Impact of very long time output variation in the treatment of total marrow irradiation with helical tomotherapy

Helical tomotherapy is an intensity modulated X-ray therapy modality that is capable of delivering highly conformal dose distributions [1, 2]. While the gantry rotates around the patients, the mutileaf collimator opens and closes to produce an intensity modulated beam with variable couch speed. The current treatment couch on the helical tomotherapy unit has a maximum travel length of approximately 150 cm.

These features allow for the delivery of intensity modulated beams for the treatment of very long targets like in the case of total body irradiation (TBI) or total marrow irradiation (TMI), which was first proposed by Hui et al. [3]. Dosimetric and clinical studies have been intensively performed and the feasibility of tomotherapy machine for TBI and TMI has been shown by several groups [4‐7].

On the other hand, TBI or TMI deliveries with tomotherapy machine sometimes require more than 30 minutes of beam-on time [3, 5, 6]. Although the new Dose Control System (DCS) has been offered to the current tomotherapy machine since 2011, most tomotherapy machines worldwide do not have DCS. Very long time output variation is therefore especially important in TBI or TMI planning delivery in the tomotherapy machines without DCS. Although several groups reported the output variations in imaging beam [8] or treatment beam modes [9‐11], no report has been published assuming very long time treatment such as TBI, and TMI.

Here we report the 30 minutes’ output variations in static and rotational procedures. Furthermore, the impact of long time output variations on TMI planning delivery was investigated.

All measurements were performed with an Exradin A1SL ionization chamber (Standard. Imaging Inc., Madison, WI, USA) connected to the 8-channel electrometer (Tomoelectrometer, Standard. Imaging Inc., Madison, WI, USA) supplied by the manufacturer. This electrometer can operate in two modes: local and software controlled. In the latter mode, the tomotherapy measurement system software (TEMS) is used to collect the measured charge in real time with a sampling time resolution that can be set as low as 100ms. In this way, the ion chamber readings vs. time can be collected throughout the overall radiation delivery. Two phantoms were used: a set of solid water stacks and a cylindrical solid water phantom, both supplied by the manufacturer (Accuray, Inc. Madison, WI, USA). Treatment planning and dose verification analysis were done with TomoTherapy Planning Station (Accuray, Inc., Madison,WI, USA).

Although the new DCS has been offered to the current tomotherapy machine since 2011, there are still a number of tomotherapy machines without DCS in clinics all over the world. In the present study, we investigated the very long time output variation in this type of tomotherapy machine which is known to have output variation, typically about ± 2%, due to the absence of a dose servo system [12, 13]. Since the treatment plans are based on a constant output, a variation between planned and delivery could result. For example, Francois et al. reported that the output in rotation during 800 seconds looked very unstable just before target change [9]. The output variation is therefore concerned especially in the treatment of TBI or TMI which requires about 30 minutes of beam-on time. To our knowledge, this is the first report that investigated the very long time variation assuming TMI delivery.

Figure 2 (a) shows the long time output variation measurements of the static beam mode. These were repeated 3 times. The red line indicates the reference dose rate of 894 MU/min that was given in the tomotherapy treatment planning system in our institute. The output gradually decreased during 30 min with an average (±SD) of −2.9% ± 0.2%, -3.4% ± 0.3%, and −3.4% ± 0.3% at 10 min, 20 min, and 30 min, respectively (Figure 2 (b)). The output variation was up to 5%. On the other hand, static output variations from reference dose rate during 30 min averaged −2.0% ± 0.4%, -2.4% ± 0.5%, and −2.8% ± 0.5% at 10 min, 20 min, and 30 min, respectively. The steep output drop-off was observed within 3 minutes from beam-on.

The same tendency was observed in rotational beam delivery. Figure 3 shows the long time output variation measurements for a rotational radiation delivery of 10 seconds/rotation gantry period. Measurements were repeated 4 times. The output gradually decreased with an average (±SD) of −2.5% ± 0.7%, -3.1% ± 0.7%, and, -3.5% ± 0.8% at 10 min, 20 min, and 30 min, respectively. The output decrease from beam-on was up to 4.5%. Again, the steep output decrease occurred within the first 3 minutes from beam-on.

Although up to 5% of long time output variations were observed, the TMI planning dose agreed well with the film and the ion chamber measurements. Figures 4 (a) and (b) show the longitudinal profiles of head and foot regions in the model TMI planning, respectively. No remarkable dose differences were observed in both head and foot regions. Figure 4 (c) and (d) showed the gamma distribution with the criteria of 3 mm/3% in head and foot region, respectively. The pass rates were 97.7% and 92.2%, respectively. The ion chamber measurements in both head and foot regions agreed within 3% with the calculated dose.

In spite of continuous decrease in output during 30 minutes of irradiation, the impact of long TMI treatment time on delivered dose was small. One possible reason is that output decreased rapidly the first 3 minutes and then slowly in both static and rotational modes. In our institute, we calibrate the machine output at a dose rate of about 903 MU/min with the static beam at the depth of 1.5 cm in the field size of 40 cm (width) x 5 cm (length) at 85 cm SSD. Considering that the overall output dose fluctuation was 3% on average, then the delivered average dose rate was only 2% lower compared to the reference dose rate (894MU/min) in the treatment planning. Assuming it is applicable for rotational output variation, the magnitude of output variation from reference dose rate would be smaller in a 30 minutes’ TMI treatment. If larger output fluctuation is observed, then the use of a higher initial dose rate would reduce the impact on real TMI or TBI treatment delivery. However, we do not find any safety issue with the tomotherapy machine even without DCS.

In the tomotherapy machine, the treatment will be terminated if (i) the monitor chamber readings differ by more than 50% from their nominal rate for more than 3 seconds or (ii) the monitor chamber readings differ by more than 5% from their nominal rate for more than three consecutive rolling 10 seconds windows [12, 13]. In a long treatment time therapy such as TBI and TMI, these kinds of interruptions should be avoided because it increases treatment time.

Duchateau et al. reported that output fluctuations in the imaging beam mode with an average imaging time of 173 seconds (range; 111–281 seconds) were between 5% and 9% during the acquisitions [8]. Although the treatment beam and imaging beam originate from the same source, the rotational variation in treatment beam mode is typically on the order of ± 2% [12, 13]. Furthermore, it has been shown through simulated data that sinusoidal rotational variations on the order of ± 5% results in systematic dose uncertainty within ± 2% [10]. More recently, Staton et al. reported that simulation of 2% or 7% rotational output change results in less than 1%, or 2% for DVH values, respectively [11]. Our treatment plan verification results were therefore reasonable.

In conclusion, TMI delivery with helical tomotherapy without DCS does not pose a risk for significant deviations from the original treatment plan regardless of the output fluctuation during treatment. The magnitude of long time output drop-off may vary depending of machine issues of which the magnetron condition is critical. For example, replacement of old magnetron with new one may improve the long time output variation. In the TBI or TMI treatment with helical tomotherapy that does not have DCS, however, quality assurance of very long time output variation should be performed before the first treatment.