"Ever since I started working with protons, I have questioned whether an RBE of 1.1 is appropriate in clinical practice," said Radhe Mohan, professor in the department of radiation physics at the MD Anderson Cancer Center. Speaking in the plenary session of the recent ICTR-PHE in Geneva, Mohan suggested that a variable RBE may be a better option.

"For carbon ion treatments, there's no question that we need to use a variable RBE," Mohan told the delegates. "But with protons, we continue to insist on using 1.1. Then just in case RBE does not equal 1.1, we avoid directing beams towards critical structures. I see no reason not to use a variable RBE – I say that 'RBE = 1.1' is a myth."

This value of 1.1 comes from averaging the results of many in vitro and in vivo experiments, mostly conducted at high doses-per-fraction and in the middle of the spread-out Bragg peak (where RBE is relatively constant). In reality, however, RBE is nonlinear a function of multiple variables, including linear energy transfer (LET, which varies along the Bragg curve), the dose-per-fraction and the α/β of the tissue under irradiation.

Mohan presented the results of experiments examining RBE at different depths for scanned proton beams and multiple cell lines. RBE was seen to be a linear function of LET up to the Bragg peak, but beyond the Bragg peak, rose nonlinearly.

So why is the generic value of 1.1 so widely applied? One justification, said Mohan, is that the increased RBE only affects a negligible region at the end of the beam, extending the biologically effective range by a millimetre or two. However, he explained, the distal edge spreads as it passes through tissue, resulting in the increased RBE encompassing a region that may be much larger, depending on the degree of tissue heterogeneity.

Advocates also state that clinical data don't suggest a need to change. But Mohan noted that we are beginning to encounter unexplained toxicities and recurrences. These may be attributed to a range of factors, one of which may be the assumption that RBE is 1.1. "It is hard to find unequivocal evidence," he said. "We need to reduce other uncertainties and see if RBE is a factor or not."

Mohan described an investigation into the correlation between high-dose/high-RBE volume and brain necrosis in a patient treated with passively scattered proton therapy. The study showed that the entire necrotic volume (as seen on MRI) was in a region where the dose and RBE were high. Whether RBE was a factor is difficult to determine, he explained, but the findings demonstrated some association between high RBE and necrosis.

Maximizing the gain

The clinical benefit of protons lies the differential between entrance dose and Bragg peak dose. Multiplying both values by 1.1 does not change that. However, using a higher RBE value at the Bragg peak confers an additional 30–40% gain – which we're not yet taking advantage of. This additional gain could be exploited by incorporating variable RBE in the optimization of intensity-modulated proton therapy (IMPT), to increase the likelihood of confining high-RBE protons to the target volume and away from normal tissues.

One way to do this is with "biological treatment planning", in which a plan is optimized by minimizing LET x dose. Mohan shared an example study in which such a plan was compared with one optimized on physical dose objectives alone. Dose-volume histograms revealed that both schemes provided similar dose to target, but the biological plan delivered reduced dose to the brainstem. "LET x dose-based constraints move planning in the right direction," he said.

Another option is to apply variable RBE-weighted optimization, using current RBE models and models yet to be developed. This work-in-progress approach, which accounts for the nonlinear dependence of RBE on dose and α/β, defines optimization criteria in terms of RBE-weighted dose or biological effect. Future RBE models would also consider the nonlinear dependence on LET.

Mohan showed some head-and-neck IMPT plans calculated using variable and fixed RBE optimization. The LET in the target was higher in all cases using variable RBE. In addition, variable RBE-optimized plans delivered less dose to the normal tissue shell surrounding the clinical target volume. "This does have some effect, we need to see if it will work in real life," he added.

To conclude, Mohan stated that current proton RBE models are based on insufficient and inconsistent data, and as such, likely underestimate RBE, especially around the Bragg peak. Using RBE=1.1 as standard may contribute to unanticipated toxicities and recurrences. "The use of variable RBE, even with currently limited knowledge and insufficiently developed models, may lead to safer and more effective IMPT," he said.

Sticking with 1.1

Tony Lomax, chief medical physicist at the Paul Scherrer Institute in Switzerland, does not entirely agree. Speaking in a conference session entitled "Revisiting RBE", he presented his own thoughts on proton RBE.

Lomax explained that in vivo RBE values are dependent upon a large number of unknowns, including tissue α/β, organ sensitivity, and tissue architecture and response, among others. And if patients do exhibit excess side effects from their treatment, there are many possible reasons for this, including, for example, smoking, surgery, hypertension, diabetes or age.

With at least four unknown factors that can influence in vivo RBE, plus at least eight additional confounding factors that can affect patient outcomes, the biological differences between photons and protons are too complicated to express in a single value, he said. In short, there is just too much uncertainty in the RBE value for human tissue to implement variable RBE.

The way forward, Lomax said, is to work with what we know. Presenting the "Lomax-eye-view", he concluded that while it may be possible to derive global RBE for tumour control, it is highly unlikely that localized RBEs in the patient could ever be deduced, given the generally low rate of severe side-effects and large number of unknowns.

Rather than developing sophisticated RBE models, we should instead analyse proton therapy outcomes in terms of measurable parameters, such as dose and LET. If unexpected outcomes do occur, we should relate these to measurable parameters and define corresponding tumour- and organ-specific dose and LET constraints.

"To do this, we need consistency across centres," Lomax explained. "That's why I think we need to stay with 1.1."

Related stories

• Topical review: Relative biological effectiveness values for proton beam therapy
• Fixed RBE is not ideal for proton planning
• Spatial distribution impacts proton RBE
• RBE variations impact proton range