Breathing guidance in radiation oncology and radiology: A systematic review of patient and healthy volunteer studies.

PURPOSE
The advent of image-guided radiation therapy has led to dramatic improvements in the accuracy of treatment delivery in radiotherapy. Such advancements have highlighted the deleterious impact tumor motion can have on both image quality and radiation treatment delivery. One approach to reducing tumor motion irregularities is the use of breathing guidance systems during imaging and treatment. These systems aim to facilitate regular respiratory motion which in turn improves image quality and radiation treatment accuracy. A review of such research has yet to be performed; it was therefore their aim to perform a systematic review of breathing guidance interventions within the fields of radiation oncology and radiology.


METHODS
From August 1-14, 2014, the following online databases were searched: Medline, Embase, PubMed, and Web of Science. Results of these searches were filtered in accordance to a set of eligibility criteria. The search, filtration, and analysis of articles were conducted in accordance with preferred reporting items for systematic reviews and meta-analyses. Reference lists of included articles, and repeat authors of included articles, were hand-searched.


RESULTS
The systematic search yielded a total of 480 articles, which were filtered down to 27 relevant articles in accordance to the eligibility criteria. These 27 articles detailed the intervention of breathing guidance strategies in controlled studies assessing its impact on such outcomes as breathing regularity, image quality, target coverage, and treatment margins, recruiting either healthy adult volunteers or patients with thoracic or abdominal lesions. In 21/27 studies, significant (p < 0.05) improvements from the use of breathing guidance were observed.


CONCLUSIONS
There is a trend toward the number of breathing guidance studies increasing with time, indicating a growing clinical interest. The results found here indicate that further clinical studies are warranted that quantify the clinical impact of breathing guidance, along with the health technology assessment to determine the advantages and disadvantages of breathing guidance.


I. INTRODUCTION
The advent of image-guided radiation therapy (IGRT) has led to dramatic improvements in the accuracy of treatment delivery in radiotherapy, with the reduction of both random and systematic uncertainties. [1][2][3][4][5][6] While IGRT has improved the accuracy of radiotherapy by utilizing information about tumor motion and positioning throughout a patient's treatment, it has also shed light on the 35 deleterious impact tumor motion can have on both image quality and radiation treatment delivery. 2, 4, 7-10 Anatomic motion due to breathing in the thoracic and abdominal regions is of great concern due to their proximity to the thoracic diaphragm, where respiratory-induced motion can be up to 5 cm. 11 In addition, heightened patient anxiety levels during imaging and treatment, 12,13 can result in increasingly irregular breathing, leading to erratic breathing motion of both internal anatomy and 40 the tumor itself. 8, 14,15 The widespread utilization of IGRT has led to the investigation of an increasing number of methods to address breathing motion and therefore tumor and organ movement and the resultant uncertainties they cause. A number of image reconstruction methods and tracking systems have been developed to ameliorate these uncertainties. [16][17][18][19] However such techniques can be expensive 45 and don't directly manage the problem of irregular breathing motion. Addressing irregular tumor motion directly at the source by managing the patients' breathing has been of increasing interest in recent times, with several breathing guidance techniques being developed from simple buzzer signals to interactive guiding interfaces to facilitate regular and predictable tumor motion.

50
The deleterious impact of irregular motion during image acquisition has been well documented for across a range of medical imaging modalities. 8,14,[20][21][22][23][24][25][26][27][28] During radiation treatment there are two fundamental types of errors: the errors occurring during treatment preparation (systematic) and the errors occurring during treatment delivery (random); 5, 29-31 both these types of errors are exacerbated by irregular breathing-motion. 9, 10, 27 55 Systematic errors typically arise from errors in the images used to plan the patient's treatment; Figure 1 demonstrates the irregular tumor motion and errors present in images due to such irregular breathing-motion.  To account for irregular breathing-motions' exacerbation of systematic and random errors, the treatment volume is expanded; 32 increasing radiation dose to the healthy surrounding tissue thus 70 increasing the risk of post-treatment radiation complications such as radiation pneumonitis. 33-39 Such complications occur in over 60% of lung cancer patients after treatment, with 47% developing at least grade 2 pneumonitis requiring clinical intervention. 34 Such clinical interventions involve the prescription of anti-inflammatory pharmaceuticals thereby increasing health-care costs for that patient's course of treatment. 36,40 To combat the increase of these systematic and random errors a 75 number of strategies directly engaging with the patient have been investigated to minimize the irregularity of patient breathing motion. These breathing guidance strategies have the advantage of being non-invasive, requiring minimal modifications to existing facilities and protocols. Given the relatively recent widespread interest in such breathing guidance strategies, a review of such research has yet to be performed. It was therefore our aim to perform the first systematic 80 review of breathing guidance intervention strategies within the fields of radiation oncology and radiology.

II. METHODS
This systematic review follows the PRISMA-Statement reporting standard (Preferred Reporting Items  85 for Systematic reviews and Meta-Analyses). 41 Table 1 presents our research questions in the PICOS approach (Patients, Intervention, Comparison, Outcome, Study design); given the relatively recent interest in such breathing guidance strategies, healthy volunteer studies were also considered in addition to patient studies.

I -intervention
Breathing guidance -technologies which monitor patient breathing and provide feedback to the patient informing them on how to adjust their own breathing in real-time on their own accord.

C -comparison
No breathing guidance of the same breathing type (i.e. non-guided breath-holds for breath hold studies, and free breathing for breathing guidance studies)

O -outcome
Regularity of breathing signal & anatomic/tumor motion, medical image quality, radiation treatment margins & coverage, medical imaging &radiation treatment times S -study design Quantitative and controlled prospective or retrospective trials. Once eligible articles were identified they were filtered in accordance to the selection criteria. The 90 objective of the selection criteria was to acquire scientific articles describing in sufficient detail a breathing guide intervention's utilisation towards some aspect of abdominal or thoracic radiology and radiotherapy application. Articles were extracted by two authors using an electronic (Microsoft Excel 2010) pro forma specifying the identified articles. Where there was disagreement between the reviewers, discussion was undertaken amongst all authors until consensus was reached. 95

II.A. Selection Criteria
Articles were included if they satisfied the following inclusion criteria: 1) Quantitatively evaluate the intervention of breathing guidance relevant to the practice of either medical imaging or thoracic/abdominal radiotherapy (prospective or retrospective) 2) Participants were human over the age of 18 (retrospective data was from adult human 100 study) 3) Reported in the English language 4) Published in a peer-reviewed journal between the years 1994 -2014 5) Had a control group for the same breathing type:  For guided breathing studies control group performed unguided free breathing 105  For guided breath hold studies control group performed unguided breath-holds Articles which excluded, even if satisfying the above inclusion criteria, if they: 1) Did not have a control group comparing intervention to no intervention for the same breathing type (free breathing or breath hold) 110 2) Lacked a statement of statistical significance 3) Did not describe, or reference to an article, in sufficient detail of the breathing guidance intervention 4) Was not a scientific paper (e.g. conference abstract, conference proceeding, book, patent) From August 1 -14, 2014 the following online databases were searched: Medline, Embase, PubMed, and Web of Science. The search for articles initially included the fields of radiation oncology and radiology using the terms: (radiation therapy OR radiotherapy OR imaging). These search results were then refined towards breathing guidance by using the terms: (respiration OR breathing) AND (audio OR visual) AND (guidance OR training OR feedback OR biofeedback). 120 The findings from the above mentioned databases, in addition to articles identified through hand searching of their reference lists and cross-referencing for previously unidentified articles which met the inclusion criteria. These articles were exported to a citation manager, Endnote X5 where duplicate articles were also removed. The process tree for attaining the search strategies results in shown in Figure 3. After duplication and filtering through the selection criteria five articles identified 125 by this hand searching method made it into the final 27 articles.

II.C. Analysis of Articles
Due to the diverse applications and results used to determine the efficacy of breathing guidance strategies a meta-analysis was not performed; however the main findings from each of these articles were organised in terms of statistical significance: achieving positive significant results, nonsignificant results, or negative results. 140 Quality assessment scoring of the identified and included articles was also performed in accordance with the Standard Quality Assessment Criteria for Evaluating Primary Research Papers From a Variety of Fields. 42 Quality Assessment Score is given based on 14 questions about the article, the reviewers award yes (2 points), partial (1 point) and no (0 points) or not applicable (N/A -question not counted in score). Overall a score out of 28 (or less if N/A is chosen) is found and then converted 145 to a percentage. Articles were scored by two authors and when discrepancies arose in the scores allocated a discussion was then undertaken until a consensus was reached.

III. RESULTS
Twenty-seven articles were included as a part of this systematic review as shown in Figure 3. After 150 duplication and filtering through the selection criteria four articles identified by this hand searching method made it into the final 27 articles. Table 2 and Table 3 detail the development of such strategies over the past 20 years, in addition to the quality assessment score of each article. The average quality assessment score was 79% (range: 54% -95%). Figure 4 also illustrates the timeline of these studies. 155      Unguided: mean D mean, CTV was 50.1 Gy  Guided breath holds: mean D mean, CTV was 50.0 Gy, a non-significant difference compared to unguided (p > 0.05) β Relative volume receiving more than 95% of the prescribed dose (V 95%, CTV ) Correlation coefficient between the best cosine fit and the original breathing signal  Free breathing: correlation coefficient was 0.66  Breathing guidance with RPM: correlation coefficient was 0.72, a non-significant difference compared to free breathing β  Breathing guidance with ABC: correlation coefficient was 0.77, significantly more regular than free breathing (p < 0.05) α Power dominant frequency (PDF) of breathing signal  Free breathing: the PDF was 0.04  Breathing guidance with RPM: the PDF was 0.08, significantly more regular than free breathing (p < 0.05) α  Breathing guidance with ABC: the PDF was 0.08, significantly more regular than free breathing (p < 0.05) α  Tables 2 and  160  Table 3. 161 Table 4 is an assembly of these 27 articles' findings and whether their results were significantly positive, negative, or non-significant. It should be noted that the number of outcomes exceeds the number of identified articles because most articles investigated more than one outcome.  (56,57,62) Breath hold stability & reproducibility 3 / 6 (44,63,68) 3 / 6 (43, 44) Gating efficiency 17 / 42 (46,50,55) 25 / 42 (47,49,50) Image Other* 5 / 11 (51,62,66) 5 / 11 (47,67) 1 / 11 (62)

IV. DISCUSSION
Findings from the 27 identified articles yielded a diverse range of breathing guidance intervention strategies being utilized on a range of different cancer types. Breathing guidance strategies ranged from buzzer signals to customized, interactive guides. Of the 27 included articles in this systematic review, 21 yielded at least one statistically significant positive outcome from the use of breathing 170 guidance, with a further 2 articles reporting non-significant improvements (or not reporting the significance of improvements) from the use of breathing guidance, and 4 articles reporting at least one statistically significant negative result. Of the 4 studies that yielded negative results, 3 investigated audio-only guidance, which resulted in larger breathing motion amplitudes, an undesirable trait in most radiation oncology and radiology procedures. 22,62,[70][71][72][73][74][75] Of the findings 175 assembled in Table 4, 63 were positive statistically significant, 82 were non-significant (or significance not reported), and 7 were negative statistically significant. It should be noted that of the 82 non-significant (or significance not reported) results, 35 noted improvements from the use of breathing guidance, 12 of which were reported to be non-significant, and 23 did not report the significance. 180 Of the 27 identified articles 12 were healthy volunteer studies and 12 were patient studies, with 3 studies recruiting both healthy volunteers and patients; the most investigated cancer type was lung cancer (12 studies), followed by breast (2 studies) and liver cancer (2 studies). Of the breathing guidance intervention strategies, most were designed to facilitate regular breathing (21 articles); 4 articles detailed breath-hold guidance, 1 study investigated both regular breathing and breath-hold 185 guidance, and 1 study investigated quasi-breath-hold breathing guidance where each exhale was extended to 3, 5, or 7 seconds. Medical imaging was performed in 15 studies, and radiation treatment was performed (or simulated) in 4 studies. Given these numbers, and as evident from Table 4, there are areas of breathing guidance which require more investigation. For example, research into the impact of breathing guidance on radiation treatment margins and target coverage 190 is limited and largely inconclusive, with all results thus far being non-significant. Further investigation into this area would be valuable as such findings would also give insight to the impact of breathing guidance strategies on patient outcomes. Further to this, of the 27 identified articles, none were randomized studies, indicating that future study designs should incorporate randomization. 195 20 of the 27 identified articles did not explicitly control for confounding, however the authors of this review paper did not consider this to bias their results. Of the 27 articles, none declared any conflicts of interest; however two articles acknowledged at least partial funding from either Phillips  2009)). However, these articles received positive quality assessment scores, as such, the authors of this review paper did not consider the results presented in these articles to be biased.

IV.A. Breathing Guidance for Breath-Holds
Breath-holds are a well-documented and frequently utilized strategy for minimizing anatomic motion during imaging and treatment. 43 44 These studies also resulted in improved image quality and intra-fraction motion management.
Breathing guidance has also been developed for deep-inspiration breath holds (DIBH). 63, 68 DIBH is often performed by the patient in left breast cancer radiotherapy to minimize the radiation damage to the lung and heart. 79 Prompts used to guide patient towards regular breathing have undergone considerable development and refinement over the years as detailed in Table 2 and Table 3. Audio-only guidance typically appeared in the form of verbal instructions or tones, 50-52, 56, 57, 62 and while the regularity of breathing was improved, it also increased the amplitude of breathing-motion. 48 (2008), utilising audio and visual prompts together poses no increase in the patient's cognitive load; i.e. it does not require additional concentration for the patient to incorporate two different sensory forms of guidance at once. 58 The guiding prompts of breathing guidance have developed from a buzzer sounding to provide a queue for breath-holds, to a patient display presenting breathing-surrogates superimposed with a 240 guiding interface. Additional constraints have been added to the visual prompts to further manage respiration, such as the displaying of inhale and exhale limits, 47, 48, 50, 60 a waveguide with fixed period and amplitude for the patient to match their own breathing to, 54 and combinations thereof. 58,64,65 In addition to the nature of guiding prompts utilized, study design has also factored into influencing patient acceptance and compliance with the breathing guidance intervention. Studies in 245 which patients used breathing guidance multiple times demonstrated improved breathing consistency with time. 50, 58, 65 Hence, to achieve optimal compliance with breathing guidance, patient training and repeated sessions are of importance to bolster their familiarity with the system; such elements have been absent in previous patient studies which yielded non-significant results. 69, 88,89 While this systematic review yielded 27 articles, it should be noted that some articles that were 250 in contention required considerable discussion between the authors to conclude on their exclusion from the final selection. The main factor influencing the decision to exclude these articles was the control group criterion; while several studies investigated a breathing guidance intervention strategy, the control group was not of the same breathing type (see inclusion criterion 5). 81, [84][85][86][89][90][91][92] While the search undertaken and review of articles by the authors was performed as objectively 255 as possible it should be noted that two of the authors of this systematic review: Sean Pollock and Paul Keall are either first-or co-authors of 3 and 9 of the 27 included articles, respectively, investigating the breathing guidance intervention: audiovisual biofeedback. Their familiarity with breathing guidance strategies led to the identification that a gap in the literature existed in that a review of such research had yet to be performed; however, unintentional bias may have permeated 260 this review towards audiovisual biofeedback. To minimize this bias, co-author Robyn Keall was invited to review and screen the identified 319 (see Figure 3); where there was disagreement between reviewers, a discussion was undertaken amongst all authors until consensus was reached. While 21 of the 27 included articles reported at least one statistically significant positive finding from the use of breathing guidance interventions, bias should also be noted that papers 265 reporting on positive results are more likely to be published than papers with negative results. 93,94 This notes the systemic bias in scientific reporting and the possibility that negative results on breathing guidance may not have been published. The largely positive results found in this systematic review indicate that further clinical studies are warranted, and should be focussed on (1) utilizing training and multiple sessions to maximize 270 patient compliance with the breathing guidance system, and (2) further determining the clinical impact of breathing guidance interventions by investigating outcomes pertaining to treatment margins, toxicity, and patient outcomes. Such factors are being explored in ongoing and upcoming studies, with some preliminary results presented thus far. 95-97 275

V. CONCLUSION
A systematic review of breathing guidance intervention strategies in radiotherapy and radiology has been performed and 27 studies were identified. In 21 studies statistically significant improvements from the use of breathing guidance were observed. No studies observed worse breathing consistency with guidance; however, audio-only guidance, while facilitating regular breathing, also 280 increased respiratory amplitude which is undesirable in most circumstances. Studies that have repeated breathing guidance across multiple sessions have observed an improvement in participant compliance from one session to the next, emphasising the importance of patient practice and training. Such insights are valuable in designing breathing guidance studies in terms of both guiding prompts used and patient familiarity with the intervention to maximize the effectiveness of the 285 intervention. The largely positive results found here indicate that further clinical studies are warranted to further assess and quantify the clinical impact of breathing guidance, along with the health technology assessment to determine the advantages and disadvantages of the use of breathing guidance strategies.

290
This project was supported by an NHMRC Australia Fellowship and the Bob and Nancy Edwards Scholarship. The authors would like to thank Informa Healthcare and Elsevier for allowing us the use of their Figures in our manuscript, as well as the authors of those papers the Figures appeared in for producing such a high quality of work we wished to utilize to improve the clarity of our manuscript's rationale. 295

CONFLICT OF INTEREST
Paul Keall is one of the inventors of US patent # 7955270 and Paul Keall, Robyn Keall, and Sean Pollock are shareholders of Respiratory Innovations, an Australian company that is developing a device to improve breathing stability. Respiratory Innovations did not provide any support or funding for this project. 300