Dong and co-researchers in Texas have conducted a systematic study to quantify uncertainties in SPR, a parameter used to calculate the water-equivalent path lengths of treatment beams. They confirmed that 3.5% is appropriate as a general recommendation to account for SPR uncertainties. Nevertheless, they demonstrated for the first time that SPR uncertainty varies significantly with tissue type and recommend the use of customized treatment margins to account for such variations (Phys. Med. Biol. 57 4095).
Existing guidance is adequate
"In our study, we found that the magnitude of uncertainties in different tissues is quite different," said physicist Ming Yang, first author of the study. "[So] although the general recommendation is adequate, the margin could be customized for each beam direction, accounting for differences in different tissues – bone, lung or soft tissues – in the beam path,"
Customization would provide coverage of the tumour, while minimizing dose to healthy adjacent tissue. "A three per cent margin would be enough for typical prostate and head-and-neck cancers," said Dong, who is now based at Scripps Proton Therapy Center (San Diego, CA).
The study was conducted as part of a larger effort to develop a dual-energy CT scanner, potentially using both kV and MV photon energies. Such a scanner provides a more accurate picture of tissue composition, enabling uncertainties in proton range and therefore tumour volume margins to be minimized.
Five sources of uncertainty
The researchers used a combination of previously published data and new, measured data to assess five contributions to SPR uncertainty across three main tissue types: low-density lung, intermediate-density soft tissue and high-density bone. Four of the uncertainty contributions correspond to steps in the stoichiometric calibration method, the most widely used method for deriving SPR values for different tissue types. The fifth originates from the dose calculation algorithm used in a treatment planning system; while SPRs vary with proton energy and consequently vary along the proton beam path, some algorithms neglect this effect in the name of simplicity.
The researchers used the uncertainties calculated for each tissue type to estimate the composite uncertainty for each beam in the treatment plans of 15 patients undergoing proton therapy for lung, prostate and head-and-neck cancer.
Uncertainty in individual tissue types, expressed to one standard deviation, ranged from 1.6% for soft tissue to 5.0% for lung. This translated to smaller composite uncertainties for a given beam path, a consequence of the relative abundance of soft tissue in the body compared to lung and bone. Expressed using the 95th percentile, beams used to treat lung tumours exhibited the greatest composite uncertainty of 3.4%. Beams treating prostate and head-and-neck cancers both resulted in an uncertainty of 3.0%.
"We were a little bit surprised by the actual numbers, which are almost identical to the commonly accepted standard of 3.5%. We expected a large uncertainty coming from the variation of elemental composition of human tissue," said Yang.
In light of the broad agreement of the study with the 3.5% reference standard, the researchers have no immediate plans to change their clinical practice. They view in-house implementation of customized margins as too cumbersome. They hope the calculations will eventually become integrated as part of a commercial treatment planning system, making their introduction to the clinic more feasible.
• Related articles in PMB
Comprehensive analysis of proton range uncertainties related to patient stopping-power-ratio estimation using the stoichiometric calibration
Ming Yang et al Phys. Med. Biol. 57 4095
A combined molecular dynamics and Monte Carlo simulation of the spatial distribution of energy deposition by proton beams in liquid water R Garcia-Molina et al Phys. Med. Biol. 56 6475
Does kV–MV dual-energy computed tomography have an advantage in determining proton stopping power ratios in patients? M Yang et al Phys. Med. Biol. 56 4499
Energy loss measurement of protons in liquid water
T Siiskonen et al Phys. Med. Biol. 56 2367
Bragg peak prediction from quantitative proton computed tomography using different path estimates
Dongxu Wang et al Phys. Med. Biol. 56 587
Theoretical variance analysis of single- and dual-energy computed tomography methods for calculating proton stopping power ratios of biological tissues
M Yang et al Phys. Med. Biol. 55 1343