When deciding whether to include an IMN field in a PMRT case, what do you consider a "safe dose" to the heart that you are willing to accept?

I agree with @Dr. First Last about keeping the heart dose as low as possible.  Aim for a mean heart dose of ≤ 4 Gy.  I prioritize the heart dose constraint, especially in younger women, women with pre-exisiting cardiac conditions and those who have had cardiotoxic chemotherapy.  This may sometimes require compromsing on PTV coverage, especially when using the current B51 standards which tend to increase the target volumes compared to pre-contouring 3D tangential techniques.  I have found that a majority of women will benefit from breath hold (DIBH or ABC) to reduce cardiac dose.  However, if you don't have breath hold, if the patient cannot cooperate or if the heart does not move, then other techniques can be used.  Possibilities include using non-tangential IMRT (there are several published techniques), custom electron bolus or electron arcs (though these techniques require some special devices and knowledge), or old fashioned electron strips, which if I must use I prefer to put on the lateral chest wall. 

I do not routinely treat IMNs in all node positive patients, so am perfectly willing to compromise on IMN coverage or omit those nodes from the PTV in order to prioritize heart dose.  I am more likely to want to include IMNs in higher risk patients:  multiple axillary nodes with inner quadrant tumor, sentinel node drainage to the IMN, locally advanced or inflammatory.  Special consideration must be taken when the IMNs are grossly involved and thus need to be treated to 66 Gy or higher, in which case a composite plan of all adaptive iterations should be evaluated up front to ensure that heart dose constraints can be maintained.

We routinely treat the IM nodes and chest wall in breath hold for left sided cases. Our average mean heart dose typically ranges from 2.5 to 3.5 Gy, which we feel is acceptable. In general, I follow the ALARA principal with heart dosing and will try to optimize my plan as much as possible, which usually ends up with a mean heart dose in that range. In certain cases, however, I've treated patients with a mean heart dose of about 5 Gy if their anatomy is unfavorable and I've done everything I can to optimize their heart dose. However, most of the time we can do better than that.

We try to make sure no direct photon beam tangent goes through heart by using a partially wide tangent or a combination of photon and electrons. We dont have ABC but that is certainly an option to avoid that.

By doing this our V25 is usually 1 to 2%. The current NRG protocol I think allows upto 5%.

@Dr. First Last, can you please share the references for non-tangential IMRT?