Fractionated vs. single fraction SRS for brain mets... What's your cut-off?

[KHE88]

I am tending to fractionate tumors I can't get 20/1 into or at least 18/1.

Under 2 cm/3cc or so I'll do 20/1
2-3cm / 3-14 cc I'll do 18/1
Towards the upper end of that range, I start thinking more about fractionated SRS, and > 14-15 cc I go 27/3.

Not doing any single fraction < 18. Is anybody else?

I don't like that grey area very much, and can't find a lot of good data on this.

Curious how everyone else is approaching these 2.5 - 3.0 cm borderline tumors.
And also for fractionated, what's your cut-off to go into 5 fraction range, and what dose are you using for 5 fraction? I've never personally gone into 5 fraction, but I see published data on 25/5, and I'm thinking why?

---

Members don't see this ad.

 

[BobbyHeenan]

I'm very similar to you. 2cm or less single fraction. If I'm seeing 12 Gy isodose volumes getting > 15 cc I push for fractionation.

Around 2cm or up I fractionate and typically have been using 27/3 per the Italian study. Without having the paper in front of me, I *think* their necrosis predictor for 3 fraction was volume of normal brain (they defined as brain minus GTV) getting 18 Gy needing to be less than 30 cc. If I can't meet that I've done 30 Gy in 5 or 27.5 Gy in 5. The paper has a nice paragraph in it about necrosis prediction in their series.

There is data that 25 Gy in 5 probably not enough:

Fractionated stereotactic radiation therapy for intact brain metastases

Limited data exist on fractionated stereotactic radiation therapy (FSRT) for brain metastases. We sought to evaluate the safety and efficacy of FSRT and further define its role in brain metastasis management.A total of 72 patients were treated with linear ...

www.ncbi.nlm.nih.gov

I *think* fractionated is the way to go for larger tumors, but make no mistake, it's still a challenge for both tumor control and radionecrosis, as I've seen anecdotally less problems with fractionated than single fraction, but still have seen necrosis and failure at 27/3 or 30/5 for bigger tumors.

Aside: Immunotherapy reactions have been bizarre sometimes too... a quiescent operative bed/tumor for upwards of a year, then immunotherapy, then crazy edema a few months after immunotherapy starts. This is not my case report but I've seen this too:

Unmasking of intracranial metastatic melanoma during ipilimumab/nivolumab therapy: case report and literature review - PubMed

These cases raise the question of whether the initial immune activation and modulation of the blood brain barrier by Ipilimumab/Nivolumab somehow "unmasks" previously clinically silent metastatic disease, rather than representing new or progressive metastatic disease. An overview of currently...

www.ncbi.nlm.nih.gov

 

[Palex80]

1 x 20 (@80% isodose) as long as V12 < 10cc, otherwise 3 x 9 (@80% isodose).

6 x 6 Gy (@95% isodose) for really big lesions.

 

[medgator]

2-3 cm I'm doing 27/3, larger than that 30/5

 

[RickyScott]

V12 for 20 gy should be around 3x volume

 

[evilbooyaa]

I'm very similar to you. 2cm or less single fraction. If I'm seeing 12 Gy isodose volumes getting > 15 cc I push for fractionation.

Around 2cm or up I fractionate and typically have been using 27/3 per the Italian study. Without having the paper in front of me, I *think* their necrosis predictor for 3 fraction was volume of normal brain (they defined as brain minus GTV) getting 18 Gy needing to be less than 30 cc. If I can't meet that I've done 30 Gy in 5 or 27.5 Gy in 5. The paper has a nice paragraph in it about necrosis prediction in their series.

There is data that 25 Gy in 5 probably not enough:

Fractionated stereotactic radiation therapy for intact brain metastases

Limited data exist on fractionated stereotactic radiation therapy (FSRT) for brain metastases. We sought to evaluate the safety and efficacy of FSRT and further define its role in brain metastasis management.A total of 72 patients were treated with linear ...

www.ncbi.nlm.nih.gov

I *think* fractionated is the way to go for larger tumors, but make no mistake, it's still a challenge for both tumor control and radionecrosis, as I've seen anecdotally less problems with fractionated than single fraction, but still have seen necrosis and failure at 27/3 or 30/5 for bigger tumors.

Aside: Immunotherapy reactions have been bizarre sometimes too... a quiescent operative bed/tumor for upwards of a year, then immunotherapy, then crazy edema a few months after immunotherapy starts. This is not my case report but I've seen this too:

Unmasking of intracranial metastatic melanoma during ipilimumab/nivolumab therapy: case report and literature review - PubMed

These cases raise the question of whether the initial immune activation and modulation of the blood brain barrier by Ipilimumab/Nivolumab somehow "unmasks" previously clinically silent metastatic disease, rather than representing new or progressive metastatic disease. An overview of currently...

www.ncbi.nlm.nih.gov

The only thing about that UAB analysis is that they mostly planned zero margin, so 5Gy x 5 with zero margin is worse than 6Gy x 5 zero margin. How that relates to 5 x 5 with say a 1mm margin is TBD.

Anything 2cm or larger is definitely getting 3Fx. I am much more stringent on V12 than 10cc, especially on a per lesion basis. I think if you're accepting V12 > 10cc you should just fractionate. I usually accept no more than V12 < 5cc on an individual lesional basis.

Lesions 1.5-2cm are probably getting 3 Fx as well but I'll evaluate. If it's close to skull so that I can do angle dose away from brain tissue then maybe I'll still SF.

 

[radiation]

1 x 20 (@80% isodose) as long as V12 < 10cc, otherwise 3 x 9 (@80% isodose).

6 x 6 Gy (@95% isodose) for really big lesions.

Question about the isodose lines - do you use FFF or flattened beams? I've heard people pick 80% because that's steepest portion of the beam profile for a single flattened beam. But now that people are doing VMAT, dynamic arcs, etc (often with FFF), the beam profiles often become quite complex for irregular lesions and I am not sure always doing 80% gives you the best dose falloff.

Can you explain the rationale for prescribing to the 95%? Seems like that would really hurt your dose fall off as well

 

[scarbrtj]

Can you explain the rationale for prescribing to the 95? Seems like that would really hurt your dose fall off as well

Same question here. So that we're all not speaking different lingoes like a bunch of nimrods, can we agree that the 100% is the max dose in all SRS plans?

 

[KHE88]

Also, since we brought up isodose lines, does anybody else in the community have trouble with dosimetry/physics looking at you with two heads when you want to prescribe to the isodose line?

When I started doing SRS plans here, the first plans they showed me were something like 160% hot so that the volume was covered by the 60% isodose line or so. I tried to explain the rationale of steepest dose falloff around 80%, which was met with, "but we're covering 99% of the volume with the prescription dose as is, why do I need to reoptimize?"

An oddly enough, I am running into the opposite problem with breast plans, as they don't want me to prescribe to a volume but instead to an isodose line. Was this the way it was done historically? I finally just stopped arguing and change the prescription in Eclipse to the isodose line instead of the volume once I've verified coverage is appropriate.

Question about the isodose lines - do you use FFF or flattened beams? I've heard people pick 80% because that's steepest portion of the beam profile for a single flattened beam. But now that people are doing VMAT, dynamic arcs, etc (often with FFF), the beam profiles often become quite complex for irregular lesions and I am not sure always doing 80% gives you the best dose falloff.

Can you explain the rationale for prescribing to the 95%? Seems like that would really hurt your dose fall off as well

I have seen publications that indicate the 80-90% isodose line is preferable for single lesions, but when treating many isos going down to 75% or so is more appropriate. Not exactly sure why that works out better but apparently it does. I guess just because of integral dose shifting the inflection point on the penumbra.

 

[Palex80]

Question about the isodose lines - do you use FFF or flattened beams? I've heard people pick 80% because that's steepest portion of the beam profile for a single flattened beam. But now that people are doing VMAT, dynamic arcs, etc (often with FFF), the beam profiles often become quite complex for irregular lesions and I am not sure always doing 80% gives you the best dose falloff.

Can you explain the rationale for prescribing to the 95%? Seems like that would really hurt your dose fall off as well

We do not use FFF.

Well, one argument is that bigger lesions may have more microscopic extension outside the visible GTV and thus a not very steep dose fall off may not be that bad. Other than that, I personally think that in big lesions and when using not high single doses there is no clear rationale to prescribe to an isodose <95%, to me a single dose of 6 Gy is not anymore "SRS-like". I don't prescribe to the 80% isodose either when I treat bone mets with 5 x 5 Gy, although I could do that and probably get more dose into the bone without risking any problems. But that's just my feeling.
Perhaps I am also a bit concerned about large volumes of tissue (irrelevant if it's tumor or not) receiving doses way >100% (although one can optimize the plan so that does not happen). It's ok if it's something like 0.5 cc in a small met and you give a single dose of 25 Gy, but if you are treating "more" ccs, maybe one should't "push" the dose too much.

 

[Palex80]

Same question here. So that we're all not speaking different lingoes like a bunch of nimrods, can we agree that the 100% is the max dose in all SRS plans?

In brain, yes. In NSCLC or liver: no.

 

[radiation]

We do not use FFF.

Well, one argument is that bigger lesions may have more microscopic extension outside the visible GTV and thus a not very steep dose fall off may not be that bad. Other than that, I personally think that in big lesions and when using not high single doses there is no clear rationale to prescribe to an isodose <95%, to me a single dose of 6 Gy is not anymore "SRS-like". I don't prescribe to the 80% isodose either when I treat bone mets with 5 x 5 Gy, although I could do that and probably get more dose into the bone without risking any problems. But that's just my feeling.
Perhaps I am also a bit concerned about large volumes of tissue (irrelevant if it's tumor or not) receiving doses way >100% (although one can optimize the plan so that does not happen). It's ok if it's something like 0.5 cc in a small met and you give a single dose of 25 Gy, but if you are treating "more" ccs, maybe one should't "push" the dose too much.

I think microscopic extension coverage is reasonable, but wouldn't you want to know what that coverage is (ie make a CTV and assess coverage) rather than guess on asymmetric dose falloff? From a purely dosimetric standpoint, the rationale to prescribe to a lower isodose line would be that the dose gradient you will get with lower isodose prescription will be numerically better and that less normal brain will get intermediate doses of radiation. Even if these intermediate doses to normal brain do not cause necrosis - low doses to things like hippocampus et al can make a difference.

The tradeoff of course will be a more heterogeneous dose within the GTV. However, it is almost always better to trade higher dose in the GTV than more low/intermediate dose in the normal brain. This is volumetrically amplified with larger lesions. It is hard to argue that increased GTV heterogeneity will lead to more toxicity if your immobilization is good and you are not worried that the high dose will go into normal tissue. I am not aware of any data that higher GTV doses causes toxicity but if you know of any that would be interesting to learn about.

The issue with prescribing to low isodose lines outside the brain for things that require a PTV, is the problem of high dose going into normal tissue, as the PTV will extend into OARs with extracranial targets and these normal tissues could theoretically get overdosed on a bad setup day. But for intracranial targets this is almost never a problem.

 

[radiation]

Also, since we brought up isodose lines, does anybody else in the community have trouble with dosimetry/physics looking at you with two heads when you want to prescribe to the isodose line?

When I started doing SRS plans here, the first plans they showed me were something like 160% hot so that the volume was covered by the 60% isodose line or so. I tried to explain the rationale of steepest dose falloff around 80%, which was met with, "but we're covering 99% of the volume with the prescription dose as is, why do I need to reoptimize?"

An oddly enough, I am running into the opposite problem with breast plans, as they don't want me to prescribe to a volume but instead to an isodose line. Was this the way it was done historically? I finally just stopped arguing and change the prescription in Eclipse to the isodose line instead of the volume once I've verified coverage is appropriate.



I have seen publications that indicate the 80-90% isodose line is preferable for single lesions, but when treating many isos going down to 75% or so is more appropriate. Not exactly sure why that works out better but apparently it does. I guess just because of integral dose shifting the inflection point on the penumbra.

I think that there is sort of a mythology that the 80% isodose has the steepest dose falloff. This is based on a single flattened beam profile. The classic picture is the one is shown here, where the line starts to become the steepest at 80%

This is for a single beam with 10x10 field etc (all the standard conditions when commissioning). But for complex targets this is almost never the case. If you ever check the beam profile from your QA physics documents on your IMRT/VMAT plans you will see the complexity of the beam profile is much higher

And even for single FFF beams its even different still

I would say the ideal IDL is usually between 40-80% and it is highly target specific based on shape and size. If you are skeptical of what I am saying. I would do the exercise of planning your next several SRS cases with different isodose line prescriptions ranging from 40 to 90% and see how it effects your intermediate dose and compactness metrics (ie gradient index if you use gamma knife, or use the RTOG metrics of R50 and D2cm)

 

[FrostyHammer]

The only thing about that UAB analysis is that they mostly planned zero margin, so 5Gy x 5 with zero margin is worse than 6Gy x 5 zero margin. How that relates to 5 x 5 with say a 1mm margin is TBD.

Anything 2cm or larger is definitely getting 3Fx. I am much more stringent on V12 than 10cc, especially on a per lesion basis. I think if you're accepting V12 > 10cc you should just fractionate. I usually accept no more than V12 < 5cc on an individual lesional basis.

Lesions 1.5-2cm are probably getting 3 Fx as well but I'll evaluate. If it's close to skull so that I can do angle dose away from brain tissue then maybe I'll still SF.

Wow, V12 <5 cc? That's ideal but perhaps not practical considering that most BMs we're sent are going to be large enough to produce a V12 higher than that. I would also add caution to use 9 Gy x3 simply because we don't have enough good and well validated DVH parameters to correlate with brain necrosis in that setting. It's like a shot in the dark tbh. Also am not a fan of 25/5. I use that dose for palliation of bone mets like Palex does... clearly the goal in brain SRS is different.

 

[scarbrtj]

I think that there is sort of a mythology that the 80% isodose has the steepest dose falloff. This is based on a single flattened beam profile. The classic picture is the one is shown here, where the line starts to become the steepest at 80%
View attachment 290119

This is for a single beam with 10x10 field etc (all the standard conditions when commissioning). But for complex targets this is almost never the case. If you ever check the beam profile from your QA physics documents on your IMRT/VMAT plans you will see the complexity of the beam profile is much higher

View attachment 290120

And even for single FFF beams its even different still

View attachment 290121

I would say the ideal IDL is usually between 40-80% and it is highly target specific based on shape and size. If you are skeptical of what I am saying. I would do the exercise of planning your next several SRS cases with different isodose line prescriptions ranging from 40 to 90% and see how it effects your intermediate dose and compactness metrics (ie gradient index if you use gamma knife, or use the RTOG metrics of R50 and D2cm)

View attachment 290123

Often times I am just too lazy to type out everything you have here and previously. Glad you were not

 

[evilbooyaa]

Wow, V12 <5 cc? That's ideal but perhaps not practical considering that most BMs we're sent are going to be large enough to produce a V12 higher than that. I would also add caution to use 9 Gy x3 simply because we don't have enough good and well validated DVH parameters to correlate with brain necrosis in that setting. It's like a shot in the dark tbh. Also am not a fan of 25/5. I use that dose for palliation of bone mets like Palex does... clearly the goal in brain SRS is different.

I am comfortable with the Minitti results, personally. I've discussed the rationale for this before. If 9Gy x 3 has a 90%+ control rate for lesions > 2cm I don't somehow think the control rate is going to decrease if the lesion is smaller. Similarly the radionecrosis rate is going to always be better with a fractionated regimen. I'm fine with it being evalauted prospectively, but Minniti was published in 2016 and I'm not waiting for randomized phase III data on something that I think has zero additional risk to the patient.

I think V12 < 10cc is an old threshold when patients weren't living as long. Longer life expectancy = higher risk for eventual necrosis. That's not to say V12 < 5cc is a hard cut-off for say a single lesion but it is a consideration in a multi lesion plan.

For example if there are 5 lesions are you going to accept V12 < 50cc on the plan sum when doing single fraction since that's V12 < 10cc per lesion? I'd be more apt to fractionate if I saw anything like V12 < 25cc. Ideally it'd be much lower than that. For me, anything over 10cc per lesion is for sure getting fractionated in my practice. This is for Brain with the GTV(s) subtracted out, which is how Minitti did it and calculated both a SF constraint (V12 < 13cc) and a MF constraint (V18 < 30cc).

 

[evilbooyaa]

Also, since we brought up isodose lines, does anybody else in the community have trouble with dosimetry/physics looking at you with two heads when you want to prescribe to the isodose line?

When I started doing SRS plans here, the first plans they showed me were something like 160% hot so that the volume was covered by the 60% isodose line or so. I tried to explain the rationale of steepest dose falloff around 80%, which was met with, "but we're covering 99% of the volume with the prescription dose as is, why do I need to reoptimize?"

An oddly enough, I am running into the opposite problem with breast plans, as they don't want me to prescribe to a volume but instead to an isodose line. Was this the way it was done historically? I finally just stopped arguing and change the prescription in Eclipse to the isodose line instead of the volume once I've verified coverage is appropriate.



I have seen publications that indicate the 80-90% isodose line is preferable for single lesions, but when treating many isos going down to 75% or so is more appropriate. Not exactly sure why that works out better but apparently it does. I guess just because of integral dose shifting the inflection point on the penumbra.

There's nothing wrong with the bolded in brain SRS. You're likely using FFF (as are we and most places that use SRS to reduce treatment times) which may have a different optimal isodose 'threshold'. While I don't really constrain the hotspot (and therefore the isodose line) I do look a little strangely if it's more than say 150% (equivalent to prescribing to the 67% isodose line, but I wouldn't ask them to re-optimize if constraints were met otherwise.

The concept of prescribing to an isodose line in breast is strange to me though.... I don't see a rationale for prescribing to anything besides the 100% isodose line.

 

[Palex80]

There's nothing wrong with the bolded in brain SRS. You're likely using FFF (as are we and most places that use SRS to reduce treatment times) which may have a different optimal isodose 'threshold'. While I don't really constrain the hotspot (and therefore the isodose line) I do look a little strangely if it's more than say 150% (equivalent to prescribing to the 67% isodose line, but I wouldn't ask them to re-optimize if constraints were met otherwise.

The concept of prescribing to an isodose line in breast is strange to me though.... I don't see a rationale for prescribing to anything besides the 100% isodose line.

Actually we don't do that.
We accept up to 110% as a hot spot, but the mean dose to the GTV is around 25 Gy, when prescribing 20 Gy to the PTV (which generally is GTV + 2 mm).
I have no data to back it up, but I do think radionecrosis can also happen within the GTV if you overdose. And certainly a voxel or two getting 160% may not hurt, but if you are treating a larger lesion and don't apply any cut-off giving 130% to a cc of GTV may cause trouble. Like I said, I don't have data to back this up, but I do think that a metastasis in the brain is not solely tumor tissue. There is healthy tissue in there too and you may end up causing vascular damage too (more than the one we are aiming to cause).

 

[evilbooyaa]

Actually we don't do that.
We accept up to 110% as a hot spot, but the mean dose to the GTV is around 25 Gy, when prescribing 20 Gy to the PTV (which generally is GTV + 2 mm).
I have no data to back it up, but I do think radionecrosis can also happen within the GTV if you overdose. And certainly a voxel or two getting 160% may not hurt, but if you are treating a larger lesion and don't apply any cut-off giving 130% to a cc of GTV may cause trouble. Like I said, I don't have data to back this up, but I do think that a metastasis in the brain is not solely tumor tissue. There is healthy tissue in there too and you may end up causing vascular damage too (more than the one we are aiming to cause).

110% hotspot is equal to prescribing to the 90.9% IDL (if you assume the max dose is 100%). That can't be right. Your description doesn't really make sense. You accept a 110% hotspot of what? 20Gy? 25Gy? Your prescription dose to PTV is 20Gy. 110% of that is 22Gy. If you're saying 110% of 25Gy (27.5Gy) then given that prescription doses are generally defined as to what is given to PTV you are accepting (27.5/20=1.375 =) 137.5% hotspot. I feel like I'm taking scarbrtj pills right now.

You guys are doing 2mm PTV for intact lesions? Even small community practices are routinely doing 1mm margin here.

GK-SRS routinely gives a 200% hotspot within the GTV (equal to prescribing to the 50% IDL as is common). If you're scared of hotspots over 130% then you're saying GKSRS is a dangerous treatment. I don't think the concept you mention is completely unreasonable, but we don't have data to suggest that. If that was truly the case, Linac-SRS would have significantly lower rates of radionecrosis than GKSRS given the innate differences in hotspot (or, what % isodose line the treatment is prescribed to), but any comparison I've ever seen basically just said "yeah it's the same" despite the stark differences in planning and heterogeneity between GK-SRS and Linac-SRS.

If a lesion calls for a hotspot of 160% based on shape and energy choice and you 'force' it to 130%, you are losing conformity even if it is still an 'acceptable' plan.

 

[Mandelin Rain]

I didn't understand that math either.

 

[Palex80]

110% hotspot is equal to prescribing to the 90.9% IDL (if you assume the max dose is 100%). That can't be right. Your description doesn't really make sense. You accept a 110% hotspot of what? 20Gy? 25Gy? Your prescription dose to PTV is 20Gy. 110% of that is 22Gy. If you're saying 110% of 25Gy (27.5Gy) then given that prescription doses are generally defined as to what is given to PTV you are accepting (27.5/20=1.375 =) 137.5% hotspot. I feel like I'm taking scarbrtj pills right now.

You guys are doing 2mm PTV for intact lesions? Even small community practices are routinely doing 1mm margin here.

GK-SRS routinely gives a 200% hotspot within the GTV (equal to prescribing to the 50% IDL as is common). If you're scared of hotspots over 130% then you're saying GKSRS is a dangerous treatment. I don't think the concept you mention is completely unreasonable, but we don't have data to suggest that. If that was truly the case, Linac-SRS would have significantly lower rates of radionecrosis than GKSRS given the innate differences in hotspot (or, what % isodose line the treatment is prescribed to), but any comparison I've ever seen basically just said "yeah it's the same" despite the stark differences in planning and heterogeneity between GK-SRS and Linac-SRS.

If a lesion calls for a hotspot of 160% based on shape and energy choice and you 'force' it to 130%, you are losing conformity even if it is still an 'acceptable' plan.

Let me clarify that.

20 Gy is the prescription dose to the PTV.
20 Gy is the 80% isodose
PTV=GTV + 2mm

The plan is optimized so that the GTV receives 25 Gy mean dose
25 Gy is the 100% isodose (20 : 0.8 = 25)

Hotspots are up to 110% within the GTV (27.5 Gy)


Do the math for CTV and PTV margins (with the van-Herk-formula) and u will see that 1 mm may not be enough.

Add all contouring uncertainties, fusion of CT/MRI, mechanical issues on the linac (table bending, rotational errors of the gantry), add errors of CBCT matching and any errors in your planning system and your MLC and last but not least include any intrafractional motion under the mask.

 

[evilbooyaa]

110% of 25 Gy, so 27.5 Gy Dmax.
20 Gy is the 80% isodose.

Do the math for CTV and PTV margins and u will see that 1 mm may not be enough. We have been over this before here and there are trials showing that margin matters and 1mm may not be enough.

OK so you're prescribing 20Gy to the 80% isodose line to your PTV.

I'm aware of data saying 0mm < 2mm, not aware of any 1mm data compared to 2mm when using sufficiently high doses, if you can reference data.

The concept of CTV in SRS treatment is not a commonly utilized one. If you're saying your facility does +1mm to CTV and +1mm to PTV then that's your prerogative, thought you were doing +2mm to PTV. While these are both usually the same thing, the rationale behind them is different.

 

[Palex80]

OK so you're prescribing 20Gy to the 80% isodose line to your PTV.

Yes.
Here's a DVH of a plan with 4 mets.

The 4 lines on the far right are GTVs the other 4 are PTVs. The orange one was tricky, quite close to an OAR, hence the not optimal coverage.

I'm aware of data saying 0mm < 2mm, not aware of any 1mm data compared to 2mm when using sufficiently high doses, if you can reference data.

Never been tested.
We just asked our physicists. They did a ton of measurements and came up with the recommendation.

The concept of CTV in SRS treatment is not a commonly utilized one. If you're saying your facility does +1mm to CTV and +1mm to PTV then that's your prerogative, thought you were doing +2mm to PTV. While these are both usually the same thing, the rationale behind them is different.

Indeed. 1 mm CTV + 1 mm PTV. But in essence its a uniform margin of 2mm around GTV in >90% of cases.

 

[evilbooyaa]

OK, we're on the same page. We do a similar concept - slightly lower dose to PTV, higher dose to GTV simultaneously. However, we just do 1mm PTV, no CTV margin.

 

[FrostyHammer]

I am comfortable with the Minitti results, personally. I've discussed the rationale for this before. If 9Gy x 3 has a 90%+ control rate for lesions > 2cm I don't somehow think the control rate is going to decrease if the lesion is smaller. Similarly the radionecrosis rate is going to always be better with a fractionated regimen. I'm fine with it being evalauted prospectively, but Minniti was published in 2016 and I'm not waiting for randomized phase III data on something that I think has zero additional risk to the patient.

I think V12 < 10cc is an old threshold when patients weren't living as long. Longer life expectancy = higher risk for eventual necrosis. That's not to say V12 < 5cc is a hard cut-off for say a single lesion but it is a consideration in a multi lesion plan.

For example if there are 5 lesions are you going to accept V12 < 50cc on the plan sum when doing single fraction since that's V12 < 10cc per lesion? I'd be more apt to fractionate if I saw anything like V12 < 25cc. Ideally it'd be much lower than that. For me, anything over 10cc per lesion is for sure getting fractionated in my practice. This is for Brain with the GTV(s) subtracted out, which is how Minitti did it and calculated both a SF constraint (V12 < 13cc) and a MF constraint (V18 < 30cc).

I hear your thoughts well for sure. What I'm concerned about with the Minniti data is that A) it's not even prospective non-randomized (IMHO phase 3 data may be too much to ask for, but phase 2 data on the other hand...), 2) doing something with the utter lack of even remotely validated brain dose constraints isn't ideal (and as you correctly mentioned, the SF dose constraint in that study was too high especially by your standards!) Which is why the radionecrosis rate was so high, 3) recall that there was an inherent selection bias in that study in that the single fraction group received pretty low doses (at least IMHO). The BED/EQD2 of the Minniti regimen is roughly the same as 18 Gy x1. Meaning that I'd much rather prefer to give a higher dose than that if I can... In which sense, fractionating it gives And if not, I would gladly give three fractions. In other words, my criteria for giving three fractions would be if the constraints for single fraction would not be met with a high enough dose. Your argument for time-dependent radionecrosis is totally correct, but recall that the 90% local control rate in Minniti is also a one year value that will only go down if they are followed longer, i.e. the analogous argument (Longer life expectancy = higher risk for local failure).

 

[scarbrtj]

I feel like I'm taking scarbrtj pills right now.

You're close!

20 Gy is the prescription dose to the PTV.
20 Gy is the 80% isodose
PTV=GTV + 2mm
The plan is optimized so that the GTV receives 25 Gy mean dose
25 Gy is the 100% isodose (20 : 0.8 = 25)
Hotspots are up to 110% within the GTV (27.5 Gy)
Do the math for CTV and PTV margins (with the van-Herk-formula) and u will see that 1 mm may not be enough.
Add all contouring uncertainties, fusion of CT/MRI, mechanical issues on the linac (table bending, rotational errors of the gantry), add errors of CBCT matching and any errors in your planning system and your MLC and last but not least include any intrafractional motion under the mask.

OK so you're prescribing 20Gy to the 80% isodose line to your PTV.

Yes.
Here's a DVH of a plan with 4 mets.
View attachment 290394
1 mm CTV + 1 mm PTV. But in essence its a uniform margin of 2mm around GTV in >90% of cases.

Ohhhhh kay.
Observations...
1) 20Gy minimum lesional dose, 27.5 Gy max dose = ~20Gy at the ~73% (20/27.5).
2) A target-within-a-target is... a target. We can talk PTV and GTV and what not but it's clear what is paramount in all these discussions is that the PTV is the real target. And keeping in line with your van Herk logic-ness, the way to evaluate a dose to a GTV/CTV is not to look at the GTV DVH but look at the PTV DVH as that is the "worst case" vs likely minimal GTV dose (if that makes van Herk sense).
3) We can see from this DVH that the fall-off is not as rapid as, e.g., a plan which prescribes to the ~70% or 65% isodose. Nothing wrong with that.
4) This plan has a max dose of ~26 Gy. The minimal lesion doses are ~20 Gy. This is a plan where 20 Gy is given at the 77%, in essence.
5) IMHO radionecrosis is MOST associated with the max dose in Gy in a plan, not % isodoses. So...

Actually we don't do that.
We accept up to 110% as a hot spot, but the mean dose to the GTV is around 25 Gy, when prescribing 20 Gy to the PTV (which generally is GTV + 2 mm).
I have no data to back it up, but I do think radionecrosis can also happen within the GTV if you overdose. And certainly a voxel or two getting 160% may not hurt, but if you are treating a larger lesion and don't apply any cut-off giving 130% to a cc of GTV may cause trouble. Like I said, I don't have data to back this up, but I do think that a metastasis in the brain is not solely tumor tissue. There is healthy tissue in there too and you may end up causing vascular damage too (more than the one we are aiming to cause).

... in this case you're essentially saying you don't like significant volumes max doses greater than 26 Gy (20 Gy * 130%), and I probably agree with that based on clinical experience. That is to say, if you're making plans where little or no volume >26Gy, it seems to be very safe. May not be the greatest BED sockdolager from the tumor's-eye-view, but it's almost always safe.

I can also somewhat assume from your DVH the lesions are pretty small. I can also see that there seems to be a push toward "GTV dose homogeneity" in your plans. Not sure if I go to that extreme; I'm more of a put beams on, calc dose, and let the inhomogeneities fall where they may kind of guy. If I don't like the inhomogeneities, I tweak. But, a truism: the bigger the expansion margin around the GTVs, the more homogenous doses therein become if there is attempt to simultaneously dose the lesion(s) which contain the GTVs (ie the PTVs). And another truism, related to the previous truism: adding more margin around GTVs sets a physical limit on how rapidly dose can fall off from the surface of the GTVs. Whether this is a big deal or not, I don't know; I think it's probably not a big deal.

 

[Palex80]

Thank you both, for your excellent comments.

And yes, scarbtj, all PTVs were in the range of 1ccm in that plan, so all mets (GTVs) were small (<0.4ccm)

 

[evilbooyaa]

5) IMHO radionecrosis is MOST associated with the max dose in Gy in a plan, not % isodoses. So...

... in this case you're essentially saying you don't like significant volumes max doses greater than 26 Gy (20 Gy * 130%), and I probably agree with that based on clinical experience. That is to say, if you're making plans where little or no volume >26Gy, it seems to be very safe. May not be the greatest BED sockdolager from the tumor's-eye-view, but it's almost always safe.

Do you have data for this? If this is the case shouldn't GK-SRS plans have much higher rates of radionecrosis (since they prescribe to the 50% IDL, meaning 200% of the prescription dose within the GTV) than linac-SRS plans (prescribe to 80% IDL normally, meaning 125% of prescription dose)?

 

[evilbooyaa]

I hear your thoughts well for sure. What I'm concerned about with the Minniti data is that A) it's not even prospective non-randomized (IMHO phase 3 data may be too much to ask for, but phase 2 data on the other hand...), 2) doing something with the utter lack of even remotely validated brain dose constraints isn't ideal (and as you correctly mentioned, the SF dose constraint in that study was too high especially by your standards!) Which is why the radionecrosis rate was so high, 3) recall that there was an inherent selection bias in that study in that the single fraction group received pretty low doses (at least IMHO). The BED/EQD2 of the Minniti regimen is roughly the same as 18 Gy x1. Meaning that I'd much rather prefer to give a higher dose than that if I can... In which sense, fractionating it gives And if not, I would gladly give three fractions. In other words, my criteria for giving three fractions would be if the constraints for single fraction would not be met with a high enough dose. Your argument for time-dependent radionecrosis is totally correct, but recall that the 90% local control rate in Minniti is also a one year value that will only go down if they are followed longer, i.e. the analogous argument (Longer life expectancy = higher risk for local failure).

Not unreasonable to have the concerns that you do. If using SF, we routinely do 18-20Gy x 1 to PTV (prescribed to 60-80% IDL) even for small lesions, so we have isoeffective (if you believe linear-quadratic for SRS/SBRT, which is a different rabbit hole I don't want to go down again) dosing regardless of what we use.

And yes, I imagine his two-year LC for MF is not 90% but will still be better than the SF done in that study.

To clarify, I still do SF for smaller lesions. If it's a single lesion I'll do SF even if V12 is 9cc. My main point was that we know V12 < 10cc is important, and I'm not aware of any data that suggests that is on a per lesion basis, but more of on a per-course basis. If you're treating multiple lesions, you need to reduce your per lesion V12 that you will accept.

 

[scarbrtj]

Do you have data for this? If this is the case shouldn't GK-SRS plans have much higher rates of radionecrosis (since they prescribe to the 50% IDL, meaning 200% of the prescription dose within the GTV) than linac-SRS plans (prescribe to 80% IDL normally, meaning 125% of prescription dose)?

I’m probably trying to have my cake and eat it too, and you can argue both ways. But in general what I’m trying to say is that toxicity correlates with dose. Max Dose is always on the right side of the equation; isodose and rx dose on the left. Maybe it’s more like in photography where we have the trinity of ISO, shutter speed, and f-stop to determine exposure. And in SRS it’s volume, isodose, and Rx dose. Can not fiddle with one without considering the other two. Exposure : photography :: Max dose : SRS.

 

[Neuronix]

I'll start by saying that I don't care at all about isodose assuming PTV is very small or zero. It's the margin and coverage that matters in my mind, and whatever isodose you need to get the best plan is what matters. This is equipment dependent.

Curious how everyone else is approaching these 2.5 - 3.0 cm borderline tumors.
And also for fractionated, what's your cut-off to go into 5 fraction range, and what dose are you using for 5 fraction? I've never personally gone into 5 fraction, but I see published data on 25/5, and I'm thinking why?

I've done 15 Gy in 1 fraction with GK for 3-4 cm tumors and done well with it. I know it's not in vogue right now, and some of the older series had poor control and high toxicity with it. One of my colleagues has been trying to get a large brain met ~15 Gy series published with nobody accepting of the results because of this.

Given that it's not in vogue and I don't see intact mets greater than 3 cm very often, nowadays I do 30 Gy in 5 fractions to 3-4 cm. I have been doing zero margin just like the UAB series. I'd say similar 5 fraction dose and similar technique is pretty popular among academics in my region.

I think most people consider 25/5 too low. The question among those I discuss this with has really has been whether 30/5 is enough and what really is the control at those sorts of doses. It's just not a common scenario to deal with unresected brain mets that large.

I am comfortable with the Minitti results, personally.

I'm not. I'll just put out there that I don't believe that three fraction is somehow magical over one fraction. I'm just skeptical that it provides significantly better control with less RT necrosis based on one series. Maybe I'm wrong and time will tell.

What bothers me is that I see people just doing 3 fraction now to everything, no concern at all about technique. It's really concerning to me what's going on in practice out there. And if fractionation is truly better for intact brain mets, why aren't we doing 5 fractions? I mean we could discuss going to even more fractions, but at least you get SBRT reimbursement up to 5 fractions to pay for what I hope is the good equipment and care necessary to give fractionated SRS.

I think V12 < 10cc is an old threshold when patients weren't living as long. Longer life expectancy = higher risk for eventual necrosis. That's not to say V12 < 5cc is a hard cut-off for say a single lesion but it is a consideration in a multi lesion plan.

The V12 data is pretty soft. Retrospective data that is all over the place, much of it in AVMs. There is almost no data in fractionated--but that's a newer technique and lack of safety data does not mean that it is intrinsically more safe.

For example if there are 5 lesions are you going to accept V12 < 50cc on the plan sum when doing single fraction since that's V12 < 10cc per lesion? I'd be more apt to fractionate if I saw anything like V12 < 25cc.

Nobody really knows what whole brain V12 should be. CE.7 limits it to V12 < 30 cc whole brain. Does anyone have a reference for where that comes from? I meant to ask Roberge that at NRG and forgot. I've discussed this with a number of CNS rad oncs and nobody agrees on this.

Ideally it'd be much lower than that. For me, anything over 10cc per lesion is for sure getting fractionated in my practice. This is for Brain with the GTV(s) subtracted out, which is how Minitti did it and calculated both a SF constraint (V12 < 13cc) and a MF constraint (V18 < 30cc).

For single fractions a reasonable approach is to just drop the dose until the V12 is less than some threshhold. For me that's 10 - 12 cc depending on the lesion. Control drops below 20 Gy dose, some people believe 18 Gy is still ok, most people don't like 15 Gy for gross disease (though for cavities it's more controversial).

 

[FrostyHammer]

I'll start by saying that I don't care at all about isodose assuming PTV is very small or zero. It's the margin and coverage that matters in my mind, and whatever isodose you need to get the best plan is what matters. This is equipment dependent.



I've done 15 Gy in 1 fraction with GK for 3-4 cm tumors and done well with it. I know it's not in vogue right now, and some of the older series had poor control and high toxicity with it. One of my colleagues has been trying to get a large brain met ~15 Gy series published with nobody accepting of the results because of this.

Given that it's not in vogue and I don't see intact mets greater than 3 cm very often, nowadays I do 30 Gy in 5 fractions to 3-4 cm. I have been doing zero margin just like the UAB series. I'd say similar 5 fraction dose and similar technique is pretty popular among academics in my region.

I think most people consider 25/5 too low. The question among those I discuss this with has really has been whether 30/5 is enough and what really is the control at those sorts of doses. It's just not a common scenario to deal with unresected brain mets that large.



I'm not. I'll just put out there that I don't believe that three fraction is somehow magical over one fraction. I'm just skeptical that it provides significantly better control with less RT necrosis based on one series. Maybe I'm wrong and time will tell.

What bothers me is that I see people just doing 3 fraction now to everything, no concern at all about technique. It's really concerning to me what's going on in practice out there. And if fractionation is truly better for intact brain mets, why aren't we doing 5 fractions? I mean we could discuss going to even more fractions, but at least you get SBRT reimbursement up to 5 fractions to pay for what I hope is the good equipment and care necessary to give fractionated SRS.



The V12 data is pretty soft. Retrospective data that is all over the place, much of it in AVMs. There is almost no data in fractionated--but that's a newer technique and lack of safety data does not mean that it is intrinsically more safe.



Nobody really knows what whole brain V12 should be. CE.7 limits it to V12 < 30 cc whole brain. Does anyone have a reference for where that comes from? I meant to ask Roberge that at NRG and forgot. I've discussed this with a number of CNS rad oncs and nobody agrees on this.



For single fractions a reasonable approach is to just drop the dose until the V12 is less than some threshhold. For me that's 10 - 12 cc depending on the lesion. Control drops below 20 Gy dose, some people believe 18 Gy is still ok, most people don't like 15 Gy for gross disease (though for cavities it's more controversial).

Every point in here is absolutely spot on. Particularly the thoughts RE: abuse of the Minnitti data and the quandary over 25-30 Gy in 5 fx.

 

[BobbyHeenan]

Great post, Neuronix.

I have a GK perfexion but tend to fractionate bigger mets more than my partners, but the data is not robust enough to be dogmatic in either direction I feel. Would love to see a randomized trial in this space but sooo many confouders like immunotherapy and different histologies at play too make it hard I'm sure. We enrolled on the post op whole brain versus SRS trial and would participate in a single versus multi fraction trial if one ever came about.

It is interesting that the V12 data for mets (Irradiated volume as a predictor of brain radionecrosis after linear accelerator stereotactic radiosurgery. - PubMed - NCBI) didn't subtract out GTV when calculating V12, but the Minniti Italian data did.

 

[Neuronix]

We enrolled on the post op whole brain versus SRS trial and would participate in a single versus multi fraction trial if one ever came about.

For cavities, a cooperative group trial of fractionated vs. single fraction recently opened: Single Fraction Stereotactic Radiosurgery Compared With Fractionated Stereotactic Radiosurgery in Treating Patients With Resected Metastatic Brain Disease - Full Text View - ClinicalTrials.gov

For intact mets it's tough because fractionated is only considered a standard option for large mets. Enrollment seems challenging since most of them get resected. Also, what would the dose be? There is no clear standard in my opinion. Maybe just pick something?

 

[BobbyHeenan]

For cavities, a cooperative group trial of fractionated vs. single fraction recently opened: Single Fraction Stereotactic Radiosurgery Compared With Fractionated Stereotactic Radiosurgery in Treating Patients With Resected Metastatic Brain Disease - Full Text View - ClinicalTrials.gov

For intact mets it's tough because fractionated is only considered a standard option for large mets. Enrollment seems challenging since most of them get resected. Also, what would the dose be? There is no clear standard in my opinion. Maybe just pick something?

You're right - it would just be picking something...but it's not like we don't do that sometimes anyway. Or could lump 27/3 in with 30/5 and make it dealer's choice if randomized to the fractionated arm.

Thanks for the heads up on the single versus multi fraction cavity study. I'm going to look into that.

 

[Mandelin Rain]

Such a rad onc thread. I love it.