Don’t mean to beat this horse to death but since “alpha/beta” (and LQ and a BED-Gy-x) will in general tend to predict/verify your clinical hypotheses, why not have confidence in it? (And specifically you’re talking about “late responding tissues,” something that makes sense given LQ but doesn’t afaik without it.) All the in vitro cell survival studies looked at fractional doses well up to and past 10 Gy. There was never any hint, that I’ve seen, of the model itself breaking down as the fraction sizes got higher well past 3 Gy.
pubmed.ncbi.nlm.nih.gov
Similarly, there is data for T1 non-melanoma skin cancer where there is good local control with single fr ~20 Gy x 1, but past that point risk of skin necrosis increases. Digging down into the references and results prior radiobiological experiments decades past, iirc there was not any benefit to fractionation for local seen until lesions were >2cm.A. Yes, for large* lesions, one follows RTOG and sure enough, lower BED to reduce risk = lower biological effectiveness.
B. For smaller lesions, there is no doubt single fraction is satisfactory. Background snip: For example, the volume of brain tissue receiving 12 Gy or more in radiosurgery appears to be correlated with the risk of radionecrosis, particularly when this volume exceeds 10–15 mL. Note, however, that this limitation appears overly restrictive, as it appears that virtually every single-fraction radiosurgery plan would exceed this limit when large lesions were treated to accepted doses.2
View attachment 363045
Thus, the only remaining question is: what is the cutoff for symptomatic necrosis (volume/dimension) where fractionation becomes a winner? Do I hear 3 cm? 4 cm? 5 cm?
ps. some patients are struggling to make it thru one fraction. for those folks, one it is based on life expectancy.
These things are not binary. The cutoff for any side effect is highly variable and patient dependent. So to me the following questions should be asked in order.
I know we've talked about this before. It's math to me. LQ is a 2nd order polynomial function fit to data that is what it is (the data would be even better approximated by a 3rd or 4th order polynomial or many types of infinite series). The fit is always best within a certain domain (fraction size). In addition, at some point in many bio systems weird things happen with curves that are unexpected and not even monotonic.
This is a fascinating and terribly interesting discussion to me. Prostate people were trying "virtual prostatectomies" with single fraction EBRT approaches. I think single fraction "can" (serious air quotes) always work, at least oncologically. (Spillover of the dose into normal tissues is another discussion obv.) But when you get up into high fraction sizes and tumor alpha/betas that are near or below normal tissue alpha/beta, you get EXTREME fractional (fraction size) sensitivity. And by that the best analogy I can think of is like with hypercars... and that above 200mph it takes an extra 8hp for every extra 1mph you try and go faster. There may not seem that much difference, mathematically, between 20 Gy and 22 Gy e.g. But in certain "LQ situations" that difference can be very CLINICALLY noticeable both in terms of LC and toxicity.
That 50/5 seems to work as well as things that are higher BED implies many interesting things, no matter the disease site. First it may imply that we are overestimating clonogen number within the apparent radiographic GTV (there is evidence eg that CT very over estimates the tumor edge extent in NSCLC, that a billion cancer cells per cc is wrong, etc). Second it may imply, for NSCLC eg, that we are dealing with slower growing and thus more late responding (and lower alpha/beta… a lot lower than 10) NSCLC cells than we might otherwise suspect; there is data for this too. Third, it may imply a mysterious radiobiology. I favor the latter implication the least!
Showing your lack of class with this reply. No need to get ugly.
I’m reporting you, this kind of post does not belong here!
pubmed.ncbi.nlm.nih.gov
pubmed.ncbi.nlm.nih.gov
Second sentence of one of the reference abstracts: “Much of the data used to generate the model are obtained in vitro at doses well below those used in radiosurgery.”*Lots of papers challenging accuracy of LQ model for higher doses
The linear-quadratic model is inappropriate to model high dose per fraction effects in radiosurgery - PubMed
The linear-quadratic (LQ) model is widely used to model the effect of total dose and dose per fraction in conventionally fractionated radiotherapy. Much of the data used to generate the model are obtained in vitro at doses well below those used in radiosurgery. Clinically, the LQ model often...Universal survival curve and single fraction equivalent dose: useful tools in understanding potency of ablative radiotherapy - PubMed
The USC can be used to compare the dose fractionation schemes of both CFRT and SBRT. The USC provides an empirically and a clinically well-justified rationale for SBRT while preserving the strengths of the LQ model for CFRT.
LQ model for clinical stuff is really based on clinical outcomes. You think an a/b for late myelopathy is roughly 3, it's based on clinical data (maybe animal data), not cell culture experiments.
Tell me where you have not seen an exponential increase in tumor kill or clinical side effects with linear increase in dose, varying fractional sensitivity depending on the tissue/tumor treated, etc. Leave the (voluminous, profound) in vitro studies out of mind. None of the things we see with our own eyeballs make sense without LQ. If something else makes better sense… what? If nothing else makes sense, let’s turn off the linacs because we are just experimenting on people.
No reason to turn off the linacs, just reason not to come up with custom, single fraction, high dose SBRT prescriptions based on the LQ model. (again, not really even a model, more of a fit).
Sick and elderly, 8Gy x 1, always safe. 3D plan.How's that song go again?
Fraction for me. Fraction for you. I love fractionssssssssssss... (but my sick elderly patients don't)
What’s weird is we are saying exact same thing, you’re using LQ to say it, and saying… don’t trust LQ. Heh.
Fair. I trust it within the 1.8-3 Gy per fraction range quite a bit. I trust it for large a/b effects (mostly linear) more than for small a/b effects (more quadratic). I use it all the time to find reasonable dose painting solutions or to moderately hypofractionate.
These graphs look great! Seem to support a linear-quadratic radiotherapy response. (Which becomes quadratic-linear when you graph on a log plot… see first graph… but I digress!)
Let me make sure I still “got it.”
