Importantly, I feel that any planning done by linac based radiosurgery system should probably be compared to the Gamma Knife system in terms of conformality index.

There is an interesting paper (published in the Red Journal recently) which speaks about similar issue while planning for VMAT.

http://www.scribd.com/full/47239321?access_key=key-1ifh4qt91exjg3l5l3le

Correct me in my conclusion then:
For a spherical lesion if it can be included in a single arc, it's worthwhile to include this in a single isocentre because any gain perhaps in terms of multiple isocentres would be marginal and increase set up time including hassles for QA.

If the lesions are wide apart, multiple isocentres make sense but one has to set up the machine for individual isocentre.

If the lesions are irregular multiple isocentres are required to ensure complete coverage of the tumor.

I think the major issue with using X Knife with multiple isocentres could perhaps be widening of the penumbra which does not seem to be a major issue with Radiosurgery.

I dont think using multiple isocentres can increase the penumbra which is primarily a function of radiation source size, distance and the location of the tertiary collimator as well as the beam energy. I think what you are referring to would be an increase in the dose spillage due to the effect of the intersecting beams

Now, is there any objective evidence that mutliple isocentres increase the conformality say in a given spherical volume without spilling off the excess dose

A very simple experiment to do on a TPS actually. I will see if any of my physicists will agree to waste some time on this so I can get back at you with concrete results. However that said I dont think using multiple isocenters would lead to a straightforward increase in the integral dose. Taking an example of brain metastasis using LINAC prescribing 18 Gy - during planning the objective is to give a peripheral dose of 18 Gy such that 18 Gy covers the entire target. So if you are using 1 or 3 isocenters the planner will adjust the dose at each isocentre so that 18 Gy covers the target with minimal tissue overlap i.e. in other words the dose gets divided between isocenters. A same thing happens in case of gamma knife. The fixed multiple beam orientation of gamma knife will not lead to a significant increase in integral dose as well the dose gets distributed all across the brain. In case of LINAC based radiosurgery for a single leison where you will have to use say 4 arcs if using 2 isocenters instead of 3 if using 1 yes the integral dose can increase - not because you are increasing the marginal dose but because you are irradiating a greater volume of tissue using a higher number of arcs. That is one of the reason we try to minimize the number of arcs in LINAC based stereotaxy. Another important reason is QA time - both pretreatment and on treatment. It takes about 45 min to deliver 18 Gy to 2 leisons using frame based sterotaxy in our center using Novalis using 2 -3 arcs on each leison. This time is less than what would be incurrred if we were doing a framless stereotaxy as then we would tend to image them with exac trac before each arc (while in frame based treatment we tend to image them once for each isocenter). The QA time is alos about 1 hour or so. I know time should not be a consideration when it comes to treatment but in terms of patient comfort and logistics it makes a lot of sense to reduce the number of isocenters if you can achieve the same plan quality.

I think the major issue with using X Knife with multiple isocentres could perhaps be widening of the penumbra which does not seem to be a major issue with Radiosurgery.

Inherently, I am veering towards the idea of a heterogeneity of dose within a tightly confined volume but as Santam rightly points out, we might as well be hitting in the dark. But then, keeping the hot spots well in the target is not harming anyone; only we don

Now, is there any objective evidence that mutliple isocentres increase the conformality say in a given spherical volume without spilling off the excess dose? I can foresee a simple experiment to set up a spherical phantom with dosimetry to the be done at the edges and say to measure the distance between the prescription isodose and 50% as cut off. I could not find any objective published reference to my assertions.

One more issue. Planning multiple isocentres would probably increase the integral dose; not an issue in palliation but definitely for the benign cases. Has the use of mMlcs solved this problem? Or perhaps the use of newer TPS? In that case how do you go about it to reduce the integral doses in case multiple isocentres are required?

Is it justified to keep a single isocentre for say spherical target? Or multiple isocentres for the same? Or any other recommendation?