Oropharyngeal Cancer: Staging, Systemic Therapy and De-escalation Trials

Hi All,

Week two of oropharynx! As always, Wednesday mornings @ 7AM, Munchkins for those who can join in person. Some of us may be in Los Angeles for the AAO-HNSF meeting next week, so if you can't make it this Wednesday, send me an email or grab any of us in the OR (ECOG is also in Philadelphia if you wanted to check out what that's all about!).

This week's topic is staging, systemic therapy, and the de-escalation trials for HPV-associated oropharyngeal cancer. Last week we established that these patients do well. This week is about how well, who doesn't ultimately do well, and what happened when cooperative groups tried to take certain aspects of treatment away too rapidly. There's a clear thread through all of it, and it sets up our two TORS weeks at the end of the month.

Staging as a Survival Prediction Tool - ICON-S and the AJCC 8th Edition

(O'Sullivan Lancet Oncol 2016; ICON-S) is a multicenter cohort study of 1,907 HPV-positive oropharyngeal cancer patients from seven centers across North America and Europe, led out of Princess Margaret in Toronto. TL;DR: the 7th edition staging system didn't work for this disease. Five-year overall survival was essentially the same for 7th edition stage I, II, III, and IVA (88%, 82%, 84%, and 81%), and only dropped for stage IVB (60%). The same was true of N category: survival was similar for N0, N1-N2a, and N2b, and only fell for N3 disease (59%). The authors rebuilt the system around what actually predicts survival, collapsing the nodal categories into ipsilateral (N1), bilateral or contralateral (N2), and greater than 6 cm (N3), with stage I as T1-T2N0-N1, stage II as T1-T2N2 or T3N0-N2, stage III as T4 or N3, and stage IV as distant metastasis.

(Lydiatt CA Cancer J Clin 2017) is the summary of how this became the AJCC 8th edition, which created a separate staging algorithm for HPV-associated oropharyngeal cancer and left extranodal extension out of its N category. The board-relevant wrinkle is pathologic staging after surgery: pN is based on the number of positive nodes, with 1 to 4 nodes as pN1 and 5 or more as pN2. Staging keeps evolving, though. (Rai Radiol Imaging Cancer 2026) reviews the AJCC Version 9 update for HPV-associated oropharyngeal cancer, which took effect January 1, 2026. The T and M categories are unchanged (with more explicit anatomic definitions of T4), but extranodal extension is back in the N category: unequivocal clinically or radiographically detected ENE now factors into clinical N staging, and pathologic ENE into pathologic N staging. Imaging-detected ENE has real interreader variability, so expect some healthy debate about it at tumor board. So the practical message for counseling a patient is that most will do well. The ones who don't tend to have bulky or numerous nodes or distant metastases, and occasionally a T4 primary, which is uncommon in this disease. Here's a question to sit with about methodology and interpretation of studies: ICON-S found that having five or more nodes didn't change survival in its clinically staged cohort, yet node count is the basis of pathologic N staging. Why might the number of nodes matter more once you've actually counted them in a neck dissection specimen? Hint: the answer is selection bias.

Systemic Therapy - Cisplatin Is Still the Standard

If a patient needs concurrent systemic therapy with radiation, it should be cisplatin. Two randomized trials, published side by side in the Lancet, settled this. (Gillison Lancet 2019; RTOG 1016) randomized 849 patients with HPV-positive oropharyngeal cancer across 182 centers in the US and Canada to 70 Gy with either cisplatin or cetuximab. Cetuximab failed non-inferiority for overall survival: 5-year overall survival was 77.9% with cetuximab versus 84.6% with cisplatin, progression-free survival was 67.3% versus 78.4%, and locoregional failure nearly doubled (17.3% vs 9.9%). And cetuximab didn't even buy less toxicity, since rates of moderate to severe acute and late toxicity were similar between arms.

(Mehanna Lancet 2019; De-ESCALaTE HPV) asked the same question in 334 low-risk patients (non-smokers or fewer than 10 pack-years) across 32 centers in the UK, Ireland, and the Netherlands, this time with toxicity as the primary endpoint. Severe toxicity was identical at 24 months (4.8 events per patient in each arm), while 2-year overall survival was 97.5% with cisplatin versus 89.4% with cetuximab, and 2-year recurrence was 6.0% versus 16.1%. TL;DR: cetuximab is worse and no gentler. For patients who truly can't receive cisplatin, carboplatin-based regimens are the usual nonsurgical fallback. Carboplatin with 5-FU has randomized data in oropharyngeal cancer (Denis JCO 2004; GORTEC 94-01), though that trial predates the HPV era and compared it to radiation alone rather than to cisplatin. Primary surgery also affords high cure rates without sacrificing survival in these patients. This is often one of the variables that shifts us towards surgery in cases where there would otherwise be an even argument for surgical versus nonsurgical therapy.

The De-escalation Story - NRG HN002 and HN005

With cisplatin locked in, NRG Oncology tried the other lever: radiation dose. (Yom JCO 2021; NRG HN002) is a randomized phase II trial of 306 patients with p16-positive, ≤10 pack-year disease (T1-T2N1-N2b or T3N0-N2b by the 7th edition) comparing 60 Gy with weekly cisplatin against 60 Gy alone. Only the chemoradiation arm met the prespecified bar to move to phase III, with 2-year progression-free survival of 90.5% versus 87.6% for radiation alone. TL;DR: even at a reduced dose, you couldn't drop the chemotherapy.

That set up (Yom Int J Radiat Oncol Biol Phys 2024; NRG HN005), a phase II/III non-inferiority trial in 382 non-smoking p16-positive patients (cT1-T2N1 or cT3N0-N1 by the 8th edition) randomized to standard 70 Gy with cisplatin, 60 Gy with cisplatin, or 60 Gy with nivolumab in place of cisplatin. Both de-escalated arms crossed the futility boundary at the interim analysis and the trial closed. Two-year progression-free survival was 98.1% in the control arm versus 88.6% for 60 Gy with cisplatin and 90.3% for 60 Gy with nivolumab. Those are good numbers in absolute terms, but the control arm did remarkably well, and that is exactly what makes non-inferiority so hard to prove in a disease with this prognosis. In the nonsurgical setting, standard-dose cisplatin chemoradiation remains the standard of care, and swapping cisplatin for a PD-1 inhibitor is not ready for prime time. These results were presented as a plenary at ASTRO 2024, and a full manuscript is still worth watching for. IMPORTANT TIME SENSITIVE ALERT! Sue Yom is super cool and is literally giving a lecture at Penn this Tuesday at 4PM if you want to meet her and probably hear about this in much more detail... (see announcements below for full details!).

Where De-escalation Goes Next

The real-world data are now lining up with the trials. (Shin Front Oncol 2026) looked at nearly 15,000 HN005-eligible patients in the National Cancer Database treated with definitive radiation from 2018 to 2022. Only 7.5% received a reduced dose (most often at academic centers), and their 3-year overall survival was worse (82.1% vs 88.9%). It's a retrospective database study with all the selection bias that implies, but the authors' conclusion is the right one: reduced-dose radiation should stay on a trial until we have better biomarkers for selecting patients.

So where does de-escalation go from here? (Topf Otolaryngol Clin North Am 2025) argues that after HN005, the most viable path to de-intensifying treatment for HPV-associated oropharyngeal cancer runs through surgery, using pathologic staging to tailor adjuvant therapy (ECOG 3311, AVOID, DART, and ORATOR2 among others). That is where we will pick up in two weeks, including the Penn trials that have been doing exactly this.

Thanks for taking the time to read this and looking forward to discussing on Wednesday!

Have a great week!

Rob