strongly favors MRI surveillance alone for cognitive failure-free survival. But the survival question is not yet fully settled.
Study design
MAVERICK enrolled patients with limited-stage or extensive-stage SCLC who had completed initial therapy and had no brain metastases on baseline MRI.
Patients were randomized to:
- MRI surveillance plus PCI
- MRI surveillance alone
PCI was delivered to 25 Gy in 10 fractions.
MRI was performed every 3 months during year 1 and every 6 months during year 2 in both groups.
Cognitive testing was performed on the same schedule using the Hopkins Verbal Learning Test-Revised, Controlled Oral Word Association, and Trail Making Test.
The primary endpoint was cognitive failure-free survival, defined as time to cognitive failure on any prespecified test or death.
The key secondary endpoint was overall survival noninferiority with MRI surveillance alone. The noninferiority margin was an upper 90% confidence interval of 1.25. Final OS analysis is planned after 190 deaths.
A total of 304 patients were enrolled, with 303 included in the analysis. Approximately 68% had limited-stage disease, and 41% had received immunotherapy with initial treatment.
Among patients assigned to PCI, 77% received hippocampal-avoidance PCI.
Median follow-up among living patients was 22 months.
Key results
MRI surveillance alone significantly improved the primary endpoint of cognitive failure-free survival:
- HR 0.60
- 90% CI 0.46-0.78
- p=0.0005
Cognitive failure-free survival at 6 months was:
- 38% with MRI surveillance alone
- 17% with MRI plus PCI
At 12 months:
- 17% vs 6%
The reduction in cognitive failure was also confirmed when cognitive failure was analyzed separately from death:
- subdistribution HR 0.62
- 90% CI 0.48-0.79
- p=0.001
The benefit appeared consistent across disease stage.
For limited-stage SCLC:
- HR 0.59
- 90% CI 0.43-0.79
For extensive-stage SCLC:
- HR 0.65
- 90% CI 0.38-1.10
There was no significant interaction by stage.
The benefit also did not significantly differ according to receipt of upfront immunotherapy.
Importantly, MRI surveillance remained superior even when compared specifically with patients intended to receive hippocampal-avoidance PCI.
Overall survival
The current OS analysis is preliminary.
At 128 deaths:
- HR for MRI alone vs MRI plus PCI: 0.90
- 90% CI 0.67-1.20
There was no significant difference in overall survival at this analysis.
The upper confidence interval remains below the prespecified noninferiority margin of 1.25, which is reassuring.
However, the protocol-defined final OS analysis will occur after 190 deaths. Therefore, overall survival noninferiority should not yet be considered formally established.
Brain metastases
Avoiding PCI came with the expected tradeoff: more brain metastases.
The cumulative incidence of brain metastases was higher with MRI surveillance alone:
- 6 months: 21% with MRI alone vs 7% with PCI
- 12 months: 30% vs 15%
The subdistribution hazard ratio was:
- 2.19
- 95% CI 1.31-3.64
PCI therefore clearly remained effective at preventing brain metastases.
However, that reduction did not translate into a detectable progression-free survival advantage.
PFS was similar:
- HR 0.96
- 90% CI 0.75-1.23
- p=0.79
The pattern of first progression differed between groups. MRI surveillance produced more initial CNS progression, while PCI was associated with more extracranial-only progression.
Patient-reported outcomes and toxicity
Patient-reported outcomes generally favored MRI surveillance alone.
MRI surveillance was associated with better PROMIS cognitive function and improved overall quality of life at early time points.
No significant difference was seen in the MDASI cognitive factor.
Treatment-related toxicity strongly favored MRI surveillance.
Grade 2 or higher treatment-related adverse events occurred in:
- 41% with MRI plus PCI
- 2% with MRI alone
Grade 3 or higher adverse events occurred in:
- 8% with MRI plus PCI
- 1% with MRI alone
- p=0.004
There was also one treatment-related grade 5 encephalopathy event in the PCI arm.
Interpretation
MAVERICK directly challenges one of the longest-standing preventive radiation strategies in lung cancer.
The trial clearly shows that routine PCI is not required to maximize cognitive failure-free survival when structured MRI surveillance is available.
Patients assigned to MRI surveillance alone were substantially more likely to remain alive without cognitive failure, experienced fewer treatment-related adverse events, and reported better cognitive and quality-of-life outcomes.
The fact that this benefit persisted despite widespread use of hippocampal-avoidance PCI is particularly important. Modern PCI techniques reduce neurotoxicity, but they did not eliminate the advantage of avoiding PCI altogether.
At the same time, MAVERICK does not show that PCI has no biological effect.
PCI approximately halved the incidence of brain metastases at 12 months. The key observation is that this reduction in intracranial disease has not, so far, translated into better PFS or OS.
That distinction is central to interpreting the study.
The most consequential unresolved question remains survival. The current data are reassuring, with an OS hazard ratio of 0.90 and the upper 90% confidence interval currently below the noninferiority boundary.
But the final protocol-defined noninferiority analysis has not yet occurred.
The authors conclude that MRI surveillance alone should become the standard of care. The randomized cognitive, toxicity, and quality-of-life data strongly support that position, but the final survival analysis remains an important outstanding piece of evidence.
Limitations
The overall survival analysis is still preliminary, with 128 of the planned 190 deaths.
Treatment adherence also complicates interpretation. Twenty-eight patients assigned to MRI plus PCI did not ultimately receive PCI, predominantly because they refused treatment.
The primary endpoint combines cognitive failure and death, although a separate competing-risk analysis of cognitive failure itself also favored MRI surveillance.
Median follow-up among surviving patients is 22 months, so longer-term cognitive and survival outcomes remain important.
Finally, the effectiveness of MRI surveillance depends on reliable access to serial brain MRI and timely salvage treatment. These results therefore assume a healthcare setting capable of reproducing the trial's intensive imaging schedule.