Introduction
Radical (chemo) radiotherapy treatment is the standard of care for certain early-stage head and neck cancers, allowing for organ preservation. Radical radiotherapy fields often extend into the lower neck, the territory of the brachial plexus (BP). The widely accepted radiation tolerance to the BP as a whole organ is 60 Gray (Gy) in 2 Gy/fraction (#), and radical doses in head and neck cancers exceed this with doses ranging from 60 to 70 Gy.Reference Emami, Lyman and Brown1 The incidence of radiation-induced brachial plexopathy (RIBP) is 1–2% in patients receiving doses less than 55 Gy; the incidence rises steeply when approaching tolerance.Reference Arya, Jinzhong and Williamson2, Reference Delanian, Lefaix and Pradat3 The morbidity associated with this late toxicity will be of increasing significance in the growing cohort of human papillomavirus patients who tend to present with malignant disease younger and have longer life expectancies.Reference Ang, Harris and Wheeler4 In the era of intensity-modulated radiotherapy (IMRT)-based treatment, the BP is becoming an increasingly important organ at risk (OAR) to outline and there are now detailed atlases available to aid with this.Reference Hall, Guiou and Lee5 Immobilisation in a 5- or 9-point thermoplastic shell is standard practice; however, studies have shown despite this, large shifts routinely occur during treatment, with lower neck structures experiencing more setup variability.Reference Neubauer, Dong and Followill6–Reference Ove, Cavalieri and Noble8 However, the impact movement has on the BP during a course of radical radiotherapy is not fully understood, with questions remaining on the degree of movement it experiences and whether our departmental standard planning at risk volume (PRV) margin of 5 mm is adequate to account for this. It is also unclear whether certain anatomical segments of the BP are more susceptible to movement than others. The clinical impact of movement could potentially result in tumour being under dosed and the OARs, such as the BP, being overdosed.
Injuries to the BP can affect the cutaneous sensation and movement of the arm. RIBP has similar clinical manifestations but is thought to be mediated by a combination of fibrosis surrounding the nerve, tissue ischaemia and effects of demyelination of axons.Reference Amira, Dirk and Fabian9 It is defined as ‘neurological impairment of a transient or permanent nature as a sequela to radiation therapy’.Reference Schierle and Winograd10 Clinically, the patient can experience pain, sensory and motor deficits, atrophy of affected muscle groups, oedema and complete paresis.Reference Gillette, Mahler and Powers11 There are two forms of RIBP described in the literature relating to head and neck cancers. Early transient brachial plexopathy is reported as occurring between 3 and 10 months after radiation treatment for locally advanced head and neck cancers.Reference Schierle and Winogard12 Late occurring RIBP has been reported in a cohort of 31 nasopharyngeal patients, developing approximately 4 years after radiation treatment. Many patients had lower cervical chain lymph node metastasis receiving a mean dose of 66 Gy, which contributed to the incidence of RIBP in this group.Reference Cai, Li and Hu13 The exact pathogenesis is not fully understood, with some reports attributing it to an autoimmune response and others to compression from post-irradiation oedema.Reference Schierle and Winogard14 The majority of evidence for RIBP exists from studies carried out in locally advanced breast cancer patients who have had post-operative chest wall and supraclavicular fossa irradiation. These studies have found rates of BP damage as low as 1·7% and as high as 73%, while using hypofractionated schedules using 2·2–4·58 Gy/#, to a total dose of between 43·5 and 60 Gy. They found that when the biological equivalent dose exceeded 55 Gy (α/β=2 Gy) the incidence of RIBP rose steeply from 1·7 to 15%.Reference Galecki, Hicer-Grzenkowicz and Grudzien-Kowalska15 Very high rates of RIBP were thought to be due to errors in radiotherapy technique, exposing the BP to higher doses than expected. There is clear evidence of the role of other contributory factors such as chemotherapy (neoadjuvant or concurrent) previous neck dissection and low cervical lymph node metastases, all increasing the incidence of RIBP.Reference Pierce, Recht and Lingos16–Reference Khan, Siddiqui and Gupta18
Accurate and reproducible immobilisation is needed to effectively deliver radiotherapy to tumour and prevent OARs from receiving doses of radiation that will cause toxicity.Reference Gilbeau, Octave-Prignot, Loncol, Renard, Scalliet and Gregoire19 Many studies have shown that by reducing positioning variation with the use of effective immobilisation, this can improve radiotherapy outcomes.Reference Kneebone, Gebski, Hogebdoom and Tumer20, Reference Rosenthal, Roche and Goldsmith21 The use of three-dimensional (3D) imaging verification with cone beam computer tomography (CBCT) integrated into the linear accelerator has been shown to be superior to two-dimensional (2D) imaging techniques.Reference Li, Zhu and Zhang22 The use of CBCTs can generate data on systematic and random errors of each treatment. As IMRT delivers highly conformal doses of radiotherapy to the target with steep dose gradients at the target boundaries, accurate immobilisation is needed to prevent setup errors. A geographical miss could result in under dosing tumour and/or overdosing OARs. Although the method of immobilisation and its impact are not under investigation in this study, we noted the variations that occured during a standard course of treatment.
This retrospective analysis aims to quantify the inter-fractional movement of the BP during radical radiotherapy and secondly determine whether there is a significant difference in movement in the superior and inferior branches of the BP. Lastly, the radiation dose received by the BP will be recorded.
Methods
A retrospective single-centre analysis was performed from January 2016 to January 2017, of 15 sequentially selected patients who were diagnosed with head and neck cancers, Tumour, Nodes and Metastasis (TNM) stages 3/4, requiring bilateral neck irradiation to a minimum of cervical lymph node levels 2, 3 and 4 as part of their radiotherapy treatment plan. Patients who had had unilateral neck irradiation, nodal irradiation not involving levels 2–4 and patients with metastatic disease were excluded. Using automated bony matching of the images, the BP was contoured on the planning scan and each weekly CBCT, the Jaccard conformity index (JCI) was then calculated. Retrospective data collection and analysis were approved by the institution audit and research department.
Patients
Patients were selected from the local head and neck cancer database at our institution, all had had treatment for cancer of the oral cavity, oropharynx, nasopharynx or larynx. Patients were treated with 2 dose levels; 65–60 Gy (2 16–2 Gy/#) in 30# to the primary tumour plus involved nodes and a second dose of 54 Gy in 30# was used to prophylactically treat uninvolved nodal levels, with or without chemotherapy. All patients were computed tomography (CT) simulated, immobilised in 9-point thermoplastic shells and had weekly CBCT during the 6-week course of treatment.
CBCT
The CBCT system consisted of a gantry-mounted X-ray tube set at 90˚ to the mega-voltage (MV) beam, and a flat panel X-ray detector at 270˚ to the MV beam. For all measurements in this work, CBCTs were acquired on the Varian Truebeam using the thorax 2-mm setting, using a full arc between the angles of 179˚ to 181˚. The BP was contoured on the planning scan and on each of the weekly CBCTs, allowing for comparison between them. CBCTs were not performed on the same day as weekly cisplatin administration to minimise the effects of oedema caused by pre- and post-chemotherapy fluid hydration.
BP delineation
BP delineation was performed using the Radiation Therapy Oncology Group (RTOG)-endorsed BP contouring atlas, which provides an accurate, reproducible and standardised approach to contouring the BP.Reference Hall, Guiou and Lee5 The BP is a difficult structure to visualise on CT imaging, the RTOG guide provides guidance on anatomical landmarks that can be used as surrogates for the route it takes. A 5-mm rollerball was used to delineate the BP on each axial slice of the planning CT scan, extending from the C4/C5 to T1/T2 interspace bilaterally and identified as structure set ‘BP Right’ and ‘BP Left’. The route of the BP was verified by a Consultant Radiologist. Once the BP had confidently been identified, the structure was divided into superior and inferior divisions; superior division extending from C4/C5 to C6/C7 interspace and inferior division extending from C6/C7 to T1/T2 interspace. The C6/C7 interspace was decided upon as the anatomical division point of the BP as this allowed for 2 vertebrae and 2·5 ventral rami to be encompassed in the superior and inferior divisions equally. These volumes were identified as structure set ‘BP right superior’, ‘BP right inferior’, ‘BP left superior’ and ‘BP left inferior’. This process was repeated on each of the six, weekly CBCTs for each patient.
The Jaccard Conformity index
CBCTs were overlaid onto the planning scan and bony matched using Varian’s automated matching software, allowing for the volume of intersection and hence JCI to be calculated (Figure 1). JCI is calculated using
$${\rm JCI} = {\rm A} \cap {\rm B}/{\rm A} \cup {\rm B},$$
that is, the intersection volume of A and B divided by the union volume of A and B. A JCI=1, equates to perfectly overlapping volumes, JCI=0·75 is 75% overlap, JCI=0·5 is 50% overlap and CI=0·25 is 25% overlap. The acceptable JCI level is unclear; however, literature suggests a poor correlation represented by a JCI<0·5, whereas others suggest an optimal level of ≥0·7.Reference Jena, Kirkby and Burton23–Reference Feuvret, Noel, Mazeron and Bay26 The JCI was calculated for BP right, BP left, BP right superior, BP right inferior, BP left superior and BP left inferior.
CBCT bony matched to planning scan. Planning scan BP volume (red), week 1 CBCT BP volume (yellow).
Statistical analysis
Descriptive statistics were calculated for patient demographics, tumour characteristics, and dose volume statistics were obtained from the radiation plans. The mean, standard deviation (SD) and 95% CI were calculated using Prism GraphPad (GraphPad Software, San Diego, California, USA, version 7). Two sample t-tests were used to assess for variation in JCI between the superior and inferior divisions of the BP. A p-value of ≤0·05 was considered significant.
Results and Discussion
The mean age of the study population was 59·5 years (range 40–79); 86·7%Reference Cai, Li and Hu13 of patients had stage 4 disease. Patient and tumour characteristics are described in Table 1.
Patient and tumour characteristics
The mean (±SD) BP volume on the left was 9·5±1·91 cm3 and right 9·4±1·81cm3. Analysis of the superior and inferior portions of the BP was performed; the mean (±SD) superior BP volume on the left was 3·29±0·67cm3 and on the right 3·27±0·82cm3. The mean (±SD) inferior BP volume on the left was 5·47±1·33cm3 and on the right was 5·45±1·59 cm3.
The mean (±SD) dose to the whole BP on the left was 49·4±5·34 Gy and on the right was 49·8±5·39 Gy.
The maximum dose (±SD) to the whole BP on the left was 58·8±4·72 Gy and on the right was 60·2±5·26 Gy.
The mean (±SD) dose to the superior BP on the left was 51·6±4·46 Gy and on the right was 52·4±5·03 Gy. The mean (±SD) dose to the inferior BP on the left was 47·2±5·22 Gy and on the right was 47·3±4·46 Gy. The maximum dose (±SD) to the superior BP on the left was 59·8±4·96 Gy and on the right was 60·9±5·4 Gy. The maximum dose (±SD) to the inferior BP on the left was 57·7±4·27 Gy and on the right was 59·5±5·06 Gy (Table 2).
Mean (±SD) and maximum (±SD) dose to right (R) and left (L), superior and inferior BP
The mean (±SD) JCI for right and left superior BP was 44·4±15·5%, whereas the mean (±SD) JCI for right and left inferior BP was 38·3±15·5% (Figure 2). There was a statistically significant difference between superior and inferior JCI, p=0·0002, 95% CI (−9·26 to −2·88). Bilateral superior BP JCI was higher, with better conformity than the corresponding inferior divisions. All patients experienced weight loss, on average 5 kg was recorded.
Mean±SD JCI for superior (right and left) and inferior (right and left) BP.
Discussion
Treatment of head and neck cancers with IMRT allows delivery of high-dose radiation with increased conformality. However, as doses are escalated and margins reduced, accuracy in delivering treatment becomes more important. Anatomical changes as a result of weight loss or tumour shrinkage could result in less accurate immobilisation, which risks under dosing the tumour and overdosing OARs. This could detrimentally impact on tumour control rates and increase the incidence of long-term irreversible toxicity. BP movement varies during a course of radical radiotherapy; this study suggests that the variation is more significant in the inferior division of the BP, which could provide evidence to increase PRV margins.
The RTOG-endorsed BP contouring guide provides a consistent method for identifying structures used as surrogates for the route of the BP and has helped to standardise BP contouring. The BP is a very difficult structure to visualise on CT imaging; the guide uses anatomical landmarks as surrogates for the route it takes through the neck and upper chest. Contouring the BP has allowed the ‘tracking’ of its movement in relation to the initial planning scan and subsequently throughout a 6-week course of treatment. As the BP lies adjacent to nodal levels II–IV, contouring of nodal groups followed by clinical target volume and planning target volume (PTV) expansions, often results in the BP coming within the PTV volume. Dose volume histogram data have shown that the maximum dose the BP received as a whole organ was 59·5±5·05 Gy. Divisions into right and left BP have shown maximum doses to be 60·2±5·26 Gy and 58·8±4·72 Gy, respectively. The mean volume receiving 60 Gy was 10·3%, well below the tolerance level set of 60 Gy to 100% of the organ.Reference Emami, Lyman and Brown1 There was a wide variation in the dose the BP received by each individual patient which could be due to anatomical variations between patients as well as differences in treatment plans. But what can be seen is all doses were within tolerance of the BP.
JCI was used to ascertain the level of overlap between the BP on the planning scan and the BP on each of the weekly CBCTs. This technique has its own limitations as it can provide a slight overestimation of the degree of movement. However, despite this, it has provided a clear indication of the effects of inter-fractional movement on the BP during a 6-week course of treatment. Wide variations in the JCI for some patients can be seen, which may be due to certain features in these patients, such as increased weight loss or tumour shrinkage, setup problems associated with mask tolerance or facial oedema to mention a few. All patients had a low JCI at week 1, indicating a significant amount of movement at the start of treatment, which may be attributed to poor setup in the thermoplastic shell. JCI clearly shows bilateral inferior portions of the BP experiencing a statistically significant lower conformity, indicating increased movement of the BP in this region. It is felt that increased movement inferiorly is due to more pronounced effects of weight loss affecting the shoulders and upper chest as opposed to the head and neck regions, rendering the thermoplastic immobilisation less effective. In this cohort, all patients experienced weight loss, on average 5 kg was experienced which is in keeping with literature.Reference Larsson, Hedelin, Johansson and Athlin27 Small patient numbers are the main limiting factor to this study, further patient numbers with a wider range of primary tumour sites would be needed to explore the impact of this work further.
Conclusion
Currently departmental protocol dictates that a 5-mm PRV margin is placed on the BP; in view of the data presented here, the inferior division of the BP experiences significant inter-fractional movement and a larger PRV should be considered, while taking into account the detrimental effect this will have on maintaining BP tolerances. The superior BP has a higher JCI than the inferior BP, but surprisingly the majority of BPs show a JCI of <50% conformity regardless of superior or inferior division, thus highlighting the importance of accurate and reproducible immobilisation techniques. The suggested movement of the BP in this study highlights the uncertain risk the BP is facing during a course of treatment; however, dosimetric data have shown radiotherapy doses using a PRV margin of 5 mm to be well clear of tolerance thresholds. But, this may be falsely reassuring and must be viewed in the context of the patient cohort used, in which the majority of patients had high-dose treatment to the upper rather than the lower cervical lymph node chain. Therefore, our recommendation would be to consider generous margins for the whole of the BP but in particular to the inferior division. This will inevitably have a detrimental impact on BP tolerances achieved but may give a more accurate assessment of dose the BP is exposed to. In the era of new radiotherapy techniques such as protons, which have steep dose gradients and smaller margins, accuracy in OAR delineation will be of increasing importance.
Author ORCIDs
Asma Sarwar, 0000-0002-4739-3986
Conflicts of interest and source of funding
None declared.
