Objectives:

Spine SRS success is dependent upon high quality imaging and accurate dosimetric calculations, aspects that can be compromised by using titanium constructs commonly employed for mechanical stabilization. Recently, radiolucent CFR-PEEK has emerged as an attractive alternative to titanium due to superior imaging attributes which aid clinicians in post-operative follow-up imaging and radiation oncologists due to minimal dose perturbations of the SRS treatment plan. Our goal was to assess the impact of titanium and CFR-PEEK constructs compared to each other and to normal, non-implanted bone, using an anthropomorphic spine phantom. These different attributes were evaluated using multiple CT imaging and metal artifact reduction techniques and multiple, advanced calculation algorithms commonly available to clinicians.

Methods:

A customized thorax phantom, sectioned coronally to permit placement of radiochromic film and that contained interchangeable spine inserts of either bone, bone with a titanium construct, or bone with a CFR-PEEK construct was CT simulated. CT acquisition was performed with both single-energy (SECT) and dual-energy (DECT) techniques and using several metal artifact reduction (MAR) algorithms, to assess image quality differences between native bone and the different constructs. Subsequently, the CT data was used to generate a series of 55 SRS spine plans for each of the three spine inserts, using two commerical, treatment planning systems (TPS) (Varian Eclipse 16.1 and Brainlab Elements 3.0 & 4.0), four dose calculation algorithms (Varian AAA, AcurosXB and Brainlab Pencil Beam and Monte Carlo) and two photon energies (6FFF, 10FFF). Target volumes were created to mimic typical clinical SRS spine cases that treated the vertebral body while attempting to maximally spare the spinal canal and spinal cord. All plans were delivered on a Varian Truebeam Edge linear accelerator using Brainlab ExacTrac for image guidance. Radiochromic films was placed between each section of the phantom to and the measured versus calculated dose distributions were analyzed using gamma analysis, point dose and line profiles.

Results:

Titanium constructs showed pronounced image artifacts compare to bone or CFR-PEEK. DECT was observed to provide some mitigation of artifact as did some of the MAR algorithms. CFR-PEEK constructs demonstrated superior canal visualization and minimal image artifacts in the raw CT data, with further reduction observed when using DECT and MAR. CFR-PEEK screws contained a small radioopaque fiducial in the tip made from tantalum which was observed to cause minor streaking effects on CT but had negligible impact in film measurements on the plan dose distributions. Film analysis of titanium insert dose distributions showed errors of 15 to 20% between the TPS within 5 mm of the construct hardware. The absence of metal in the large majority of the CFR-PEEK construct resulted in films that agreed closely with the TPS and were comparable to measurements in native bone. Distant from the implanted hardware and along the spinal cord itself, all TPS algorithms agreed within 5% or better to film.

Conclusion(s):

In close proximity to the construct CFR-PEEK showed improved dosimetric accuracy, consistent with in-bone only measurements, when compared to conventional, titanium constructs. Moreover, CFR-PEEK resulted in superior image quality, even with only SECT and no MAR technique, compared to titanium. This would be a significant advantage for facilities that do not have access to more advanced image processing features of the newest generation of CT-simulators. The improved visualization of the vertebral body and spinal canal with CFR-PEEK is advantageous for target and spinal cord delineation and in MRI-CT image registration which is often challenging when titanium hardware is utilized. In many cases, the presence of titanium hardware in patients results in the need for a CT myelogram to aid in spinal cord identification for treatment planning. Follow-up imaging or retreatment of disease may also benefit from CFR-PEEK. All TPS algorithms predicted spinal canal dose to sufficient accuracy, even in the presence of titanium, provided the distance to the spinal cord from hardware exceeded 5 mm. Our study demonstrates several advantages for the SRS planning process of utilizing CFR-PEEK constructs compared to conventional, titanium hardware.