Incorporating detailed biology in hyperthermia treatment planning: a necessary condition for progress

Hyperthermia serves as an adjunct to radiotherapy and is particularly helpful in scenarios where radiotherapy alone achieves suboptimal local response. The most prominent example is when a past history of radiotherapy limits the radiation dose. Concurrent hyperthermia is known to be a potent radiosensitizer. In this issue of the IJH, van Leeuwen et al. [1] presented clinical treatment planning software that allows the practitioner to estimate the impact of a predicted temperature distribution on radiotherapy response. Arguably radiobiological modelling is more important to practitioners of thermoradiotherapy than to radiotherapy alone. The impact of hyperthermia can be substantial, but can vary sharply with both time and temperature–temperature changes of 1 or 2 degrees can markedly alter the likelihood of response. The paper by van Leeuwen et al. demonstrates how substantial these effects can be and, thus, confirms the usefulness of a biological treatment planning programme.

As the authors point out, there remain substantial gaps in our knowledge of how to model heat-induced radiosensitivity (HIR). Even with substantial error bars due to the present level of uncertainty, they are able to demonstrate the usefulness of biological modelling based pre-planning. The paper therefore provides support for future biology experiments to enhance our ability to accurately model HIR.

Although HIR has been known for some time, the detailed time and temperature dependence of the impact of hyperthermia on basic radiation response parameters is not well characterised [2–5]. It is to van Leeuwen et. al.’s credit that their group [6] has already contributed much to what we currently know.

Because each cancer type is heterogeneous, there will always be uncertainty in biological modelling of radiation response. For example, we know that, with radiation alone, the surviving fraction after a dose of 2 Gy (SF2Gy) typically ranges by a factor of ∼3 when multiple human cell lines for a given disease type are considered [7]. This does not mean that we cannot make predictions of the of impact of hyperthermia on radiation response. But, to do so, we would need functions that describe (or, more accurately, bound) the temperature and time dependence of the radiation response parameters (i.e. the coefficients α and β in the linear quadratic model of radiation response). This means studying radiation survival for one cell line in great detail: at multiple radiation doses between ∼1.8 and ∼5 Gy, multiple temperatures between 37 and ∼44 °C, and at multiple times (with the nearest time from hyperthermia to radiation ranging from ∼−1 h to ∼ +1 h). This would help to determine if the impact on radiation response parameters (α and β) is best described by an additive [8] vs. multiplicative factor [1] vs. a combination. It would help to determine how best to fit the temperature dependence of that effect, whether, e.g. the temperature dependence is better fitted to a single coefficient exponential (as van Leeuwen et al. use) or (as the Sapareto Dewey [9] formula would predict) the exponential dependence is bi-phasic with a break point at ∼42 °C.

In addition to intensively studying one cell line to determine which functions to use, other cell lines for the same disease type should be studied for at least two temperatures and two radiation doses, in order to verify the appropriateness of the formulas and to bracket the fundamental uncertainty for the fitted coefficients. This process would be repeated for other disease types.

All of this is time-consuming work. As stand-alone biology it might seem prosaic, even subject to the lethal, from a funding agency standpoint, label of “fishing expedition”. From a pure biology standpoint, it is always more appealing to discover a new effect than to characterise the details of an already known one. But, for clinical hyperthermia, as technologies like MRI-based thermometry and treatment guidance bring us into an era of much more precisely planned and delivered treatment, knowing how to model radiation response modification becomes a necessary step. A well-constructed multi-disciplinary research programme can be both exciting and innovative, even if one component of it asks questions that might otherwise be deemed dull.

It is very much hoped that funding agencies will support research programmes that, like the work represented by van Leeuwen et al., include the biology that the physics needs.