Head and Neck Cancer: Elective Nodal De-Escalation
A New Era in elective nodal dosing is replacing a 60+ year standard.
For over six decades, 50 Gy has been the cornerstone of head and neck radiation for uninvolved nodal areas. I believe the data now points lower.
Quick note: It’s been a while. I know :) But data in this particular region of our field has shifted in the past few years. It appears to change something that has stood the test of time for decades. So I write this as a one place reference document for me - and hopefully a few residents, maybe some physicians out there in practice, and perhaps even a few patients looking to reduce treatment toxicity.
As always, investigate and do your research. I did dig a bit and found nothing concerning in our literature regarding reduction in elective nodal dose. And with that…
A Bit of History:
I was blessed to train at MD Anderson. That training shaped my head and neck experience and helped me to later build a pretty busy, relatively large volume head and neck practice for the first 17 or so years after leaving training.
In residency at MD Anderson, we’d have at least 20-30 or so under beam that were “mine”. I believe I had 3-4 eight (or twelve) week rotations in a dedicated head and neck setting with a brilliant team - collectively the brightest group of people I’ve ever worked with. Since that time, I’ve had a strong interest in head and neck treatment. In fact, this patient cohort is why I moved for protons. I often describe treatment to that site as “a life altering experience” and I envisioned technology as a path to do better.
Gilbert Fletcher MD (In Memoriam from RSNA, Lester Peters Dedication) started the program at MD Anderson and was chairman from 1948 - 1981. Among other accomplishments, he worked on the construction of the first dedicated clinical Cobalt 60 machine and helped develop the Fletcher-Suit-Delclos applicator for gynecologic cancers. I arrived two chairmen later but clearly within the large imprint of his legacy.
Dr. Fletcher pioneered the concept for a subclinical dose and established that subclinical necks can be treated to a lower dose of 4500-5000 cGy. Here is his paper from 1972. (REF - Click PDF for file - Note: a single author - times have certainly changed.)
The research supporting this paper dates back to the 1950’s and 1960’s. Based on his work, this became the worldwide standard that has remained in place for over six decades. An amazing accomplishment. Truly one of the founding fathers of our specialty.
But as we hit 2025, I think more and more data supports that we can accomplish equal tumor control outcomes with less dose and therefore, less toxicity. Today, we’ll look at that data in a bit more depth and see if we can’t tweak what he developed back in the 50s.
April 15th, 2025: UPGRADE-RT Trial Landed.
(UPGRADE-RT: REF)
I’ll start here because it was the recent impetus for the article. It appears to be a “final straw” from my perspective that should make us re-assess and move with more confidence towards adjusting these elective doses from well over sixty years ago.
This Dutch trial compared “standard” dose in the elective neck to a “reduced” dose in the elective neck. There is quite a bit of nuance and details to understand as the dosing and treatment varies from US standards.
~75% were HPV negative with 60% Stage III or IV (so advanced, largely HPV-).
It was accelerated WITHOUT any chemotherapy.
High dose level was 68 Gy in 34 fractions over 5.5 wks - 6fx / wk
An optional intermediate level of 60 Gy in 34 fractions was available for intermediate-risk nodes in dose reduced arm (used in 20% of cases).
Low risk elective regions were treated to 50 Gy or 43 Gy. (so 1.26 Gy per fraction in the low dose arm - accelerated to 5.5 wks without chemo EQD2 ~34 Gy).
The randomization was 2:1 towards the low dose arm - 300 randomized, 295 included in the publication. The primary outcome was the normalcy of diet score at 1 year (REF for diet scoring).
In simple terms, it worked - kind of. The 2-year recurrence rate in electively irradiated nodes was 4.9% in the dose reduced arm and 4.3% in the control arm. There was a less than 2% risk of failing in the lower dose regions. So it controlled the disease.
That said, the primary trial outcome was negative - no difference in normalcy of diet. To me, primary outcomes still matter a lot and should be highlighted. Here it was negative for any difference.
Feeding tube rate was about 1/2 in the reduced dose arm (19.2% vs. 10.7%) and patient reported saliva metrics were “improved” on QoL surveys with “p value of 0.006!” But below are the curves for that difference:
The main difference at 3 months. Curves that appear much more similar than the p value might suggest. Again, the primary chosen toxicity endpoint showed no difference with both arms right around 92% diet normalcy at 1 year.
But from a safety aspect, there is no evidence of an increase in failures within the lower dose elective neck. Here is Overall Regional Control.
The trial is open access! Lots of good KM curves. Too many to include here.
Trial 2: Dose De-escalation to the Elective Nodal Sites in Head and Neck Cancer
Back in 2020, this randomized trial was released. It landed at a crazy time in the world and very likely received far less attention than it deserved. It is out of Belgium and randomized 200 patients to either an elective nodal dose of 40 Gy or 50 Gy equivalent dose.
The primary outcome was radiation-induced dysphagia and the secondary endpoint was tumor control.
The 5-year relapse rate was 14.0% in the 40 Gy arm and 7.5% in the 50 Gy arm. The p value says not significant (p=0.10), but it clearly isn’t in the direction that anyone would like to see. That said, 9 of the 13 recurrence were felt to be in the high dose PTV and 2 were beyond the fields. So only 2 in the 40 Gy region. Fortunately, the OS data is supportive of lower doses with a 5-year OS in the 40 Gy arm of 56.5% and 49.6% in the 50 Gy arm. p=0.56, but “leaning towards” the low dose.
The trial did show some evidence favoring the lower dose arm in some toxicity as noted in the abstract but the primary outcome of the trial was negative.
Results: A trend toward less decline in QoL during treatment was observed in the 40 Gy arm compared to the 50 Gy arm. Statistically significant differences for global health status, physical functioning, emotional functioning, speech problems, and trouble with social eating in favor of the 40 Gy arm were observed. A clinically relevant better outcome in the 40 Gy arm was found for physical functioning at the end of therapy.
And I think that is why people have waited. Yes - it probably appeared ok, but relapse rate trended in the wrong direction and wasn’t a clear home-run. Pair that with a negative primary QoL endpoint and many land on: possibly valid, but reasonable to be rather cautious and see more data.
Broader Context: Backing up a Step
But today, I believe one can make the case that there are two randomized trials demonstrating safety. We have randomized about 500 people looking at lower doses and there is no clear safety signal that this is wrong.
So two randomized trials both appearing quite safe with lower elective nodal doses. In the US a move to 40 Gy in 20 fractions makes sense. We give concurrent chemotherapy rather routinely. At 40 Gy, you are higher than the Dutch trial and concurrent chemotherapy should lessen any risk of de-escalation.
Certainly, it is not absolutely clean - rarely in medicine are things cut and dry. In the Belgium trial, there were a few more region relapses (due to failures within the high dose region). Not statistically significant, but worth noting. And in the Dutch trial there are many variances from practice standards between the US and Netherlands.
Fortunately, the UPGRADE-RT trial does offer more flexibility in covering more “questionable” disease - a not uncommon occurrence in planning these cases. I think it adds to both control and flexibility of any transition lower.
Here are the exact definitions used:
High-risk LNs comprised overtly metastatic LNs and were identified by (1) positive cytology or (2) central necrosis on imaging or (3) increased FDG uptake (≥2× background).
Intermediate-risk LNs comprised LNs with limited macroscopic tumor volume and were identified by (1) summed long- and short-axis diameter ≥17 mm and (2) increased FDG uptake (≥1.5 and <2× background).
Low-risk LN volumes comprised the elective LN regions including all LNs not classifying as high- or intermediate-risk.
So of the 500 patients randomized, about 40 total patients (300 were reduced dose IMRT) had an intermediate dose volume. I’m not certain how much this contributes to the success of the UPGRADE-RT data, but it certainly adds some flexibility. Again, in the Belgium study, only 2 failures were within the reduced low dose region.
HPV+ Data Continues to Mature in the Same Direction
First off - a reminder for HPV+ disease:
Adjusting treatment for gross disease or exchanging the cisplatin requirement has failed - REPEATEDLY.
We have at least three randomized prospective trials showing we must be cautious even in the setting of HPV+ disease within the actual disease. Here are three de-escalation trial failures:
NRG HN005: 70 with Cis vs. 60 with Cis vs. 60 Gy with Nivolumab
RTOG 1016: 70 Gy with Cetuximab vs. Cis
De-ESCALaTE HPV: 70 Gy with Cetuximab vs. Cis
It’s really one of the only sites where we have three failed “non-inferiority” trials. Quite a sad accomplishment honestly. And I think this data, likely, is in part for the confusion / caution related to lower elective nodal dosing.
Mentally separate de-escalation of the primary from the elective neck.
On the elective nodal dose for HPV+ disease, there is interesting / compelling data out of Memorial Sloan Kettering. It mirrors the above success in lowering the elective nodal dose. Here is there study from January 2022:
Evaluation of Substantial Reduction in Elective Radiotherapy Dose and Field in Patients With Human Papillomavirus–Associated Oropharyngeal Carcinoma Treated With Definitive Chemoradiotherapy (REF).
(Note: they have a second similar yet different approach using hypoxia and a FMISO PET scan - that is a different approach and not ready for widespread use in my opinion - keep them separate mentally).
This publication included 276 patients: 31% with T3-T4 disease and 23% with N2-N3 disease. Local control is 97%. Stepwise planning - initial to 30Gy to comprehensive volume and then boost disease to 70Gy.
Progression free at 88% with OS at 95%. One failure in the low dose region - that single failure was retrospectively thought to have had disease (but that is risk with this approach). 87 of 276 (around 31%) did have an intermediate dose adjacent to gross disease treated to 50Gy. Per report, mainly in year one during the transition period. (so again less than 1/3 and assuming now well less than 1/4).
To me, this dataset is blazingly consistent with the new UPGRADE-RT data. Clearly this study is retrospective, but in my conversations with the members of the team, it apparently is up to around 500 patients and has become their “standard” approach. The series hasn’t been formally updated, but they haven’t seen any safety signals. Again trust in full the publication data and less my conversations, but multiple physicians have confirmed this as their continued approach (resident project idea - simple update - we need more data in this area to move the SoC broadly).
When paired with the above two randomized trials, I believe the they argue that HPV+ elective dose is appropriately ~30 Gy. In the UPGRADE-RT trial, the effective dose is thought to be 34 Gy at 2 Gy equivalent without chemotherapy (low dose per fraction but accelerated) for a population that was 75% HPV-. Certainly seems reasonable for the HPV+ ENI dose to be slightly lower. In a similar fashion, the intermediate dose is moved down 10 Gy for indeterminate nodes. Again consistent with risk levels.
Additional References to be aware of:
Phase 2 De-escalation trial: UTSW - covered below in a bit of detail.
My Summary:
Today we covered at least 700 patients - I believe closer to 1000 patients. 500 are from randomized controlled trials where we have successfully shown no clear risk to de-escalation of the neck (even the “worst” example of the three is very limited differences). The “Additional References” add over one hundred more treated in Phase II de-escalation trials.
Almost certainly, these dose reductions come with less toxicity. I liken it to the move that we have made in allowing more treatments to be unilateral. Even if there is some very limited risk, it is countered by a clear justification by reducing toxicity.
Below are my reasonable treatment standards off trial based on this data:
2025 Reasonable De-Escalation Standards (EQD2):
HPV-: ENI of 40 Gy, 70 Gy to gross disease.
HPV+: ENI of 30 Gy, 70 Gy to gross disease.
Intermediate dose of 60 Gy in HPV- and 50 Gy in HPV+ is warranted in a minority of cases for suspicious indeterminant nodes (Defined above).
Note: Post-operative doses are muddied by this shift. Too much for one article.
If you made it this far, please hit the like button. Even if you disagree, it is the “easiest” way to support the work - can always also clarify with comment. These do take about 20-40 hours to put together and engagements drive all social algorithms beyond payment models which I’ve chosen to avoid.
A Note on Planning: “A Return to Old School” Stepwise
I think most looking in this direction have reverted to multiple sequential plans due to the dose differences. For a number of reasons, I think this is easier than very large cGy / fx differences across the plan.
So for example: an IMRT to 40 Gy (HPV-) or 30 Gy (HPV+) to comprehensive volumes at 2 Gy. Then a reduction to 60 Gy (if applicable - see criteria above) and then a final plan to 70 Gy to encompass disease.
I’ve planned multiple cases out using this step-wise approach and it is amazingly easy - especially when not treating an intermediate volume. Uniform dose initial plan and then a boost plan. Both uniform and clean - really volumes and rings and go. And the lower low dose makes meeting cord, constrictors etc. magnitudes easier.
A Note on Exclusion of Low Risk Nodal Basins:
“De-escalation” can also include out-right exclusion. The historical example would be in unilateral cases where the contralateral neck is avoided or avoiding dose to nodal basins not at risk in a particular case. I want to spend just a moment on more aggressive volume de-escalation in order to be more comprehensive. Specifically, this trial stands out to me:
It is the most interesting approach I found. They attempted to reduce dose to 40 Gy for elective nodal treatment AND excluded some traditional volumes. Here is their approach for exclusion:
Patients with oropharynx cancer had their bilateral level II lymph nodes included in the CTV, except T1-2, well-lateralized tonsil primaries; a total of 5 tonsillar primary patients received ipsilateral radiation therapy. If a station had an involved node, that level and the adjacent cervical lymph node chain level were included in the CTV. For example, if level II included a positive node, levels II and III were included in the CTV. If a level contained a suspicious node, only that level was treated. The cranial border of level II for an involved neck was the jugular foramen.
In patients with laryngeal cancer without subglottic extension, bilateral levels II and III were treated in all patients. Level IV was treated if the ipsilateral level III contained an involved node or if there was a suspicious node in that level. Bilateral level IV and VI nodes were treated if subglottic extension was present or an emergency tracheostomy was required.
For both sites, level IB was only treated if it contained an involved or suspicious node or if the primary tumor invaded the oral cavity. Level V was only treated if it contained an involved or suspicious node or if 2 other ipsilateral levels included a positive node. All lymph node stations were treated to 40 Gy in 20 fractions.
Areas beyond these rules were excluded from all dose. Further they have an SIB intermediate dose level to 64 Gy in 35 fractions - so many interesting thoughts - worth a full read.
Ultimately, there were 7 nodal failures, 5 of which were in the 70 Gy volume and two more were basically synchronous failures leaving zero solitary elective nodal failures. Note: I had this trial show up in some AI reviews as a negative / “bad” outcome based on these 7 nodal failures. Both the authors and myself don’t think that is correct, but as always, do your own research for your patients.
The level of data with this approach, to me, steps lower than moving the dose, but in talking with several clinicians, they are as comfortable or more with this concept - after all we have a long history of eliminating some regions based on the primary. Especially, in the lower necks, I think this gives some additional flexibility and / or support for lessening our prescriptions.
If zero dose carries, at least, limited risk, then 30 Gy in HPV+ might seem more comfortable.
Additional reference of Exclusion:
Today’s New Standard
I see us on the verge of replacing a more than half a century standard dose level.
Well, kind of…
See…If you go back up to the original 1972 paper, the recommendation was 4500-5000 cGy - without chemotherapy. One can argue quite easily that 4500 cGy vs. 4000 cGy (often with chemotherapy) isn’t in fact a change. The error may been history simply landing on the top side of the dose recommendations. And when we went to SIB IMRT, we again erred high pushing into the mid 50s. So it made for a catchy title today but maybe the answer is more accurately.
Dr. Fletcher was correct: 4500 at 180 per day without chemotherapy is adequate to eradicate subclinical disease.
Few will ever write something that holds up that well for the majority of a century in medicine.
Today, I think physicians can shift lower to the standards above now based on the totality of our literature - at a minimum begin planning both ways and picking cases to move lower. Hopefully this article can help support that shift. Read the literature and decide. My guess is that this “change” has snuck up on many in practice.
There is nothing wrong in my mind with doing more levels or a slight adjustments in the dose levels in your plans while the data continues to mature and broaden - but in doing so, make certain you are cooler in the ENI regions and that your overall approach lowers integral dose while maintaining full dose in disease. (Easiest approximation of integral dose is mean dose to external contour IMO).
Personal Caveat: Remember, in these three trials a minority (no more than 1/3rd, probably closer to 15%) have an intermediate dose level. I’ve seen many a good plan “de-optimized” (my words) by physicians expanding their intermediate dose volumes being “overly cautious”. I’d attempt to mirror the trials. Lowering ENI while having large expansions of the 60 Gy zone would very likely undercut nearly all benefit. Goal is lower integral dose.
I’ve said for years, the proton therapy data will push the field forward faster than if protons didn’t exist. IMRT simply will push harder / faster along a de-escalation path than if protons didn’t exist in order to compete. I think here we are beginning to see this come to fruition and I believe it is a huge win for our field. I purposefully kept protons out of this discussion - it got too long and became too windy of a road. So hopefully I made one clear point some can integrate into their practice.
Exciting times. As always, keep pushing for better.
Thanks for reading along. Hopefully this inspires some thought - even if you disagree. Please comment if you see things differently - it is about learning and pushing forward. Happy to host dissents or agreements. It’s been a while and certainly no way I will repeat the routine schedule of work I did a few years back, but when a topic strikes that I think my experience can benefit even a few, I’ll likely put some thoughts on paper.
Pleasure to follow your chain of thought, as always. Might we think something similar for post-op prostate bed during early salvage RT..?
Post-op has such limited data by comparison. I think I'll work on a "thoughts" / opinion post on that front, but I wanted to keep this more data driven. That one will clearly have more opinion just due to the lack of big randomized data.