Transcript
So, as Dr. De Salles already mentioned about the title, at UCLA, we changed the protocol during three periods of time. And we decided to do a retrospective analysis comparing which was the impact on facial numbness. So, the purpose was then to compare, if progressive increased doses to the root entry zone and therefore to the Pons would have an impact on the facial numbness outcome. That's the most common complication from radiosurgery for trigeminal neuralgia. And the rationale is that if we're increasing the dose to the transition myelin zone that's more radiation-sensitive, and that's where it's believed to be the injury towards the trigeminal system, we would be pumping up the rate of complications we would expect.
So, the inclusion criteria was only patients who were treated with 90 GY:REZ, at the beginning of the series at UCLA, there was a dose-escalation done and we didn't want to mix the data with also variation into the dose. So, we only selected patients receiving exactly the same dose, treated with exactly the same collimator size that was the 5-millimeter collimator and had at least six months of clinical follow-up.
So, our main outcome was facial numbness. Our main predictor was time period what translates into radiation dose to the root entry zone getting the trigeminal nerve was seen by kinda very key studies lies from 2 to 3 millimeters from the emergence from the Pons. And other covariates analyzed were gender, it's believed that female tend to report more facial numbness after treatment for trigeminal neuralgia due to any ablative technique. Etiology, we know that patients with secondary trigeminal neuralgia, they already have some kind of injury to the trigeminal pathway and that may lead them to be more prone to develop numbness after another ablative procedure, in prior procedures for the same reason I just described to you.
So, this is pretty much our follow-up questionnaire that patients fill up in clinic when they come for follow-up. Those that don't come for follow-up we go over this questionnaire over the phone. And there is this question there. Did you experience decreased facial sensation? We also have the neurological exam performed and that's in their clinical reports. And we evaluated those two sources to get our rates.
Regarding the time periods first, period ranged from August '95 to December '99. In the isocenter was positioned with the 20% isodose line tangential to the brainstem surface. That was the beginning of this series, we had only 12 patients because many of these patients treated with the 20% isodose line tangential to the brainstem also had less than 90 GY:REZ. And this is just a magnification to see this is the 20% isodose line.
That's the Pons surface, here you'll have a more global view, so you can have a better understanding of this magnification here. And the purple is the 30% isodose line. Then from January 2000 to February 2003, we had the prescription of the 30% isodose line and then we treated 50 patients like that again 30%, 30% here. And finally, oh, another point this is another example, with this prescription to the 30% isodose line and this is the 10. And we also had some changes in the arcs distribution throughout this period. And this is another example. So the same 30% isodose line, but a slightly different distribution, because of the arcs positioning.
And after March 2003, until present, but in this instance, until December '05, the 50% isodose line tangential to the brain stem, so you're just seeing probably the two most extreme groups presenting here. Dr. De Salles has the most anterior target and we're among one that have the most posterior targets. And in this analysis, 40 patients treated like that. And here you can see the 50% isodose line. This is the 30%, this is the 20%, so, pretty much, we have 20% isodose line inside the brainstem. In here's just to show the gasserian extension and what would be the dose distribution at this region.
So the demographics according to period, I'll just breeze through this. The majority of the pa...I mean, the age variation was really narrow. They were in average, our mean is 64 years, majority of the patients were female, majority of the patients had essential trigeminal neuralgia. The secondary causes were zoster, multiple sclerosis, and benign tumors. Regarding distribution, majority of the case had V2 and V3 pain, if you add these up, it won't be 12, 40 and 50, because many patients had many branches pain.
In previous procedures, we use median, and interquartile range because this variable had a skewed distribution. And what we can gather from here is what has been happening throughout the treatment with radiosurgery. So, we started by treating the very refractory patients, so they had many prior procedures. But as time goes by, and it's established as a very minimally invasive therapy with good results and acceptable level of complications, more and more patients without, even candidates for microvascular decompression, prefer to have a first try with radiosurgery. That's precisely again, what we see here. So, less patients in this group with prior procedure, majority of the prior procedures were radiofrequency whereas others were all microvascular decompression. And pain outcome, so here we basically average 67% of good results grade 3 and 4 simplifications Dr. De Salles went through with you. Here we go a little bit up and then here we are at 82.5%. But also the range of follow-up is different. So, we cannot drive any conclusions that this represents this group did better than this so far.
Numbness incidence, in our protocol, and we have a very careful evaluation of this patient, we have a high rate of facial numbness, but this is any facial numbness. It doesn't mean they have important facial numbness, and we grade that from one to five, and five is anesthesia, one is barely they notice. So, in average, our patients have grade two facial numbness. Now, what we're using now, so it gets clear. This is our current protocol, we use the 50% tangential to the Pons. And that's the distribution. Right now we also prescribe the 90 GY:REZ. We can generally just change to the 4-millimeter collimator, we use 7 non-coplanar arcs equally spaced to two passes per arc and in average the duration of the treatment is 45 minutes.
Regarding statistical analysis, we used SAS software. And we compared the proportion of numbness among the three treatment groups for a period of time and computing 95% Confidence Interval for the between-group differences. And we added the covariates I already went through with you. And we use logistic regression since this is a bi-variant outcome.
So in the first model, we included only the time period. And you can see that the odds ratio estimate there was a decrease of 60% odds to develop facial numbness in those treated with the 20% isodose line in comparison to the reference group that was the 50%. But the confidence interval crosses the no value, therefore it was not statistically significant. This number became even more prominent, slightly more prominent when we added gender and age to the model.
In the final model we added all the covariates. And again, the number one means the group we used for reference in the model. And here for the first period, we still didn't reach a statistical significance. But the second period in comparison to the third what means the 30% isodose line in comparison to the 50% isodose line showed that if we use the 30% in comparison to the 50%, there is a 60% decreased odds to develop numbness in the group that was treated with the 60%. So just confirming what we were expecting. Now, we didn't reach significance here but again, we had only 12 patients, and we did power calculations before and we were 80% powered, I mean just makes sense to talk about power, because we did that before, to detect the difference of 40%. So, if you have less than 40% difference, we wouldn't be able to detect with this fixed sample size. So, it doesn't mean it doesn't exist.
Regarding model discrimination, it showed to be better in the third model that included all the covariates.
So, in discussion, our finds are in agreement, what is out there in the literature. But many of the authors arrive to this conclusion, just analyzing their data without varying the protocol. It's just like the random difference on the trigeminal nerve length, or the volume of the cistern and they didn't on purpose vary the dose to the root entry zone. So, our experience by changing protocols over time, offers a unique opportunity to look for that.
It's fair to assume that it is interindividual variation because we know nerve lengths, we will alter the final dose to the Pons, even if you're using a close target to the brainstem. The angle that the nerve exit the Pons also have an influence on the amount of brainstem receiving the dose. But it's fair to assume that this was randomly distributed among these three groups. So, this is a teleology difference, the study showing increased numbness according to pre-determined higher doses of radiation to the brainstem.
The risk difference, if you make the calculations based on those models I showed to you was 19.7%, when I compared the 30% to 50%, that's the 95% confidence interval. What would translate clinically is that we would need to treat six patients with the 30% isodose line, to avoid one numbness in the group receiving a closer approach to the brainstem.
And so, but all of this was a retrospective analysis. And we only can be sure that these trends are happening if we develop the randomized control trial. And I think the main message is, I'm just showing to you the complication aspect of it. But the rationale behind is that, if we can prove that also better pain outcomes are achieved, with a closer target to the brainstem and recurrency rates are lower then it's acceptable to propose this to the patient and let him know or her know in advance that they are also running a higher risk of facial numbness. If better pain outcomes and less recurrence rates are not confirmed, then we should move away from the brainstem, but that needs to be proved in a randomized control trial.
And I'd like to thank and mainly I'd like to thank you, Dr. Selch, Dr. Agazaryan, Dr. De Salles because there have always been very supportive and enthusiastic of this research. Thank you.
So, the inclusion criteria was only patients who were treated with 90 GY:REZ, at the beginning of the series at UCLA, there was a dose-escalation done and we didn't want to mix the data with also variation into the dose. So, we only selected patients receiving exactly the same dose, treated with exactly the same collimator size that was the 5-millimeter collimator and had at least six months of clinical follow-up.
So, our main outcome was facial numbness. Our main predictor was time period what translates into radiation dose to the root entry zone getting the trigeminal nerve was seen by kinda very key studies lies from 2 to 3 millimeters from the emergence from the Pons. And other covariates analyzed were gender, it's believed that female tend to report more facial numbness after treatment for trigeminal neuralgia due to any ablative technique. Etiology, we know that patients with secondary trigeminal neuralgia, they already have some kind of injury to the trigeminal pathway and that may lead them to be more prone to develop numbness after another ablative procedure, in prior procedures for the same reason I just described to you.
So, this is pretty much our follow-up questionnaire that patients fill up in clinic when they come for follow-up. Those that don't come for follow-up we go over this questionnaire over the phone. And there is this question there. Did you experience decreased facial sensation? We also have the neurological exam performed and that's in their clinical reports. And we evaluated those two sources to get our rates.
Regarding the time periods first, period ranged from August '95 to December '99. In the isocenter was positioned with the 20% isodose line tangential to the brainstem surface. That was the beginning of this series, we had only 12 patients because many of these patients treated with the 20% isodose line tangential to the brainstem also had less than 90 GY:REZ. And this is just a magnification to see this is the 20% isodose line.
That's the Pons surface, here you'll have a more global view, so you can have a better understanding of this magnification here. And the purple is the 30% isodose line. Then from January 2000 to February 2003, we had the prescription of the 30% isodose line and then we treated 50 patients like that again 30%, 30% here. And finally, oh, another point this is another example, with this prescription to the 30% isodose line and this is the 10. And we also had some changes in the arcs distribution throughout this period. And this is another example. So the same 30% isodose line, but a slightly different distribution, because of the arcs positioning.
And after March 2003, until present, but in this instance, until December '05, the 50% isodose line tangential to the brain stem, so you're just seeing probably the two most extreme groups presenting here. Dr. De Salles has the most anterior target and we're among one that have the most posterior targets. And in this analysis, 40 patients treated like that. And here you can see the 50% isodose line. This is the 30%, this is the 20%, so, pretty much, we have 20% isodose line inside the brainstem. In here's just to show the gasserian extension and what would be the dose distribution at this region.
So the demographics according to period, I'll just breeze through this. The majority of the pa...I mean, the age variation was really narrow. They were in average, our mean is 64 years, majority of the patients were female, majority of the patients had essential trigeminal neuralgia. The secondary causes were zoster, multiple sclerosis, and benign tumors. Regarding distribution, majority of the case had V2 and V3 pain, if you add these up, it won't be 12, 40 and 50, because many patients had many branches pain.
In previous procedures, we use median, and interquartile range because this variable had a skewed distribution. And what we can gather from here is what has been happening throughout the treatment with radiosurgery. So, we started by treating the very refractory patients, so they had many prior procedures. But as time goes by, and it's established as a very minimally invasive therapy with good results and acceptable level of complications, more and more patients without, even candidates for microvascular decompression, prefer to have a first try with radiosurgery. That's precisely again, what we see here. So, less patients in this group with prior procedure, majority of the prior procedures were radiofrequency whereas others were all microvascular decompression. And pain outcome, so here we basically average 67% of good results grade 3 and 4 simplifications Dr. De Salles went through with you. Here we go a little bit up and then here we are at 82.5%. But also the range of follow-up is different. So, we cannot drive any conclusions that this represents this group did better than this so far.
Numbness incidence, in our protocol, and we have a very careful evaluation of this patient, we have a high rate of facial numbness, but this is any facial numbness. It doesn't mean they have important facial numbness, and we grade that from one to five, and five is anesthesia, one is barely they notice. So, in average, our patients have grade two facial numbness. Now, what we're using now, so it gets clear. This is our current protocol, we use the 50% tangential to the Pons. And that's the distribution. Right now we also prescribe the 90 GY:REZ. We can generally just change to the 4-millimeter collimator, we use 7 non-coplanar arcs equally spaced to two passes per arc and in average the duration of the treatment is 45 minutes.
Regarding statistical analysis, we used SAS software. And we compared the proportion of numbness among the three treatment groups for a period of time and computing 95% Confidence Interval for the between-group differences. And we added the covariates I already went through with you. And we use logistic regression since this is a bi-variant outcome.
So in the first model, we included only the time period. And you can see that the odds ratio estimate there was a decrease of 60% odds to develop facial numbness in those treated with the 20% isodose line in comparison to the reference group that was the 50%. But the confidence interval crosses the no value, therefore it was not statistically significant. This number became even more prominent, slightly more prominent when we added gender and age to the model.
In the final model we added all the covariates. And again, the number one means the group we used for reference in the model. And here for the first period, we still didn't reach a statistical significance. But the second period in comparison to the third what means the 30% isodose line in comparison to the 50% isodose line showed that if we use the 30% in comparison to the 50%, there is a 60% decreased odds to develop numbness in the group that was treated with the 60%. So just confirming what we were expecting. Now, we didn't reach significance here but again, we had only 12 patients, and we did power calculations before and we were 80% powered, I mean just makes sense to talk about power, because we did that before, to detect the difference of 40%. So, if you have less than 40% difference, we wouldn't be able to detect with this fixed sample size. So, it doesn't mean it doesn't exist.
Regarding model discrimination, it showed to be better in the third model that included all the covariates.
So, in discussion, our finds are in agreement, what is out there in the literature. But many of the authors arrive to this conclusion, just analyzing their data without varying the protocol. It's just like the random difference on the trigeminal nerve length, or the volume of the cistern and they didn't on purpose vary the dose to the root entry zone. So, our experience by changing protocols over time, offers a unique opportunity to look for that.
It's fair to assume that it is interindividual variation because we know nerve lengths, we will alter the final dose to the Pons, even if you're using a close target to the brainstem. The angle that the nerve exit the Pons also have an influence on the amount of brainstem receiving the dose. But it's fair to assume that this was randomly distributed among these three groups. So, this is a teleology difference, the study showing increased numbness according to pre-determined higher doses of radiation to the brainstem.
The risk difference, if you make the calculations based on those models I showed to you was 19.7%, when I compared the 30% to 50%, that's the 95% confidence interval. What would translate clinically is that we would need to treat six patients with the 30% isodose line, to avoid one numbness in the group receiving a closer approach to the brainstem.
And so, but all of this was a retrospective analysis. And we only can be sure that these trends are happening if we develop the randomized control trial. And I think the main message is, I'm just showing to you the complication aspect of it. But the rationale behind is that, if we can prove that also better pain outcomes are achieved, with a closer target to the brainstem and recurrency rates are lower then it's acceptable to propose this to the patient and let him know or her know in advance that they are also running a higher risk of facial numbness. If better pain outcomes and less recurrence rates are not confirmed, then we should move away from the brainstem, but that needs to be proved in a randomized control trial.
And I'd like to thank and mainly I'd like to thank you, Dr. Selch, Dr. Agazaryan, Dr. De Salles because there have always been very supportive and enthusiastic of this research. Thank you.