TRIDENT NCT04471844
ForndGBM (WHO 2016), post maximal safe resection or biopsy, pre-RT/TMZ
HR 0.953
95% CI 0.82-1.10, p=0.52; mOS 17.7 vs 17.5 m, not met
TL;DRNo OS gain starting TTFields with RT/TMZ vs at maintenance TMZ in ndGBM: mOS 17.7 vs 17.5 m, HR 0.953, p=0.52.
The RT read is toxicity without payoff: wearing TTFields through RT/TMZ raised device-related AEs to 76.2% vs 59.2% (mostly grade 1-2 skin) with no OS, PFS or ORR difference. RT dose, fractionation and array handling during RT are not reported in source, so the trial gives no reason to build arrays into the chemoRT course.
In ndGBM pts about to begin RT/TMZ after maximal safe resection or biopsy, this does not support moving TTFields start forward from maintenance TMZ; it does not extend to selecting MGMT-methylated pts for early start, which rests on a post hoc subset.
Wearing TTFields through RT/TMZ raised device-related AEs to 76.2% vs 59.2% (mostly grade 1-2 skin) with no OS, PFS or ORR difference. RT dose, fractionation and array handling during RT are not reported in source, so nothing here argues for building arrays into the chemoRT course.
Sequencing stays as is: TTFields with adjuvant TMZ gave mOS 17.5 m vs 17.7 m with early start (HR 0.953, p=0.52). The MGMT-methylated signal (HR 0.683, p=0.03) comes from a post hoc per-protocol-usage subset with no n reported, so MGMT status is not yet a selector for earlier start.
7 details
Randomized, open-label, global phase 3, 1:1, randomized before RT/TMZ. N=981, accrued Dec 2020 to Jan 2024. Stratified by MGMT promoter methylation status and extent of resection.
Adults with ndGBM (WHO 2016 classification), recovered from maximal safe resection (biopsy allowed), KPS threshold 70 (inequality sign not rendered in source), RT/TMZ planned <8 weeks after surgery. Median age 60 (22-86), 65.1% male, 38.9% MGMT methylated, 5.4% IDH mutant.
TTFields (200 kHz) started concurrently with RT/TMZ (Early Start, n=492) or with adjuvant TMZ (Maintenance Start, n=489). TTFields recommended until second disease progression in both arms. Median device use 7.7 m vs 6.1 m.
RT with TMZ in both arms. Dose, fractionation, target volume and array management during RT are not reported in source.
Primary: overall survival. Secondary: PFS, 1-, 2- and 3-year survival rate, ORR, safety. Kaplan-Meier with stratified log-rank test.
Primary not met in ITT (see table). PFS and ORR not different between arms, values not reported in source.
| Endpoint | Early Start (with RT/TMZ) | Maintenance Start (with adjuvant TMZ) | Effect |
|---|---|---|---|
| mOS (95% CI) | 17.7 m (16.0-18.8) | 17.5 m (15.9-19.0) | HR 0.953 (95% CI 0.82-1.10), p=0.52 |
| 1-year survival | 70.9% | 72.0% | NS |
| 2-year survival | 33.9% | 31.6% | NS |
| 3-year survival | 22.5% | 18.4% | NS |
| Device-related AEs | 76.2% | 59.2% | mostly grade 1-2 skin |
| Median TTFields duration (range) | 7.7 m (0-49) | 6.1 m (0-43) | not reported |
Device-related AEs 76.2% Early Start vs 59.2% Maintenance Start, mostly grade 1-2 skin-related. Grade 3+ rates and RT interruption not reported in source.
EF-14 established TTFields added to maintenance TMZ after chemoRT, which the source names as current standard and which is the control arm here. TRIDENT asks only whether starting earlier adds to that, and in ITT it does not.
The between-arm exposure contrast is small (median use 7.7 m vs 6.1 m) and ~25% never started maintenance, which dilutes a timing effect. The MGMT signal conditions on device usage after randomization, with no subgroup n, interaction test or ITT MGMT-methylated result in source.
A timing-null result, not a device-null one. The post hoc MGMT-methylated finding (HR 0.683, p=0.03) is a hypothesis for a prospective trial, and the abstract's "clinically meaningful" framing outruns a per-protocol-usage subset.
CONSORT flow
Randomized phase 3 with prespecified OS primary not met, so maintenance-start TTFields stands. MGMT-methylated benefit is post hoc in a per-protocol-usage subset, not grounds for change.
- Does early TTFields improve OS in MGMT-methylated ndGBM when tested prospectively?
- Scalp dose and RT interruption with arrays worn during RT/TMZ
📚 Sources · 📄 1 paper
Abstract
The longer read
The question here was narrow, and the result should be read at the same width. TRIDENT did not test whether TTFields work in ndGBM. Both arms received the device, with the control arm following the maintenance schedule that EF-14 established and that the abstract itself describes as standard. The only thing randomized was whether the arrays went on at the start of RT/TMZ or at the start of adjuvant TMZ. A flat OS comparison therefore says nothing for or against the device, and should not be recruited into either side of that older argument.
As a test of timing, the trial is large and the primary endpoint is the right one. OS is insensitive to the open-label design, which matters more for the PFS, ORR and device-attributed AE comparisons. The ITT estimate of HR 0.953 with a 95% CI of 0.82 to 1.10 excludes a large benefit from early start but leaves room for a modest one. Two features make a modest effect hard to detect. The exposure contrast the randomization actually produced was small, with median device use of 7.7 m vs 6.1 m. And ~25% of pts across both arms never started the maintenance phase, so a meaningful share of the control arm likely had little or no device exposure while the experimental arm had at least the chemoRT window. That should have favored Early Start if anything, and it still did not separate the curves.
The MGMT-methylated finding needs to be handled with care. It is conditioned twice: first on per protocol device usage during RT/TMZ, a post-randomization behavior that tracks with fitness and with not progressing early, and then on methylation status. The abstract does not state how many pts are in it, how the comparison group was defined for an arm that by design wore no device during RT, or whether an interaction test or any multiplicity control was applied. Medians of 33.8 m vs 23.8 m in both groups sit far above the ITT medians of 17.7 m and 17.5 m, which is itself evidence of how strongly the subset was selected. MGMT status was a stratification factor, so an ITT comparison within methylated tumors would be the cleaner analysis, and it is not in the source.
For a radiation oncologist the practical read is that the abstract supports feasibility but supplies none of the parameters needed to act on it. RT dose and fractionation, how arrays were managed around daily treatment, scalp dose, treatment interruptions and grade 3+ skin toxicity are all unreported. What is reported is a higher device-related AE rate, 76.2% vs 59.2%, mostly grade 1-2 skin. The WHO 2016 eligibility also admitted 5.4% IDH-mutant tumors, which current classification would not call glioblastoma. Maintenance start remains the reference schedule. Early start in MGMT-methylated disease is a hypothesis that would need its own prospective trial to be wrong or right.