One simple question that has been bothering me. Why should we aim for homogeneity of dose within the tumor volume when inherently the tumor is inhomogeneous? Is it required for comparing the plans or an esoteric ideation? Can we do without it?
Reality is merely a persistent illusion
very valid point since IMRT by definition relies on the concept of "planned" dose inhomogeneity.For starters , SIB itself delivers different dose per fraction to different volumes within the total treated volume . Then again , taking the concept even further, dose painting by numbers on theragnostic imaging attempts to deliver " planned " different doses per fraction to different sub-volumes within the individual dose level volumes based on the local tumor characteristics as visualized on biological ( functional) imaging—admittedly still some way from wide clinical implementation , but with excellent image guidance and with volumetric arc therapy significantly cutting short treatment times and therefore intrafraction motion , it seems increasingly possible with every passing day . Whether it will improve clinical results or whether it will increase second malignancies ? —long way to go from generating conclusive evidence , but the debate is on , and hotly so , with greats like E.J. Hall ( against the motion ) and Soren Bentzen ( for the motion ) producing plenty of literature on the subject.
what do our members feel?
1. Eschmann SM, Paulsen F, Bedeshem C, Machulla HJ, Hehr T, Bamberg M, et al. Hypoxia-imaging with 18F-misonidazole and PET: changes of kinetics during radiotherapy of head-and-neck cancer. Radiotherapy and Oncology. 2007;83(3):406–410.
This is a very interesting article that discusses something that is well known with head and neck cancers. The areas that are hypoxic today may not be there tomorrow. So when you are targeting them with theragnostic imaging are you going to PET them before each fraction? So making IMRT doses inhomogenous based on these markers is premature to say at the very least. Infact we will have to go a lot further in terms of computation speed, auto segmentation algorithms and deformable image registration before we can actually fit a IMRT dose distribution adaptively to changes over each week to say nothing of the changes that happen from day to day. You may argue with CBCT you can ..but really how many hoops you have to jump through now to get there?
Another issue is imaging resolution of course - PET has a pathetic resolution as anyone doing contouring on PET images will know - auto-contoured zones based on SUV look like cubes instead of spheres without extensive smoothing by software. So even if you detect an area which is supposedly having a more radioresistant area how sure about the boundaries you are? When you add in the PTV to that area can you imagine the zone of uncertainity? When we are talking about delivering high doses to a well defined zone in tumor we must have imaging that tells us reliably where that zone is without requiring assumptions and presumptions.
There is another very important issue with SIB that was raised by the physicists early on - what about normal tissue embedded in the target? You can blast the tumor with 3 Gy each fraction easily but the cost in terms of normal tissue toxicity may be prohibitive. Another important thing with SIB is the OAR dose fraction. Typically it will be less than 1.3 - 1.5 but it is quite easy to overlook a higher dose per fraction. E.g. Nasopharynx CA temporal lobe 2% gets 62 Gy —- great ? what if that 62 is there in 30 # ? What is the cost to that?
An inhomogenous IMRT plan is unlikely to be more carcinogenic than a homogenous IMRT plan and my gut feeling is with IMRT we may see an increase in tumor control and survival leading to a greater number of 2nd cancers thanks to the bad genes that will get blamed on RT.
Another important issue is the resolution of the dose distribution that we get with IMRT. Even with helical tomo - conceivably the best intensity modulation we can have it is not possible to reduce doses by more than a couple of gray in a span of 2 - 3 mm. Considering most head and neck cancers we have are about 5 -8 cm in size what is the maximum inhomogenity we can get within the tumor - my guess is of the order of 10 Gy or so. So even if we could potentially delineate areas inside the tumor which are supposedly bearing "radioresitent clonogens" - can we hit them with a high enough dose while respecting the normal tissue tolerance? I am talking of doses in order of 100 Gy here (yes … what brachy can achieve today)
Perhaps the biggest lesson regarding the lack of usefulness of homogenity provided you have got a minimum target dose given is brachytherapy - compare the dose distributions there and those we obtain with conventional RT - and even for HN for a 2 cm leison in tongue brachy will beat EBRT any day for LC. So its not that homogeneity is the goal - its just that we dont have a way of producing a sufficiently useful inhomogeneity in a relatively well known zone where we know that the inhomogenity should be placed
So does this mean we will continue to have homogenity in dose as a goal for ever? No my hunch is few developments will change the way we use RT in future:
In fact, we will have to go a lot further in terms of computation speed, auto segmentation algorithms and deformable image registration before we can actually fit a IMRT dose distribution adaptively to changes over each week to say nothing of the changes that happen from day to day.
We are running just such a protocol here in WLG! Weekly rescanning, revoluming and replanning of H&N patients. We are all manual at present, and have a turn around time of <24 hours on IMRT.
Anyone got algorithms for autosegmentation or deformable image registration that they want tested? I have the data on 9-10 patients that can act as an objective basis for performance testing.
An inhomogenous IMRT plan is unlikely to be more carcinogenic than a homogenous IMRT plan and my gut feeling is with IMRT we may see an increase in tumor control and survival leading to a greater number of 2nd cancers thanks to the bad genes that will get blamed on RT.
If we lift the cure rates for cancers by 5-10%, the number of induced malignancies will be well and truly traded.
Should we being asking patients whether they want more cures and tissue sparing now, or a lower second malignancy rate in 10-20 years time?
Even with helical tomo - conceivably the best intensity modulation we can have it is not possible to reduce doses by more than a couple of gray in a span of 2 - 3 mm. Considering most head and neck cancers we have are about 5 -8 cm in size what is the maximum inhomogenity we can get within the tumor - my guess is of the order of 10 Gy or so. So even if we could potentially delineate areas inside the tumor which are supposedly bearing "radioresitent clonogens" - can we hit them with a high enough dose while respecting the normal tissue tolerance? I am talking of doses in order of 100 Gy here (yes … what brachy can achieve today)
I think we can actually, especially with 5-8cm tumours. We haven't really tried and we don't know much about what happens to tumour size over the weeks to make the delivery safe. I suspect that we could plan a double-dose to the high risk PTV a week at a time. Problem will be smaller tumours where you can't get an inverse movement margin to dump some dose.
So does this mean we will continue to have homogenity in dose as a goal for ever?
Even now, I am considering how to get more dose into the high risk area. the EUD predicts that any positive dose heterogeneity (i.e., never anything less that 95%, but peaks of unrestricted? higher doses) will improve the control rates. Problem I have is how to produce this in a plan on a routine basis.
Let me add more to the confusion!!
1) Most of the biological models assume a uniform radiosensitivity (alpha/beta ratios) in order to "simplify" it.
2) Radiosurgical principles induce deliberate inhomogeneity.
3) Tumor by itself is not homogeneous. We know it from Hypoxia experiments (although I don't know about the TUNEL experiments for molecular markers).
4) EUD is a concept defined to explain the conformal radiation (including IMRT) because that has the been the prevailing practise in the past.
5) IMRT by itself creates fluences (not uniform doses) so while it may be prudent to compare it on basis of conformality but homogeneity is a suspect!!
Reality is merely a persistent illusion
Level "10" evidence , no doubt , but does add fuel to the fire —interesting reads (eye-openers?)
http://www.cksociety.org/document/docdownload.aspx?docid=651
this innocent appearing point made by Abhishek is escalating to a full fledged discussion —do we want a fresh thread on LATTICE Radiotherapy?
