The Sharp End of the Photon

One of the biggest surprises for many of our new patients in radiation therapy is how much time it takes to get treated. They expect that they will come in, get treated, and be finished on that day. Imagine their surprise when they are told they will have to get treated each workday for up to nine weeks or more.

Of course, there is a solid biological reason for extending the treatment over such a long period of time. In my last post, I wrote about the need to balance killing the tumor with damaging the surrounding normal tissue. In this post I will show how using fractionation, or dividing the total dose into a number of smaller doses, can help improve what is referred to as the therapeutic ratio, the chance that the tumor will be killed versus the chance of a normal tissue complication.

I’ve tried to keep the discussion so far at a high level, but the easiest way to understand these concepts is visually so I’m going to use some figures to help illuminate my points. An important point to keep in mind is that the following discussion is a simplified model based on experiments done with cells in a lab. Still, to a certain extent, the basic features correspond to what we see clinically.

In this first figure, I’ve graphed two curves. These curves show the fraction of cells that survive after being irradiated with a certain dose (the x axis is dose in units of Gray, the SI unit for radiation dose). The red curve is for tumor cells and the blue curve is for normal tissue cells. Note that since cell damage from radiation is random, these survival curves are averages. If you make some statistical assumptions, you can say that a dose which kills a certain number of cells will have a probability of killing all of the clonogens in a tumor and therefore killing the tumor itself. At that dose, indicated by the vertical line on the chart, you can see from the other curve how many normal tissue cells are killed. We can convert that to a probability of normal tissue side effects for the curative dose. For most cases, the number of surviving cells from normal tissue will be higher than for the tumor. (The numbers on this chart are for illustrative purposes and are not what is given clinically.)

One thing to notice about the curves is that at low doses the curve is almost a straight line, while at higher doses it bends more sharply. Suppose we were to give the tumor a small dose, one that stayed within the linear part of the curve, and then wait until sublethal DNA damage had been repaired. Each small dose would then start back along the cell survival curve from zero. We could continue to give these small doses until we had achieved the same probability of killing the tumor as before.

This figure shows the results. The red curve shows the survival fraction of tumor cells, and the blue curve is for normal tissue cells. You can see the cell survival curve for a single dose and also a dashed line. This dashed line shows the effect of delivering small dose fractions and then waiting for DNA repair to complete. The straight black line connects the end points of each short hop along the single dose curve. It takes a much higher total dose to achieve the same fraction of tumor cells killed and, therefore, the same chance of tumor control. However, if you look at the curve for normal tissue, the survival fraction of normal tissue cells is also much higher. This means that the chance of side effects is lower.

If we massage the statistics a little bit more, we can generate curves of tumor control or side effect probability versus total dose for a single dose and a series of small doses. The red curve shows the probability for tumor control while the blue curve shows the probability for normal tissue complications. If we were to give the dose indicated by the vertical line, there would be a small chance of normal tissue complications, but a much greater chance of killing the tumor. The ratio between the two probabilities is called the therapeutic ratio. If we could somehow shift the tumor curve to the left (or towards less dose), the chance of tumor control would increase dramatically while the chance of side effects would remain the same. This is shown in the bottom figure. Conversely, if we could shift the normal tissue curve to the right (towards higher dose) and increase the dose to maintain the same complication rate, then we would also have a greater chance of tumor control. When we fractionate the dose, both curves move towards higher dose, but the normal tissue curve moves further than the tumor curve. In this way we can improve the therapeutic ratio.

Improving the therapeutic ratio is the central issue in radiation therapy. Fractionation is one way to do it. Determining fractionation schemes is difficult, however. Cell survival curves can be measured in a laboratory using cultured cells, but cells in the body react differently to radiation than cells in a Petri dish. Tissue tolerance can be tested in animals, but applying the data to humans is problematic. The best way to test fractionation schemes is to conduct a clinical trial, looking at the rate of complications and the rate of tumor control across a large number of patients. Clinical trials are time consuming and expensive and are well beyond the capabilities of small clinics. In practice, most clinics use well established schemes, using fractions of 1.8 to 2 Gray each weekday with a 2 day break on the weekend. There are other schemes such as treating a patient twice a day or using higher doses per fraction, but these are specific to the type of tumor and staging of the patient.

Another way of improving the therapeutic ratio is to decrease the amount of dose that normal tissue receives relative to the tumor. This will reduce the normal tissue complication rate without sacrificing tumor control. In practice, this is where most clinics spend their effort. The process is called treatment planning, and all clinics have at least one person called a dosimetrist who spends the majority of their time trying to accomplish this goal. There are a lot of different things a dosimetrist can try, but explaining them involves some knowledge of how the dose of radiation is delivered to the patient. So in the next few posts in this series, I will give an overview of how a linear accelerator works, and how we can use it to shape the distribution of dose in a patient.

References
All of the figures in this post are inspired by figures in the following references.
Hall, E and Giaccia, A (2005) Radiobiology for the Radiologist, Lippincott Williams and Wilkins
Perez, C et.al. (2007) Principles and Practice of Radiation Oncology, Lippincott Williams and Wilkins