Session Item

Treating patients with metallic implant with protons is in general (much) more complex than with photons, due to the high sensitivity of the proton range with respect to the density and composition of the material in the beam path.
At the same time, many patients affected by tumors that are considered well suited for proton therapy (e.g. chordomas) have metallic implants in the treatment field because of surgery. Therefore, solutions for a safe and effective treatment of these patients should be sought. The reality of clinical practice is that every proton therapy center has its own policy for accepting or refusing patients with metal implants, and these policies may change considerably from center to center.

For accurate treatment (planning) of proton therapy in presence of metallic implants, the main issues to be tackled have to do with:

•    Imaging
•    Material characterization
•    Dose calculation
•    Plan optimization and evaluation
•    Developing workflows that are sustainable in clinical practice.


High quality CT imaging is essential to acquire a description of the implant that is geometrically correct. Errors at this step are systematic. In the past 10 years the increased use of algorithms for reducing artifacts due to metallic material significantly improved the CT image quality in several clinic. Such algorithms should be considered a must for proton therapy. There may be situations (e.g. dental implants, or hip prosthesis), where the combination between high Z material and volume of the implant create artifacts that are so significant to make the treatment with protons impossible.

CT imaging may provide a good geometrical description of the implants, but it does not help determine the composition, as most metals will have a Hounsfield Unit (HU) equal to the maximum. Information on the composition must therefore be acquired in different ways. Ad hoc choices in the CT calibration curve may help an automatic association (i.e. without additional contouring) between the highest CT number and the metal most often encountered in clinical practice (e.g. titanium).

Dose calculation with pencil beam algorithm is known to be associated with potentially high errors, and the combination of metal implants, grazing beam incidence and the use of a pre-absorber is probably the most challenging scenario one can encounter in terms of dose calculation. The good news is that Monte Carlo algorithms are increasing the norm in proton therapy, and they are associated with markedly improved accuracy even in this situation.

Robust optimization and plan robustness evaluation have been developed with the aim of explicitly compensating for setup and range uncertainty, and they can be very useful in the specific case of metal implants. When possible, robust optimization on multiple CT datasets, where each dataset has a slightly different shape, or a different composition, of the metal implant, can help when such quantities are not totally certain. More robust dose distribution may not come “for free”, so one may need to decide case-by-case where the deterioration of the nominal dose is worth the additional plan robustness. One can also introduce variations in the metal implants shape or composition only during plan evaluation and not during plan optimization.

All the aspects mentioned above should be combined in clinical workflows that are sustainable in terms of time and personnel resources, and that can be part of a treatment process where re-planning may be needed.

There may still be situations where, despite all the efforts and the available tools, the proton dose distribution is overall inferior to a photon dose distribution. Proton center should always have this in mind and be able to refer to photons a patient initially intended for protons.