AREST
ForpT1-2N0 oral SCC post adequate resection, ≥1 intermediate-risk feature
HR 0.52
95% CI 0.30-0.91, p=0.02; 3yr LRFS 89.2% vs 80.9%
TL;DR3yr LRFS 89.2% vs 80.9% with adjuvant RT after adequate resection of intermediate-risk pT1-2N0 OSCC; HR 0.52, no OS gain.
Transfer hinges on the surgery: benefit was shown only after margins ≥5mm and a ≥16-node level I-III dissection, so a lesser neck operation is not the population studied. Per-protocol the effect strengthens (HR 0.43, 91.1% vs 80.9%), and competing-risk LRF ran 10.6% vs 18.9%. Dose was 60Gy/30fx to bed plus at-risk nodes.
In an intermediate-risk pT1-2N0 oral tongue resection, this is the first randomised evidence supporting adjuvant RT for loco-regional control, with no survival gain shown; buccal mucosa benefit looked smaller and stays exploratory, and node-positive or close-margin disease sits outside the trial.
The target was the resected bed plus at-risk neck at 60Gy/30fx, so this transfers directly to standard post-op practice with no unusual technique requirement. Per-protocol the estimate strengthens to HR 0.43 (91.1% vs 80.9%), so the ITT figure likely understates a completed course. No toxicity data to set against it.
RT was only tested after margins ≥5mm and a ≥16-node level I-III dissection, so the referral question sits downstream of node yield and margin width. A neck below that bar is outside the evidence entirely, and the features driving the referral (DOI ≥5 to ≤10mm, PNI, LVE, poor differentiation) come off the specimen.
| Arm | 3yr LRFS (95% CI) | HR (95% CI) | p |
|---|---|---|---|
| Adjuvant RT | 89.2% (84.3-93.3) | 0.52 (0.30-0.91) | 0.02 |
| Observation | 80.9% (74.6-86.1) | reference | n/a |
8 details 3 trials watching
Multicentre open-label phase III RCT from India, 1:1 randomisation, N=392 (191 adjuvant RT, 201 observation), stratified by oral cavity subsite, PNI/LVE and differentiation. Median follow-up 47.2 months (IQR 30-59.4).
pT1-2, pN0 OSCC after adequate surgery, defined as clear margins ≥5mm plus at least ipsilateral level I-III neck dissection yielding ≥16 nodes. At least one intermediate risk factor required: DOI ≥5 to ≤10mm, PNI, LVE, or poor differentiation. Baseline characteristics reported as balanced.
60Gy in 30 fractions over 6 weeks to the resected tumour bed and the at-risk neck nodal region. Technique, target volume detail and dose constraints are not reported in the source.
Primary: loco-regional recurrence-free survival, from randomisation to first documented local and/or regional recurrence of the index cancer. Kaplan-Meier 3-year point estimates with log-rank comparison; DFS and OS secondary.
Primary endpoint met on both ITT and per-protocol analysis, and reproduced in the competing-risk analysis. DFS and OS did not differ between arms.
| Analysis | Adjuvant RT | Observation | HR (95% CI), p |
|---|---|---|---|
| 3yr LRFS, per-protocol | 91.1% | 80.9% | 0.43 (0.23-0.80), p=0.01 |
| Cumulative LRF, ITT | 10.6% | 18.9% | 0.52 (0.30-0.91), p=0.021 |
| Cumulative LRF, per-protocol | 8.7% | 18.9% | 0.43 (0.23-0.79), p=0.007 |
The randomised adjuvant evidence in resected head and neck cancer (EORTC 22931, RTOG 9501) tested chemoradiation against radiation in high-risk disease defined by positive margins and extranodal extension, leaving the intermediate-risk indication to retrospective series, which the abstract itself names as the basis for the debate. This is the first randomised test of that indication in an adequately resected node-negative cohort.
The observation arm reached 80.9% 3-year LRFS against the 70% the sample size assumed, so the trial ran event-poor and the DFS and OS comparisons are underpowered rather than reassuring. No toxicity, xerostomia or quality-of-life data appear in the source, so the price of the local control gain is unquantified. The subsite effect is exploratory.
The whole content of the result is loco-regional control at a median 47.2 months, with no survival separation, so the decision turns on how the reader values preventing a recurrence in a cohort whose failures are visible and often salvageable. The exploratory oral tongue over buccal mucosa split, if it replicates, would narrow the indication rather than extend adjuvant RT to every intermediate-risk resection.
CONSORT flow
First randomised test of an indication previously grounded in retrospective data; primary endpoint met, but open-label and the gain is loco-regional only, no DFS or OS.
- Does the oral tongue vs buccal mucosa subsite effect replicate? recruiting A Study of Radiation Therapy After Surgery in People With Oral Tongue Squamous Cell Carcinoma Phase 2n=24 · primary completion 2027-06 · pT1-2 pN0-2b oral tongue only, post-op IMRT
- Toxicity and QoL cost of the loco-regional control gain recruiting A Study of Radiation Therapy After Surgery in People With Oral Tongue Squamous Cell Carcinoma Phase 2n=24 · primary completion 2027-06 · post-op IMRT sparing tongue site; QoL questionnairesnot yet A Study Evaluating the Contribution of Non-Coplanar Beam Arrangement in Reducing Toxicity in Radiotherapy for Upper Aerodigestive Tract Cancers Phase NAn=70 · primary completion 2028-02 · non-coplanar vs coplanar EBRT, OAR dose + xerostomia
- Whether longer follow-up reveals any survival difference
📚 Sources · 🐦 1 tweet · 📄 1 paper
#ASCO26
— Dr Rishabh Jain (@DrRishabhOnco) May 27, 2026
🗣️ The AREST trial tackles one of the biggest gray zones in oral cavity cancer.
After adequate surgery in pT1-2N0 OSCC with intermediate-risk features:
✅ Adjuvant RT improved loco-regional control
❌ No OS benefit observed
3-year LRFS:
🔹 89.2% vs 80.9%
🔹 HR 0.52… https://t.co/qsALPFX032 pic.twitter.com/F4XzTMETih
Abstract
The longer read
The value here is not the direction of the result, which retrospective series have pointed at for years, but that the question was finally randomised in a population defined tightly enough to interpret. The load-bearing design choice is the definition of adequate surgery: clear margins of at least 5mm and at least an ipsilateral level I-III dissection yielding 16 or more nodes. Much of the retrospective literature on intermediate-risk features is contaminated by inadequate necks and close margins, where an apparent benefit of adjuvant radiotherapy may be salvage of an incomplete operation rather than treatment of occult residual disease. By fixing surgical quality first, this trial isolates the question a radiation oncologist actually faces at the post-operative tumour board.
The most interesting number in the abstract is one the headline graphic does not lead with: the observation arm was assumed at design to reach a 3-year LRFS of 70%, and it reached 80.9%. That gap cuts in two directions. It means this surgical cohort did better than the historical series the indication was built on, which further undercuts the retrospective evidence base. It also means the trial accrued fewer events than it was sized for, so the primary endpoint was met on a thinner event count than planned and the disease-free and overall survival comparisons are correspondingly weak. Reading no survival benefit as a positive finding of equivalence would be a mistake; the honest reading is that this trial was never positioned to detect one.
That leaves a result whose entire content is loco-regional control, and the reader has to decide what that is worth. The competing-risk analysis is the cleaner framing than the Kaplan-Meier estimate, because death is a real competing event in this population, and it puts loco-regional failure at 10.6% with radiation against 18.9% without. The clinical trade is six weeks of radiotherapy to the tumour bed and neck set against that failure rate. What the abstract does not supply is the other half of the trade. There is no toxicity, xerostomia, swallowing or quality-of-life reporting, and in an oral cavity cohort with long expected survival, late RT morbidity is exactly the variable that decides individual cases. Without it the result supports the indication but cannot price it.
The exploratory subsite finding is the part most likely to change how the indication is applied and the part least safe to act on. Oral tongue was reported to derive greater benefit than buccal mucosa, but no subgroup hazard ratios appear in the source, subsite was a stratification factor rather than a powered comparison, and buccal mucosa carries a different failure pattern and a different relationship between depth of invasion and margin adequacy. If it replicates, the indication narrows and buccal patients with a single intermediate feature become reasonable candidates for observation. If it does not, the trial applies as randomised, across intermediate-risk subsites.
For the primary result to be wrong, the observation arm would have to have been surveilled differently or its failures captured differently from the irradiated arm, which an open-label design cannot formally exclude. Loco-regional recurrence in the oral cavity is a hard and visible endpoint, though, which makes that the weakest objection available and the survival question the one genuinely left open.