When performing IMRT treatment planning for head and neck cancer, how do you instruct your dosimetrists to manage the optimization process for targets that are very close to the skin surface (or with PTVs that extend into air), yet the skin itself is not at risk?

For elective target volumes it makes sense in most cases to pull in the PTV 3-5mm from the skin for planning, but for gross disease close to the skin surface (e.g. bulky adenopathy in someone with a thin neck; or even some elective PTVs in a patient with a very thin neck) I have seen the following approaches used: 1. Placing physical bolus material at the time of simulation over areas at-risk with no change in the PTV volume with planning. Downside: gives high skin doses. 2. Placing virtual bolus material within the planning system for the optimization process, with no change in the PTV volume, then a re-calculation without the bolus to give the actual plan. This tricks the optimization system into "flashing" the skin surface over the PTV with the beamlets. The assumption is that the tangential IMRT fields will give you enough dose just below the surface, even if not modeled accurately on the planning system. Dowside: higher dosimetry workload. 3. Still just subtract PTV 3-5 mm from skin and assume that the tangential IMRT fields and perhaps bolus effect from the mask will compensate at superficial depths.