Why this matters
Does dividing cisplatin into weekly doses actually make definitive head and neck chemoradiotherapy less toxic?
NRG-HN009 tested that question directly. Its phase II endpoint measured the overall burden of treatment-related grade 3-4 adverse events, rather than focusing on an individual complication.
The newly reported p16-negative cohort did not demonstrate the expected toxicity reduction with weekly cisplatin. The presentation also confirmed that the previously reported p16-positive cohort had not met the criteria for phase III continuation.
The clinically important distinction is that weekly dosing changed the toxicity profile without demonstrating a lower overall severe-toxicity burden.
Study design
NRG-HN009 was designed as a randomized phase II/III comparison of definitive radiotherapy with two cisplatin schedules:
- Weekly: cisplatin 40 mg/m² with 70 Gy RT.
- Every three weeks: cisplatin 100 mg/m² with 70 Gy RT.
The trial evaluated separate p16-positive and p16-negative/non-oropharyngeal cohorts. The latter included laryngeal, hypopharyngeal, and p16-negative oropharyngeal cancers.
The phase II primary endpoint was the T-score, defined as the number of treatment-related grade 3-4 adverse events during treatment or within 180 days afterward. This is an event-count measure, not the percentage of patients experiencing any severe toxicity.
For the p16-negative cohort to proceed to phase III, weekly cisplatin had to demonstrate a lower mean T-score using a one-sided significance threshold of p<0.10. The difference in estimated six-month locoregional failure, weekly minus three-weekly, also had to be no greater than eight percentage points.
The new cohort included 234 randomized patients, with 116 assigned to weekly and 118 to three-weekly cisplatin. The displayed toxicity distributions included 224 treated patients: 115 and 109, respectively.
Median age was 60 years. Laryngeal cancers accounted for 57% of the cohort, oropharyngeal cancers for 28%, and hypopharyngeal cancers for 15%. Ninety-one percent completed follow-up through six months.
Key results
The primary toxicity endpoint was negative.
Mean T-scores were 2.48 with weekly cisplatin and 2.23 with three-weekly cisplatin. The corresponding 95% confidence intervals were 1.97-2.99 and 1.81-2.65.
The mean T-score ratio, weekly versus three-weekly, was 1.11, with a 90% upper confidence bound of 1.33 and one-sided p=0.77.
Weekly cisplatin therefore did not demonstrate the required reduction in overall acute grade 3-4 toxicity. The numerically higher mean score should not, however, be presented as proof that weekly treatment was more toxic overall.
Individual toxicities moved in different directions.
The following selected grade 3-4 adverse events differed by at least five percentage points:
| Adverse event | Weekly cisplatin | Cisplatin every three weeks |
|---|---|---|
| Nausea | 4% | 12% |
| Dehydration | 1% | 9% |
| Acute kidney injury | 3% | 11% |
| White blood cell decrease | 28% | 14% |
| Hypomagnesemia | 8% | 2% |
These are descriptive, event-specific comparisons. Weekly treatment had fewer reported renal and selected gastrointestinal toxicities, but more white blood cell and magnesium decreases.
Early locoregional failure warrants attention.
In the p16-negative cohort, six-month locoregional failure was 9.6% with weekly cisplatin versus 4.6% with three-weekly cisplatin.
The absolute difference was 5.0 percentage points, with a reported 95% CI of 1.0-8.9 percentage points.
The observed difference was below the protocol's eight-point cutoff for the early continuation rule. Nevertheless, the higher failure estimate with weekly treatment is clinically relevant and requires longer follow-up. Meeting an early screening rule is not equivalent to establishing long-term efficacy noninferiority.
The presentation also summarized the p16-positive cohort.
Among 241 randomized p16-positive patients, the investigators reported similar overall acute toxicity scores between schedules, again without meeting the criteria for phase III continuation.
Six-month locoregional failure in that cohort was 2.7% with weekly versus 3.5% with three-weekly cisplatin. These results should remain separate from the p16-negative findings rather than being treated as a pooled efficacy comparison across all 475 patients.
Interpretation
NRG-HN009 challenges the assumption that weekly cisplatin necessarily means a gentler overall treatment course.
The distinction between individual adverse events and aggregate toxicity is central. Fewer episodes of acute kidney injury, dehydration, and nausea are relevant, but they do not establish an overall toxicity advantage when other severe events become more frequent.
Conversely, a similar aggregate T-score does not make the two schedules interchangeable from a patient's perspective. Counting events does not fully capture their duration, reversibility, or functional consequences. A transient laboratory abnormality and a persistent complication may contribute to the same numerical endpoint while having different clinical implications.
The investigators specifically noted that longer toxicity follow-up and analysis of pretreatment and six-month audiograms remain important. The present report therefore should not be used to settle the comparison of long-term hearing outcomes or chronic toxicity.
The efficacy interpretation also requires restraint. The p16-negative cohort had a higher six-month locoregional failure estimate with weekly treatment, whereas the p16-positive cohort did not show the same direction. These are early cohort-specific results, not mature evidence of equivalent or inferior long-term cancer control.
The reason the trial did not meet its phase III continuation criteria was the absence of the required toxicity advantage. That is a meaningful negative result, but it is not a completed phase III verdict on which cisplatin schedule provides better survival.
Limitations
This report concerns phase II acute toxicity and six-month locoregional outcomes. Mature disease-control, survival, and chronic toxicity results are not provided.
The detailed p16-negative toxicity distributions include treated patients rather than every randomized participant. The final report will need to clarify how treatment non-initiation and incomplete follow-up were handled across analyses.
The T-score measures the number of severe adverse events, not their cumulative duration or patient-reported burden. Similar scores therefore should not be interpreted as equivalent quality of life.
The findings also concern the definitive 70 Gy regimens and disease populations studied. They do not establish the same comparison for other cisplatin doses, postoperative treatment, or different radiotherapy schedules.