This Article Abstract Figures Only Full Text (PDF) Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Essapen, S Articles by Tait, D Search for Related Content PubMed PubMed Citation Articles by Essapen, S Articles by Tait, D

Accuracy of set-up of thoracic radiotherapy: prospective analysis of 24 patients treated with radiotherapy for lung cancer

Departments of ¹Radiotherapy and Oncology and ²Computing and Information, Royal Marsden NHS Trust, Sutton, Surrey SM2 5PT, UK

	Abstract

Top Abstract Introduction Methods Results Discussion Conclusion References

 
In thoracic radiotherapy, a number of factors hinder the use of portal films and electronic portal imaging devices for measuring field placement errors (FPEs). The aim of this study was to assess the accuracy of treatment set-up using simulator check films (SCFs) in radiotherapy for lung cancer. Prospective evaluation was performed on 24 patients. During their radiotherapy, patients returned to the simulator weekly for a minimum of four SCFs, for which the parameters from the original simulator planning film were set, positioning being achieved without fluoroscopy. A total of 96 SCFs were taken. FPEs in left–right (L–R) and superior–inferior (S–I) direction, as well as coronal rotational errors, were measured. The mean absolute FPE was 0.35 cm in the L–R axis and 0.43 cm in the S–I axis. Statistically, the FPEs in the S–I direction were greater than those in the L–R direction (p<0.001). A margin of 0.93 cm between the clinical target volume and the planning target volume would cover 95% of FPEs in the L–R direction, whilst a margin of 1.13 cm is needed for this degree of certainty in the S–I direction. Mean coronal rotational error was 1.6°. Systematic errors were greater than random errors. This study demonstrated that the FPEs were within clinical tolerance (0.7 cm) in 84.9% of the measurements. The planning margins used in our clinical practice compare favourably with the FPEs in this study.

	Introduction

Top Abstract Introduction Methods Results Discussion Conclusion References

 
Quality assurance has become a major factor in radiation treatment. As more complex treatment set-ups become an integral part of radiotherapy, the requirements for accuracy become paramount. Quality control is a complex issue involving many different aspects, one of which is the accuracy and reproducibility of radiotherapy treatment set-up. A number of clinical studies have used electronic portal imaging devices (EPIDs) or portal films to quantify the errors associated with patient set-up for radiotherapy in the pelvis or head and neck region [1–4]. There have been relatively few studies quantifying these errors associated with radiotherapy for lung cancer.

Results of lung cancer treated with radical radiotherapy are disappointing. Attempts to improve this situation include dose escalation with conformal techniques, which allow increased sparing of normal tissue by more precise definition of the target volume and critical normal structures. Such precision imposes strict requirements for accuracy, as small shifts in patient movement or beam alignment may result in tumour under dosage and/or normal tissue over dosage.

In thoracic radiotherapy, the lack of fixed, identifiable bony reference points hinders the use of portal films and EPIDs for acquiring accurate measurement of variation in field placement. The vertebral bodies are poorly visualized using megavoltage energy (6 MV or 10 MV), and in the setting of a limited field area it may be impossible to define which particular vertebral bodies are being seen on a portal film. Although the trachea and bronchi are commonly used as reference structures, as they are often situated near the planning target volume (PTV) and are visible on most portal films, these are not appropriate landmarks because of their movement with respiration.

For the reasons given above, and after our initial unsuccessful attempts to use the EPID at our centre, the present trial was designed to use simulator check films (SCFs) (with diagnostic quality X-rays) to quantify the treatment set-up error in those patients receiving radiotherapy to the lung. The original simulator planning films (SPFs), taken as part of the normal planning procedure, were used as the gold standard, against which four simulator check films (SCFs), taken without fluoroscopy during the course of each patient's radiotherapy, were compared. These SCFs were acquired on the same facility as the original SPF, and with all parameters constant. On these films, the vertebral bodies were clearly visualized and allowed any displacement of the field centre in the left–right (L–R) and superior–inferior (S–I) axes, and any rotation in the coronal plane, to be measured.

Although lateral verification films are routinely taken prior to the 3-field phase 2 treatments, we did not attempt to measure movement in the anteroposterior (AP) direction as this would have been feasible for only half the study group, and therefore less useful for analysis.

24 patients were entered into the trial and the results of the accuracy and reproducibility of treatment set-up in thoracic radiotherapy are reported here.

	Methods

Top Abstract Introduction Methods Results Discussion Conclusion References

 
Trial design and patients
Between February 1995 and December 1997, 24 patients, who were planned to receive a course of radiotherapy to the lung, were recruited into this prospective study. Consent was obtained from all those entered. The group included 12 patients who had planned fields derived from CT images, positioned with their arms up above their heads for planning and treatment (known as the arms up group; AUG), facilitating the use of oblique treatment fields. The remaining 12 patients were planned using parallel opposed fields with their arms down by their sides (the arms down group; ADG). The aims of the study were to quantify the field placement errors (FPEs) and to assess the impact of these errors on clinical practice. In addition, an assessment of the random errors, i.e. errors due to day-to-day variations in patient set-up during radiotherapy, and systematic errors, i.e. differences in patient set-up between simulator and treatment, was made.

Treatment planning
All patients were positioned supine, except one who was positioned prone. Standard headrests on a hard, flat couch top were used to aid reproducibility. No immobilization devices were employed as this is not our standard practice. Patients planned for single-phase AP opposed fields were simulated with their arms by their sides (ADG). Fluoroscopy was used to define the treatment field required, as determined by the pre-chemotherapy gross tumour volume (GTV) if this was encompassable within an acceptable volume. A single anterior tattoo, in the centre of this field, was performed to assist in patient set-up for subsequent treatment. The prescribed dose for this group was 40 Gy mid-plane dose in 15 fractions over 3 weeks.

Those patients having fields planned with CT were positioned with their arms elevated above their heads, supported by an armband (AUG). At the time of scanning, anterior and lateral reference tattoos were made on the chest wall, in this position, to be used for subsequent treatment set-up. Axial images at 1 cm intervals were acquired, following which the GTV, PTV and critical structures were defined [5]. All patients in this second group were initially treated with AP opposing fields, followed by a second phase of treatment according to a 3-field plan. The combined phases of treatment aimed to deliver a total dose of 60 Gy in 30 fractions over 6 weeks to the PTV, with a maximum spinal cord dose of 46 Gy.

For all patients, an AP simulator planning radiograph was taken prior to treatment as verification of the intended treatment field. These SPFs were taken using a fixed focus-to-film distance (FFD) of 140 cm, and a suitable X-ray exposure was used to define the anatomy of the thoracic spine. These radiographs were used as the gold standard against which future SCFs were compared.

Simulator check films
During the course of their radiotherapy, patients returned to the simulator for a minimum of four SCFs. Parameters from the original simulator visit were set for each patient, as would be done on the treatment unit, using the instructions on the prescription sheet. Patient positioning was achieved by the use of skin tattoos and lasers, and fluoroscopy was not used prior to taking the SCF. The procedure adopted was to try and reflect practice on the treatment unit as much as possible. AP radiographs were taken at an FFD of 140 cm using the same X-ray exposure as the original simulator visit. These films were taken weekly for 4 weeks, but 13 patients whose treatment course lasted less than 4 weeks attended twice during the last week of treatment.

SCFs were taken by simulator staff, but did not necessarily include the staff involved in the original SPF. The simulator staff comprised permanent radiographers and those rotating from the treatment units.

Measurements
To obtain the FPEs, each SCF was superimposed onto the corresponding SPF, aligning all visible vertebral bodies, which were clearly defined on these films. The shift in the field centre was calculated by direct measurement on the film; magnification was determined using a known FFD. Deviations in the anterior L-R and S-I directions were documented. The coronal rotational error (CRE) was determined by measuring the angle between the central axes of the two superimposed films using a protractor. All measurements were performed by a single observer (CK).

Statistical method
Simulator–treatment displacements are referred to as "actual" measurements where a negative value indicates displacement of the field inferiorly, to the left, or anticlockwise for the S–I, L–R and coronal measurements, respectively. The actual displacement measurements all follow a normal distribution. "Absolute" displacements were calculated by removing the negative sign to give an indication of the absolute displacement from the SPF.

Estimates of random error (RE) and systematic error (SE) for all patients were obtained using an analysis of variance (ANOVA) of the actual displacement measurements. The estimates of the variance components, random error variance (²_r) and systematic error variance (²_s), and the range covered by 95% of cases for these components (2 standard deviations, i.e. 2) were calculated according to standard methodologies [6].

Four observations were taken for each patient over 3–4 weeks. To account for the multiple measurements per patient, an ANOVA with repeated measures method was employed to examine differences of actual measurements between subgroups including arm position, sex and age.

	Results

Top Abstract Introduction Methods Results Discussion Conclusion References

 
The characteristics and treatment details of the 24 patients entered into this study are given in Table 1. They were all of WHO performance status 0–2. Most of the patients in the ADG were those with small cell lung cancer who had achieved a complete or partial response to their chemotherapy. These patients therefore received consolidation radiotherapy to the mediastinum and site of the primary tumour via parallel-opposed portals.

View larger version (90K): [in this window] [in a new window]   Figure 1. Summary of mean actual treatment–simulator displacement showing field placement errors from a single simulator check film (). The crossing point of the fine t-bars represents the mean of these differences and their length, two standard deviations from this mean. The thick t-bars indicate the narrow confidenceinterval.

View larger version (89K): [in this window] [in a new window]   Figure 2. Pie charts showing overall maximum field displacement errors in the (a) left–right axis and the (b) superior–inferior axis, both with reference to the clinical threshold of 0.7 cm, and (c) overall maximum field displacement errors in the coronal plane.

Sub-group analysis
The distribution of field placement errors found in the AUG and ADG sub-groups are shown in Figure 3, and the summary statistics for the sub-group analysis are shown in Table 3. Statistically, there was no significant difference in the L–R (p=0.34) or S–I field (p=0.12) displacements when comparing arm position during treatment (Table 3 part c). This may have been owing to the relatively small number of observations. In the AUG, 13.5% of FPEs were greater than 0.7 cm, with 85% of the SCFs with this magnitude of error being in the S–I direction. In the ADG, 16.7% of FPEs exceeded the tolerance value of 0.7 cm, with 56% of the SCFs with this error being in the L–R axis.

View larger version (130K): [in this window] [in a new window]   Figure 3. Distribution of treatment–simulator field placement errors. Arms up group; (a) left–right direction, (b)superior–inferior direction, and (c) rotational errors. Arms down group; (d) left–right direction, (e) superior–inferior direction, and (f) rotational errors. A minus sign indicates a field displacement to the left, inferiorly, or anti-clockwise.

	Discussion

Top Abstract Introduction Methods Results Discussion Conclusion References

 
The accuracy of treatment set-up is clearly dependent on the site being treated, with the thorax being among those sites with greater discrepancies [8, 9]. Quantifying FPEs in the thorax can be technically demanding, even with the use of EPIDs with digitally reconstructed images of the simulator film. For fundamental reasons, the quality of the images produced by EPIDs is poor, as good resolution of the anatomical structures cannot be obtained with these megavoltage devices [11]. A study using an EPID to determine FPEs in thoracic irradiation found that displacement in the field beyond 1 cm ranged from 10% to 46% [12]. It emphasized the limitations of EPIDs in the thorax where there were marked discrepancies in the FPEs recorded dependent on whether the portal boundary or anatomic landmarks were used.

It is for this reason that we elected to use SCFs of diagnostic quality to quantify, with greater accuracy, the magnitude of the FPEs associated with thoracic radiotherapy, and assess their impact on our clinical practice. Accuracy of patient set-up was better in the L–R axis. The mean observed FPE in the S–I direction was 0.43 cm, compared with 0.35 cm in the L–R direction. We found that a margin of 0.93 cm between the clinical target volume (CTV) and PTV would cover 95% of FPEs in the L–R direction, whilst a margin of 1.13 cm was needed for this degree of certainty in the S–I direction. The overall mean absolute FPE of 0.39 cm in the current study compares very favourably with that found by Rabinowitz et al, where the mean field discrepancy was 0.58 cm in thoracic radiotherapy, whilst the worst FPE (defined as the value which 20% of fields exceed) was 1.17 cm [8].

Respiratory movement is obviously an important factor in the reproducibility of set-up in chest radiotherapy. It has been demonstrated using fluoroscopy of the GTV in the thorax during quiet respiration that the mean movement in the cranio–caudal direction is 3.9 mm, with a range between 0 mm and 12 mm [13]. Ekberg et al [13] admit to having ignored the occasional deep breathing when assessing the movements of the lung tumour caused by respiration. Since many patients with lung cancer have compromised respiratory function, respiration can potentially become more laboured at unpredictable times, and inevitably contribute to the REs in treatment set-up.

The current study and others [8, 13], give an indication of the margin of uncertainty needed in chest radiotherapy planning to prevent geographical miss of the CTV. In our planning procedure, the margin between the CTV and PTV is usually 1.5–2 cm in the axial plane, whilst in the S–I axis, a margin of at least 2 cm is used to allow for movement of the CTV with respiration.

The other important issue in patient set-up errors in thoracic radiotherapy is whether there is a need for immobilization devices, in particular when treating patients who are required to have their arms elevated above their heads to facilitate oblique fields.

We found no statistically significant difference in the FPEs in the L–R or S–I direction relative to arm position during treatment. As shown in Figure 3, the actual FPEs mainly follow a normal distribution. However, the inferior field displacements in the AUG were greater than those in the ADG (Figure 3b, e). The most obvious explanation for this observation is that in the AUG, the arms are simply supported above the head in a band. The technique is therefore prone to subjective positioning by both patient and treatment radiographer. Any tendency for the patient to relax the arms downwards would displace any tattoos on the chest wall inferiorly, with the potential for displacing the field inferiorly also. This factor is likely to have contributed to the SEs seen in the current study. Whether or not better immobilization of the arms in this sub-group would improve the FPEs, particularly in the cephalad–caudal direction, is unclear.

Bentel et al [14] looked at the impact of cradle immobilization on set-up reproducibility for lung cancer radiotherapy. They reviewed 726 port films from 60 patients and looked at the frequency of physician-requested isocentre shifts in the immobilized vs non-immobilized patients. Although there was a lower frequency of isocentre shifts in the immobilized patients compared with the non-immobilized group, this difference was not statistically significant, and the differences reached statistical significance only for the oblique fields. However, this study did not quantify the FPEs, and the use of physician-requested isocentre shifts gives no indication of the tolerance used to define unacceptable beam/patient misalignment. We feel that there is still uncertainty as to whether immobilization devices significantly improve patient set-up reproducibility in thoracic radiotherapy. Further studies are needed in this area.

Figures 3c, f shows that the CREs were greater in the ADG. These patients would normally only have an anterior tattoo in the centre of the field and would not routinely get lateral tattoos, as would the AUG. This would allow more rotation to go undetected during treatment set-up in the ADG. This latter observation raises the question of whether all patients receiving radical radiotherapy via parallel opposed portals should also have lateral tattoos, in addition to the routine anterior one, to reduce rotation of the patient and, effectively, the PTV during treatment.

Our results, like those of others, show that simulation-to-treatment variability, i.e. SEs, is a significant source of error in patient set-up during radiotherapy. Rabinowitz et al [8] showed that SE were greater than RE when they looked at the accuracy of radiotherapy to a number of anatomical sites, including lung. They suggested that a possible reason for this finding was a technical difficulty in accurately reproducing patient positioning between simulator and treatment machine.

We accept that our study may not reflect the true SEs, as the FPEs were never actually measured on the treatment unit. However, given the poor assessment of FPEs in the thorax using EPIDs or portal films, the method described in this study was considered to be the best option in mimicking simulator–treatment errors. The patients returning for SCFs were being set up using the same protocols as used on the treatment units, with no help from simulator imaging. Furthermore, the simulator staff included radiographers on rotation from the treatment units.

It is neither feasible nor practical, in a busy radiotherapy department, to routinely use the simulator for checking treatment set-up errors. Ideally, the method described here would be a useful tool for continued audit, but is impractical due to the use of valuable simulator time. It may be that future advances in digital image display will improve the resolution of bone reference points in the thorax to facilitate more accurate measurements of field displacements.

The clinical importance of quantifying FPEs associated with thoracic radiotherapy becomes increasingly relevant as dose escalation studies, using conformal techniques, become more prevalent. In such studies, the aim is to limit normal tissue toxicity whilst escalating the dose to the tumour in the hope to improving local tumour control and, perhaps, survival. Using these techniques, the treated volume is closely conformed to the PTV, and therefore any small shift in patient position or beam alignment will lead to loss of tumour control.

The real answer may lie in more complex treatment planning systems that can take into account patients' movements. One such four-dimensional treatment planning system has been developed and piloted by Shirato et al [15]. Here a linear accelerator synchronized with a fluoroscopic real-time tumour tracking system tracks the three-dimensional co-ordinates of a 2 mm gold marker placed within, or near, the lung tumour. The linear accelerator is then triggered to irradiate the tumour only when the marker is located within a region defined by planned co-ordinates relative to the isocentre.

	Conclusion

Top Abstract Introduction Methods Results Discussion Conclusion References

 
Reproducibility of patient set-up is of paramount importance if the PTV is to receive the prescribed radiotherapy dose. In lung radiotherapy, patient set-up may be prone to greater SEs and unpredictable respiratory movements may lead to significant REs. This study allowed us to quantify the FPEs in lung cancer radiotherapy and to audit the accuracy of our treatment. However, further studies are needed in this technologically challenging area.

Received for publication October 24, 2000. Revision received August 3, 2001. Accepted for publication August 17, 2001.

	References

Top Abstract Introduction Methods Results Discussion Conclusion References

 

1. Bijhold J, Lebesque JV, Hart AM, Vijlbrief RE. Maximizing set-up accuracy using portal images as applied to a conformal boost technique for prostate cancer. Radiother Oncol 1992;24:261–71.[Medline]
2. Huddart RA, Nahum A, Neal A, McLean M, Dearnaley D, Law M, et al. Accuracy of pelvic radiotherapy: prospective analysis of 90 patients in a randomised trial of blocked versus standard radiotherapy. Radiother Oncol 1996;39:19–29.[Medline]
3. Mitine C, Leunens G, Verstraete J, Blanckaert N, Van Dam J, Dutreix A, et al. Is it necessary to repeat quality control procedures for head and neck patients? Radiother Oncol 1991;21:201–10.[Medline]
4. Rosenthal SA, Galvin JM, Goldwein JW, Smith AR, Blitzer PH. Improved methods for determination of variability in patient positioning for radiation therapy using simulation and serial portal film measurements. Radiother Oncol 1992;23:621–5.
5. International Commission of Radiation Units and Measurements. Prescribing, recording and reporting photon beam therapy. Report no. 50. Bethesda, MD: ICRU, 1993.
6. Armitage P. Statistical methods in medical research. Oxford, UK. Blackwell Scientific Publications, 1971.
7. Byhardt RW, Cox JD, Hornburg A, Liermann G. Weekly localisation films and detection of field placement errors. Int J Radiat Oncol Biol Phys 1978;4:881–7.[Medline]
8. Rabinowitz I, Broomberg J, Goitein M, McCarthy K, Leong J. Accuracy of radiation field alignment in clinical practice. Int J Radiat Oncol Biol Phys 1985;11:1857–67.[Medline]
9. Valicenti RK, Michalski JM, Bosch WR, Gerber R, Graham MV, et al. Is weekly port filming adequate for verifying patient position in modern radiation therapy? Int J Radiat Oncol Biol Phys 1994;30:431–8.[Medline]
10. Bijhold J, van Herk M, Vijlbrief R, Lebesque JV. Fast evaluation of patient set-up during radiotherapy by aligning features in portal and simulator images. Phys Med Biol 1991;36:1665–79.[Medline]
11. Boyer AL, Antonuk L, Fenster A, Van Herk M, Meertens H, Munro P, et al. A review of electronic portal imaging devices (EPIDs). Med Phys 1992;19:1–16.[Medline]
12. Lam WC, Partowmah M, Lee DJ, Wharam MD, Lam KS. On-line measurement of field placement errors in external beam radiotherapy. Br J Radiol 1987;60:361–7.[Abstract/Free Full Text]
13. Ekberg L, Wittgren L, Holmberg O. What margins should be added to the clinical target volume in radiotherapy treatment planning of lung cancer? Radiother Oncol 1995;37(Suppl. 1):S19.
14. Bentel GC, Marks LB, Krishnamurthy R. Impact of cradle immobilisation on set-up reproducibility during external beam radiation therapy for lung cancer. Int J Radiat Oncol Biol Phys 1997;38:527–31.[Medline]
15. Shirato H, Shimizu S, Kitamura K, Nishioka T, Kagei K, Hashimoto S, et al. Four-dimensional treatment planning and fluoroscopic real-time tumour tracking radiotherapy for moving tumour. Int J Radiat Oncol Biol Phys 2000;48:435–42.[Medline]

This article has been cited by other articles:

				C M THOMPSON, C S HAMILTON, and J VAARKAMP Thorax set-up verification with multiple oblique treatment portal images Br. J. Radiol., November 1, 2009; 82(983): 950 - 955. [Abstract] [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Essapen, S Articles by Tait, D Search for Related Content PubMed PubMed Citation Articles by Essapen, S Articles by Tait, D