Introduction
The planning target volume (PTV) concept was firstly introduced in the International Commission on Radiation Units and Measurements (ICRU) Report 50 Reference Jones1 . The PTV includes the clinical target volume (CTV) – which is the volume that encompasses the clinical and subclinical disease – plus a margin to account for internal movements and set-up errors Reference Jones1 . According to the ICRU 83 report, PTV margin size should be calculated for each radiotherapy department, this is because set-up procedures, treatment modalities and imaging modalities are some of the department-specific factors that can influence the number of movements and set-up errors that must be accounted for in this margin. This report, however, did not specify the method to be used for margin calculation Reference Gregoire and Mackie2 .
There are also several PTV margin formulas such as Van Herk, Stroom, ICRU 62 formula, Antolak, Bel, McKenzie and Parker, and the selection of the method of calculation could have an influence on the margin size Reference Gill, Reddy, Campbell, Chen and Pearson3 .
A systematic literature review was conducted to identify the methodological approaches used to calculate the PTV margin in the head and neck region with volumetric arc therapy (VMAT) across published literature. To achieve this aim, the objectives of this review were to identify the formulas used to calculate the PTV margin and whether the reviewed studies considered the necessary factors to calculate the margin size with the formula used.
Methods
A systematic literature review was performed to answer the question: What methods are employed by studies to calculate the PTV margin for patients treated with VMAT to the head and neck region?
The Preferred Reporting Item for Systematic Review (PRISMA) checklist was used to guide the write-up of the systematic literature review protocol (PRISMA checklist, 2009). The following demonstrated the Population, Intervention, Comparison, Outcome and Study Design (PICOS) framework which was used to guide the literature search Reference Eldawlatly, Alshehri, Alqahtani, Ahmad, Al-Dammas and Marzouk4 :
Population – Patients receiving radiotherapy in the form of VMAT to the head and neck region.
Intervention – Calculation of the PTV margin.
Comparison – No comparison made
Outcomes – Identification the PTV margin formula and the factors considered.
Study Design – Quantitative studies.
A systematic search on databases in medicine and radiotherapy: MEDLINE, PubMed, CINAHL, ProQuest (Nursing and Allied Health), Scopus, tipsRO and ScienceDirect. The Hydi search engine was also used to find related studies.
Keywords were validated by two experts in the field and content validity was used to assess the validation of the keywords. There was 100% mean agreement and minor suggestions were taken on board.
The Medical Subject Headings (MeSH) thesaurus was used to search for other related words that could be used as keywords. The asterisks (*) next to the keywords identified other terms that are written in different ways and Boolean operators were also used to allow combination of words and phrases to expand the search.
The following combinations of keywords were used to search for related studies:
-
Nasopharyn*/Nasal cavity
-
Oropharyn*
-
Laryn*/Supraglottis/Subglottis/Glottis
-
Hypopharyn*
-
Oral cavity/Mouth/Tongue
-
Sinus*
-
Thyroid
-
Lymphoma
-
Head and Neck
-
Set-up/setup/set up
-
Error/errors
-
VMAT/Volumetric-Modulated Arc Therapy/Volumetric Modulated Arc Therapy/RapidArc Therapy
-
PTV/Planning Target Volume
Publication period (1 January 2007 to 30 December 2020), full-text and human species filters were used to aid in the selection process. Figure 1 is a representation of the search strategy that was used on PubMed.
Search strategy.
Only quantitative studies related to the calculation of the PTV margin for patients being treated with VMAT to the head and neck region were included. The review was also restricted to English language studies.
A dual independent literature search was done by two researchers with over 5 years of clinical experience. The reviewers performed a separate search for the literature using the same research criteria. The search was done between April and December 2020. In the first phase of the review, the literature was screened for the inclusion criteria based only on the title and abstract. For the second phase of the review, a full-text reading of the studies was performed on the eligible studies that were selected in the first phase. Disagreements with regard to data suitability were resolved by consensus between reviewers.
Meta-analysis was not performed due to the heterogeneity in the methodological, statistical and clinical sources; therefore, a narrative synthesis approach was selected Reference Snilstveit, Oliver and Vojtkova5 .
Results
A total of 4341 articles were found. After removing duplicated articles and screening the studies for eligibility, a total of seven relevant studies were found. Figure 2 shows the PRISMA flow chart and Table 1 presents a summary of the study characteristics.
PRISMA 2009 flow chart (Adapted from: Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group (2009). Preferred Reporting Items for Systematic Reviews and Meta-Analyses).
Summary of characteristics of studies included in the narrative synthesis
N = sample number; S = superior; I = inferior; A = anterior; P = posterior; ML = medial-lateral; SI = superior-inferior; AP = snterior-posterior; C3 = cervical spine level 3; C6 = cervical spine level 6; TP = thermoplastic; HR = head rest; HFW = HeadFIX® mouthpiece; SYR = 10 mL/cc syringe barrel; EPID = electronic portal imaging device; CBCT = cone beam computed tomography; PI = portal imaging; eNAL = extended no action level protocol (imaging on first three fractions, followed with once weekly imaging); NAL = no action level protocol (imaging on first three fractions); VHMF = Van Herk’s margin formula.
Methods to calculate the PTV margin
The PTV margin was calculated for all the studies in the review using the Van Herk formula. Both inter- and intra-fraction motion were measured in Bruijnen et al. Reference Bruijnen, Stemkens, Terhaard, Lagendijk, Raaijmakers and Tijssen6 and Yin et al. Reference Yin, Sun and Chi7 studies, whilst the other studies derived the margin by evaluating the inter-fraction errors. None of the reviewed studies assessed and included target delineation variation in the margin formula. Table 2 demonstrates the methods opted by the reviewed studies to calculate the PTV margin.
PTV margin methods
PTV margin size
There was a discrepancy in the PTV margin results of the reviewed studies; however, this was expected since the factors influencing the margin differ from department to department.
The largest discrepancy was found in Anjanappa et al. Reference Anjanappa, Rafi and Bhasi8 and Yin et al. Reference Yin, Sun and Chi7 studies. These studies analysed the margin in the nasopharyngeal region, however, Yin et al. Reference Yin, Sun and Chi7 made use of cone beam computed tomography (CBCT) imaging and assessed inter- and intra-fraction errors, whilst Anjanappa et al. Reference Anjanappa, Rafi and Bhasi8 made use of two orthogonal images and evaluated the inter-fraction errors. Intra-fractional contributes to create a larger PTV margin Reference Stroom, Gilhuijs and Vieira9 ; however, in these two studies, the smallest PTV margin result was in the study that did not evaluate intra-fraction errors. This discrepancy in the margin result could be attributed to the imaging modalities. The chosen modality has an impact on the set-up error that is detected. CBCT should be the modality of choice, since it allows for better observation of the volumes of interest Reference Martins, Couto and Barbosa10 .
Oh et al. Reference Oh, Baek, Kim and Kim11 and Kukolowicz et al. Reference Kukolowicz, Mietelska and Kiprian12 (post-no action level protocol) had the most similar PTV margin result. Both studies, however, varied in immobilisation devices, imaging protocols and outcomes. The similarity of results was most likely by chance. In the reviewed studies, the medial-lateral (ML) margin was not measured in the larynx and oropharynx region. Comparison of margins in the different regions of the head and neck was also not possible, since the studies did not provide data regarding the different areas.
Inter-fraction errors
The reviewed studies obtained similar results for population systematic errors with the standard deviation (SD) of the systematic errors of the reviewed studies measured to be 0·4 mm, 0·5 mm and 0·5 mm in the ML, superior-inferior (SI) and anterior-posterior (AP) direction, respectively. The SD for population random errors resulted to be slightly higher than that of the population systematic errors, with each direction (ML, SI and AP) obtaining a value of 0·6 mm.
Deb et al. Reference Deb, Chaudhuri and Panda13 study obtained the highest population random error. This study treated patients without daily imaging, instead a total of 10 CBCT images were acquired for each patient. Population systematic and random errors can be corrected prior treatment with daily imaging; however, in studies where daily imaging are not performed, random error can not be compensated. For this reason, the PTV margin of Deb et al. Reference Deb, Chaudhuri and Panda13 study resulted to be measured larger when compared with other studies in the review.
Rotational errors were analysed in four studies. Oh et al. Reference Oh, Baek, Kim and Kim11 study compared rotational errors in different anatomical regions. The rotational error for head and neck region was below 3o and this value was small when compared with other anatomical regions. Norfadilah et al. Reference Norfadilah, Ahmad, Heng, Lam, Radzi and John14 study also calculated rotational errors with the aim of comparing two tongue immobilisation devices. Average rotational errors result for headFIX® mouthpiece and syringe mouthpiece were 0·00°±0·65° and 0·34°±0·59°, respectively. In Kukolowicz et al. Reference Kukolowicz, Mietelska and Kiprian12 rotations larger than one degree were seldom observed; therefore, these errors were not taken into consideration. This study, however, performed 2D imaging, therefore, rotational results from multiple perspective were not analysed. Small values of rotations were also observed in Yin et al. Reference Yin, Sun and Chi7 study, with the number of fractions rarely exceeding 2° for pitch, roll and yaw directions.
Intra-fraction translational errors
Bruijnen et al. Reference Bruijnen, Stemkens, Terhaard, Lagendijk, Raaijmakers and Tijssen6 measured intra-fraction errors from 2D cine magnetic resonance imaging (MRI) and deformable image registration. In this study, respiratory tumour motion, swallowing, tongue motion and set-up errors were investigated to determine the PTV margin size. When the tumour motion was incorporated into the PTV margin formula, the margin expanded by 0·6 mm for oropharyngeal tumours, 0·2 mm for nasopharyngeal tumours and 1·7 mm for laryngeal tumours Reference Bruijnen, Stemkens, Terhaard, Lagendijk, Raaijmakers and Tijssen6 .
In Yin et al. study Reference Yin, Sun and Chi7 , the intra-fraction population systematic error during the 5–9 min VMAT period ranged from 0·2 mm to 0·4 mm, and the population random error ranged from 0·5 mm to 0·6 mm.
Discussion
PTV margin equations
In radiotherapy, there is an issue on the method selected to determine the PTV margin. The Van Herk formula is a widely used strategy for PTV margin calculation, and this equation was used in all the reviewed seven studies. The reason for selecting the formula was not specified and lack of comparison of this formula against other options for head and neck was identified as a major gap in the literature.
A study by Namysł-Kaletka, Tukiendorf and Wydmański Reference Namysł-Kaletka, Tukiendorf and Wydmański15 used three formulas; Van Herk, Stroom and ICRU, to assess PTV margin results based on set-up errors for gastric cancer patients. The margin results were compared, and the study revealed that the formula being used has an impact in the PTV margin. As such, the formula should be selected carefully.
The Van Herk’s formula assumes that the CTV is spherical in shape, the tissue is homogeneous, conformal beam penumbra and that the number of fractions is infinite Reference van Herk, Remeijer, Rasch and Lebesque16 . As such, the use of this formula for the PTV margin calculation should be used with caution when these assumptions do not apply.
Due to the exclusion of rotational errors and shape variation, the Van Herk’s formula should be considered a lower limit for the delivery of safe radiotherapy. The formula guarantees that 90 per cent of patients receive a minimum of 95 per cent of the recommended dose in the CTV Reference van Herk, Remeijer, Rasch and Lebesque16 . Therefore, this formula seems to be adequate for the calculation of PTV margin in head and neck patients.
PTV margin size
The studies in the review had different PTV margin size and this continues to necessitate the importance of the departments to calculate their own specific margins. Margin sizes seem to be affected by the imaging protocols and immobilisation devices.
Yin et al. Reference Yin, Sun and Chi7 study demonstrated the effect of daily imaging on the margin size for nasopharyngeal patients. The resulting margin size was small when compared to other studies in the review, since the PTV margin was calculated on the set-up errors obtained after CBCT correction. Kukolowicz et al. Reference Kukolowicz, Mietelska and Kiprian12 stated that daily online correction was slightly better than the NAL (no action level) protocol for patients having treatment to the head and neck region. However, the study failed to analyse rotational errors and could not assess anatomical changes since these were not visible on portal images.
Daily imaging protocols should be adhered for patients treated with a tight PTV margin, such as intensity-modulated radiotherapy or VMAT. Daily imaging aids in verifying the set-up position, identifying the location of the target, assessing tumour shrinkage and making the necessary corrections prior to each exposure Reference Oh, Baek, Kim and Kim11 .
Norfadilah et al. Reference Norfadilah, Ahmad, Heng, Lam, Radzi and John14 study assessed the impact of variation in tongue immobilisation on the margin size, and the results were indicative that immobilisation devices influence the PTV margin size.
According to Anjanappa et al. Reference Anjanappa, Rafi and Bhasi8 study, the lower neck region requires a larger PTV margin in the ML and AP direction. Another similar result was obtained in Cheo et al. Reference Cheo, Loh, Chen, Lee and Tham17 study, where the set-up errors were evaluated in different levels of the neck and the largest displacement was found to be in the ML direction (6·52 mm). As compared to the SI direction, the PTV margin findings in the other reviewed studies do not appear to indicate any substantial difference in the margin size of the ML and AP direction. Anjanappa et al. Reference Anjanappa, Rafi and Bhasi8 suggestion of increasing the margin size in the ML and AP direction should therefore not be considered for all clinical situations.
Inter-fraction and intra-fraction errors
All the studies in the review analysed inter-fraction errors; therefore, the method of calculating the margin varied solely on whether intra-fraction error was being assessed. Van Herk et al. suggested to include target volume delineation variation and intra-fractional errors in the margin estimation, as well as including the SD of these errors in quadrature Reference Van Herk18 .
The systematic literature found that inter-fraction errors were generally higher than intra-fraction errors, which indicates that maintaining the position during treatment leads to less errors than reproducing the set-up between treatments.
Population-based margins should account for internal motion, therefore intra-fraction movements should be quantified Reference Bruijnen, Stemkens, Terhaard, Lagendijk, Raaijmakers and Tijssen6 . Bruijnen et al. Reference Bruijnen, Stemkens, Terhaard, Lagendijk, Raaijmakers and Tijssen6 and Yin et al. Reference Yin, Sun and Chi7 were the only studies that analysed both intra-fraction errors and inter-fraction errors. Intra-fraction error is related to internal organ motion and patient movement during treatment, therefore it is a random error Reference Shikha and Kataria19 .
In Bruijnen et al. Reference Bruijnen, Stemkens, Terhaard, Lagendijk, Raaijmakers and Tijssen6 study, when the tumour motion was incorporated into the PTV margin formula, the margin expanded mostly in the laryngeal region. This indicates that this region is subjected to greater tumour motion due to swallowing Reference Bruijnen, Stemkens, Terhaard, Lagendijk, Raaijmakers and Tijssen6 . These results were similar to a study by Gurney-Champion et al. Reference Gurney-Champion, McQuaid and Dunlop20 where the tumour motion analysed with MRI imaging was found to be significantly larger in the larynx and hypopharynx when compared to oropharynx.
In Yin et al. Reference Yin, Sun and Chi7 study, intra-fraction errors were assessed via CBCT images. The study focused on patients receiving treatment to the nasopharynx and intra-fraction errors were based on patient movement during treatment.
Few studies obtained intra-fraction results on patients treated with VMAT and the need to calculate this error was raised by Yin et al. Reference Yin, Sun and Chi7 . There was no significant correlation between treatment delivery time and intra-fraction errors. This study had some limitations since there was a limited data for analysis and statistically significant results could not be obtained due to the narrow range of treatment time. These results are contradictive to Hoogeman et al. Reference Hoogeman, Nuyttens, Levendag and Heijmen21 , who stated that with an increment in time, intra-fraction systematic errors increase.
Van Herk’s formula does not consider rotational errors Reference Van Herk18 . Four studies from the review still investigated the rotational set-up errors on their population sample. In high-precision treatments, it is important that rotational errors are not ignored, especially when the distance from the isocentre to the target is large or when the tumour has a non-spherical shape Reference Chang22 . In most clinical departments, these errors are not corrected due to couch limitations Reference Zhang, Song, Chan, Burman and Yamada23 . Rotational errors were minimal in the four studies.
Target volume delineation
Calculation of the PTV margin should incorporate the systematic errors obtained from target volume delineation Reference Yin, Sun and Chi7,Reference Anjanappa, Rafi and Bhasi8 . Even though target volume delineation might have the largest impact in margin size since it is fixed throughout the treatment Reference Vinod, Jameson, Min and Holloway24 , none of the studies in the review evaluated this factor. This type of systematic error is commonly ignored in studies that investigate PTV margin calculation Reference Vinod, Jameson, Min and Holloway24 leading to a potential underdosage to the tumour. This is important for all head and neck locations, but more so for the oropharyngeal cancer since interobserver variability in target delineation is greater in this region Reference Segedin and Petric25 .
Limitation
The researchers ran a thorough search and all the body of evidence found was analysed and discussed; however, the availability of data on PTV margin calculation to the head and neck region was limited and this resulted in a small sample of studies. This could have limited the findings of significant relationships.
Also, some studies were not reliable since they obtained a weak quality evaluation after been analysed by the Joanna Briggs Institute tool, as such, the researchers were careful in drawing conclusions from these publications.
Since the review relied on pre-existing data, the obtained results were dependent on the methodology of the studies. Self-reported data bias could be introduced from relying on pre-existing data. The studies were also limited to the English language; therefore, 46 non-English language studies were excluded in the first phase of the review, this resulted in the review to be susceptible to reporting bias and language bias Reference Kirkham, Dwan, Altman, Gamble, Dodd, Smyth and Williamson26 .
Comparison of study results was limited due to the studies being heterogenous in terms of the key characteristics of the studies and methodology design.
Although the findings of this study should be interpreted with caution, this review represents a comprehensive examination of studies that analysed PTV margin to the head and neck region with VMAT.
Conclusion
The Van Herk formula was used in all the studies in the review and none of the studies made use of other PTV margin formula. This margin seems adequate to calculate the PTV margin for head and neck patients treated with VMAT.
Most studies only included inter-fraction errors into the van Herk formula. However, PTV margin should incorporate target volume delineation, intra-fraction errors and inter-fraction set-up errors to ensure a well-defined margin.
The result indicated that tumours in the laryngeal region where more susceptible to motion when compared to those found in the nasopharynx and oropharynx.
Inter-fraction translational errors were assessed in all the studies from set-up errors that were registered by the imaging software. The SD of population random errors was found to be a little bit higher than that of population systematic errors. The systematic and random errors of set-up rotational errors were not considered in most studies, and the obtained values were not included in the PTV formula since this formula assumes that the CTV is spherical and, therefore, is unaffected by rotation.
The findings of the review where in line with other studies that stated that different anatomical regions, immobilisation devices, imaging frequency, treatment modality, set-up procedures and patient collaboration influence the size of the PTV margin.
Supplementary Material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1460396921000546.
Acknowledgements
None.
Financial Support
None.
Conflict of Interest
The authors declare none.
