I would suggest a slight modification to this:
1. It seems when the GTV directly abuts a critical structure protons alone do not add value and may be deleterious (prostate/pancreas/posterior fossa/central lung/some H&N).
-> Note - PTV under coverage is very reasonable; as
@ramsesthenice notes above; w/ Crane/Ablative style panc dosing and OAR PRV prioritization.
2. When the PTV abuts a low Z medium (i.e. chest wall ranging into low-z lung), your integral dose could be worse than a photon plan
3. As such, protons add value only in very specific settings for sparing organs in parallel. They do not add value for sparing organs in series and may actually harm these organs due to RBE uncertainty and lateral penumbra. This contradicts much of the original proton dogma. Q.E.D.
@IonsAreOurFuture - I understand what you are saying; but honestly I am not sure you full understand the what is going on w/ the SOBP and the summation/integral of the downrange dose. If you look at a lot of the physics work from the 2018 (I think?) ASTRO special session on brain stem necrosis; LET and RBE uncertainty, and ways to mitigate it (many of which you address). Your point about a "pure" Bragg peak just does not make sense in regards to the clinical sequalea.
Correct me if I am wrong; but what I imagine happening is similar to how an electron beam creates downrange atomic interactions and builds a fluence/flux of photons that builds up to a asymptotic level over depth.
The crux of the Proton "problem" is that instead of a fluence/flux of photons, it includes a flux of subatomic particles with directionality and energy; and whose RBE is difficult to calculate and/or model; and whose ultimate penetration does not map onto the proton range.
A Bragg peak in the middle of a SOBP mathmatical distribution (i.e. one in the GTV) as you describe; one in which is weighted less than the distal-most Bragg peak; would still possess all the same qualities of the distal Bragg peak (generating the equivilent of subatmoic shotgun blasts of potentially high RBE subatomic buckshot); but with an x% multiplier given the dose at that point is a sum of the different energies. It would still participate in the Asymtotic build-up of flux. We just don't notice that as much since it gets adjusted for in the Proton RBE multiplier and/or it is in the GTV; but it most likely participates in the lateral beam uncertainty/penumbra.
As such; I would propose the "problem" of protons is inseperable from the technology. Unless better mapping/modelling is generated.
Robustness can help but it is an inherent source of error/uncertainty.
Thanks Upgrayedd. You make some really good points, and I think it's probably fair to say that a lot of the original proton dogma was/is wrong and I can still get into warm conversations with my physicists who grew up in the passive scatter era, from which some rules no longer apply.
One concept that I still hear is "use as few beams as possible, because it's faster and the integral dose is lower." Integral dose is usually not that different from an arc plan vs 3D conformal or even APPA; only the high dose vs low dose volumes are different - the main exception being a target that is really close to the skin on one side of the body, where you might use electrons instead of a multi-beam Xray plan or proton plan.
The speed of delivery argument is not so true in a modern pencil beam facility where rotating the gantry is pretty fast and the therapists don't have to go back in the room - in the passive scatter days (still the norm for most proton rooms built circa 2012 and earlier - e.g. Loma Linda, MDACC, MGH, the Procure facilities) - every single beam needs its own: 1. Custom-machined brass aperture block. 2. Range compensator/distal edge shaper, both need to be inserted into the head of the machine like an electron cutout and a custom-printed bolus from dotDecimal. Imagine going in and out of the room 8 times to treat a 7 field IMRT plan. Therapists would get >15 miles worth of steps in per day. VMAT is simply impossible.
More than 2 fields of passive scatter a day = really slow treatment - so some centers, and I was pretty sad to learn this, because it seems kinda stupid, didn't treat every field every day. I'd heard of this being done on Linacs back in the early days of 3D conformal, treating a 4-field pelvis AP-PA one day and opposed laterals the next, so you only need to use 2 sets of custom blocks per day in the pre-MLC era (yes, there was such a time).
The thinking was, "It'll average out into a 4-field box," but we all know that 2.2 Gray per fraction through the whole rectal width every other day is not the same biologically as 1.1 Gray daily.
You are also correct to cite the 2018 workshop on brainstem necrosis as the time that RBE/LET really hit the field hard and people had to look in the mirror and change their ways. One of the things I gleaned from it was that one should simply treat every field every day, and come of the brainstem as much as possible for the boost, and not give more prescribed dose than needed. Just common sensical stuff.
NCI Workshop on Proton Therapy for Children: Considerations Regarding Brainstem Injury - Haas-Kogan 2019 IJROBP - it's a classic paper in proton therapy, free full text online
MGH would treat 2 fields: PA & LPO beam on one day and PA & RPO on the next. One of my physicists once suggested this as being somehow okay and I was pretty upset.
This DVH and plan below from Haas-Kogan, et al, shows a pretty dramatic increase in brainstem dose, going from the assumed fixed RBE = 1.1 to a brainstem RBE = 1.2. Not pretty; you go from 55 Gy max nominal dose to 60 Gy RBE-weighted dose in the brainstem. One recommendation was to not aim more than one beam terminating in the brainstem (usually the PA beam).
Today I can create this type of comparison DVH on Raystation in clinic, and I have already started using it for better beam angle selection. The closer I get to an arc plan (more beams) the less this type of scenario happens, because I don't have all the distal Bragg peaks (the hot pristine ones) ending up in the organ at risk, but rather fanned out all around. LET-based optimization is also in this version of Raystation but I haven't cracked the hood open on that one, yet...
To really learn more about proton LET/RBE and variable RBE, please read AAPM TG-256. It's really quite well done and has about 6 strategies for coping with the problem:
https://aapm.onlinelibrary.wiley.com/doi/10.1002/mp.13390 - behind the AAPM paywall but your librarian or physicist can get it for you
To answer your question about, Does the TPS account for dose from secondary particles like scattered electrons, neutrons, etc. Short answer is yes, it's all Monte Carlo based modeling like you would do for an atomic bomb. Raystation is awesome, they even account for induced deuterium and alpha particle formation from rare nuclear interactions; recently learned from a meeting talk I attended. Here's their website. They also talk about 4D motion mgmt for particles, robust CTV-based optimization, LET/RBE, and step-and-shoot proton arcs. Super smart people:
Treatment Planning System - RayStation | RaySearch Laboratories One of the few software companies that releases a major upgrade every 6 months.