In vivo dosimetry in pelvic brachytherapy

Brachytherapy is an effective treatment in the curative management of prostate and gynaecological cancers. With advances in technology, brachytherapy has increased in complexity in recent years. Human error, equipment malfunction, patient organ motion and radioactive source displacement can result in substantial deviation of delivered dose from planned dose. To limit adverse clinical outcomes, adequate steps to improve the robustness of pathway processes, ensure the implementation of appropriate treatment margins and confirm the delivered dose must be considered. In vivo dosimetry is one such method of dose validation which, if implemented appropriately within clinical practice, is an attractive technique for reducing dosimetric uncertainties and identifying potential errors. This review aims to describe the dosimetric uncertainties and potential errors associated with brachytherapy, the potential for in vivo dosimetry in adaptive brachytherapy as a key method of dose validation, and the clinical considerations and future directions of in vivo dosimetry. Advances in knowledge This paper describes the potential role for in vivo dosimetry in the reduction of uncertainties in pelvic brachytherapy, the pertinent factors for consideration in clinical practice, and the future potential for in vivo dosimetry in the personalisation of brachytherapy.


INTRODUCTION
The use of ionising radiation for radiotherapy is an effective cancer treatment strategy which induces cancer cell death through direct and indirect DNA damage. 1In brachytherapy, radioactive sources such as strontium-90 ( 90 Sr), iridium-192 ( 192 Ir) and iodine-125 ( 125 I) which emit radiation in the form of β particles ( 90 Sr) 2 or γ rays ( 192 Ir, 125 I) 3 are directly inserted within or in close proximity to the radiotherapy target.These are permanently inserted in the case of low dose rate (LDR) brachytherapy sources (e.g. 125I) which emit radiation at <2 Gy per hour or inserted for a short period of time in the case of high dose rate (HDR) brachytherapy sources (e.g. 192Ir) which emit radiation at >12 Gy per hour. 4,5achytherapy is a common method of treatment for prostate and gynaecological malignancies and has several advantages over external beam radiotherapy (EBRT).These include the ability to deliver a much higher dose of radiation directly to the cancer. 4Internal source placement with brachytherapy is associated with rapid radiation dose fall-off as a result of the inverse square law.This advantageous dose distribution improves the therapeutic ratio resulting in the capability of delivering higher radiation doses to the tumour and/or reduced dose to adjacent organs at risk (OARs) compared with EBRT, thereby increasing the probability of cure and/or reducing the likelihood of adverse treatment effects while maintaining high tumour control rates. 6rachytherapy is also associated with a far shorter time commitment and fewer visits required on the part of the patient. 4e high dose of radiation delivered by brachytherapy can result in adverse clinical outcomes if any deviation from the prescribed radiotherapy plan occurs.Deviations can occur due to uncertainties in dose delivery following movement of OARs or radioactive source positioning and displacement, or as a result of human or equipment error.In vivo dosimetry has the potential to identify some of these deviations and thereby allow for their rectification.This review aims to summarise the need for and current status of in vivo dosimetry for clinical brachytherapy, describe considerations for integration of in vivo dosimetry within routine clinical practice and propose future developments.

THE NEED FOR IN VIVO DOSIMETRY IN BRACHYTHERAPY
The brachytherapy pathway involves multiple steps, each of which could potentially be associated with dosimetric Brachytherapy is an effective treatment in the curative management of prostate and gynaecological cancers.With advances in technology, brachytherapy has increased in complexity in recent years.Human error, equipment malfunction, patient organ motion and radioactive source displacement can result in substantial deviation of delivered dose from planned dose.To limit adverse clinical outcomes, adequate steps to improve the robustness of pathway processes, ensure the implementation of appropriate treatment margins and confirm the delivered dose must be considered.In vivo dosimetry is one such method of dose validation which, if implemented appropriately within clinical practice, is an attractive technique for reducing dosimetric uncertainties and identifying potential errors.This review aims to describe the dosimetric uncertainties and potential errors associated with brachytherapy, the potential for in vivo dosimetry in adaptive brachytherapy as a key method of dose validation, and the clinical considerations and future directions of in vivo dosimetry.Advances in knowledge This paper describes the potential role for in vivo dosimetry in the reduction of uncertainties in pelvic brachytherapy, the pertinent factors for consideration in clinical practice, and the future potential for in vivo dosimetry in the personalisation of brachytherapy.
uncertainty and also the potential for human or equipment error.These steps include the insertion of applicators, imaging, target delineation, applicator reconstruction, radiotherapy planning and delivery. 7As a high dose per fraction is delivered with HDR brachytherapy compared to EBRT, and LDR brachytherapy is usually limited to a single procedure, the potential for deviation of treatment dose from planned dose is much greater and so errors and uncertainties should be minimised where possible. 5e American Association of Physicists In Medicine (AAPM) classify absolute dose deviations of 10-20% and positioning differences of >5 mm between planned and actual treatments as being 'very wrong' and highly likely to result in a serious adverse clinical outcome. 8eviations classified as 'wrong' include dose deviations of 5-10%, in addition to positional deviations of 3-5 mm. 8 This includes relatively small discrepancies between the measured and delivered dose in each step in the brachytherapy treatment pathway which can cumulatively amount to clinically significant adverse events. 5

Human and equipment error
Several aspects of the planning and delivery pathway which have the potential for human error can be mitigated by in vivo dosimetry.These include errors in patient setup, applicator or needle catheter placement, guide tube connections, applicator and seed reconstruction, image fusion and calculation of appropriate source time in each position, based on the residual radioactivity of the source. 5Administrative mistakes can also result in clinically significant errors, e.g. in Philadelphia when an incident occurred in 2008 in which a patient received 125 I seeds of incorrect strength (0.38 mCi instead of 0.509 mCi) due to an error in the ordering process. 9This resulted in the insertion of radioactive seeds giving 25% less than the intended radiation dose.
The potential for equipment errors exists due to the complexity of brachytherapy.For example, a serious incident occurred in Indiana in 1992 when the radioactive 192 Ir source detached from the guide wire and remained undetected in a patient receiving treatment for anal carcinoma for 5 days resulting in a substantial radiation overdose and subsequent death as a direct result of radiation exposure.More than 90 other individuals were also exposed to the radioactive source as a result of this equipment failure. 10Other potential errors include defects in source loading time and positions within the HDR afterloader of up to 2.0% for multiple interstitial needle applicators due to either software or motor malfunctions. 11,12

Imaging uncertainties
The quality of imaging modalities in brachytherapy can also contribute to uncertainties in brachytherapy planning.Kim et al 13 found that random displacements of HDR prostate brachytherapy catheters by one CT slice thickness resulted in average dose errors of 0.7, 1 and 1.7% for slice thickness values of 2, 3 and 5 mm respectively.The partial volume effect, in which more than one tissue type occurs in a voxel, can result in the blurring of tissue boundaries. 14Uncertainties associated with pixel resolution can also impact fusion, contouring and dose reconstruction. 15

Internal organ motion
For HDR gynaecological brachytherapy, there is a delay between applicator insertion, CT and/or MRI imaging, and insertion of the radioactive source during which OAR motion may occur. 16or LDR prostate brachytherapy, while radioactive seeds are inserted under real-time ultrasound guidance, the position of the target and OARs may change following insertion, and may have already changed in position since planning in the case of preplanning. 5veral studies have evaluated the movement of OARs during gynaecological brachytherapy and the resulting dosimetric impact (Table 1).Variability in findings exists, however, the majority of studies report increases in dose to the bladder and rectum as a result of OAR movement between planning and treatment delivery.Anderson et al 22 compared planning and pre-treatment MRIs in HDR cervical brachytherapy and found >10% deviation in the minimum dose received by the most irradiated 2 cc (D2cc) of the bladder in 38.9% of fractions, rectum D2cc in 58.3% of fractions and bowel D2cc in 52.8% of fractions.Mazeron et al., 19 Yan et al 17 and Rey et al 25 found significant increases in rectal dose due to OAR movement between treatment planning and delivery, Nomden et al 21 reported significant increases in rectal dose among outliers in their study and Lang et al 23 found non-significant changes in rectal dose with the rectum dose constraint met in all cases.
In a study of 31 patients treated with pulsed dose rate (PDR) prostate brachytherapy, Dinkla et al 26 found the distance between the prostate and the rectum as measured on CT decreased from an average of 7.1 to 5.9 mm after 24 h and to 5.3 mm after 48 h.This resulted in an increase in the rectum D2cc from planned dose of an average of 14.8% after 24 h, and 17.3% after 48 h.Similarly, due to OAR movement, the bladder D2cc increased by an average of 25.4% after 24 h and 24.8% after 48 h and the urethra D0.1cc decreased by an average of 2% after 24 h and 3.2% after 48 h.Milickovic et al 27 evaluated urethral and rectal movement in HDR brachytherapy.The greatest movement occurred between the planning ultrasound and post-treatment ultrasound with mean movements of 1.1±1.3mm for the urethral base and 0.4±0.4mm for the rectum.

Radioactive source displacement
Studies of observed radioactive source displacement in a clinical context in gynaecological and prostate brachytherapy are summarised in Tables 1 and 2 respectively.The dosimetric impact resulting from positional displacement of radioactive sources (e.g. 192Ir) can be quantified in respect of deviations in the D90 of the high risk clinical target volume (HRCTV) in gynaecological brachytherapy.Variable impact is reported in studies, with minimal dosimetric impact in the study by Nomden et al, 21 an intrafraction mean decrease of 2.5±10.8% in the study by Nesvacil et al, 24 and a statistically significant mean decrease of 4.1% on the second day of brachytherapy and 5.7% on the third day compared with the original plan in the study by Rey et al. 25

In vivo dosimetry in pelvic brachytherapy
The majority of catheter displacements during prostate brachytherapy tend to occur in a caudal direction relative to the prostate gland.In a study of 20 patients receiving HDR prostate brachytherapy, Simnor et al 38 found more than 70% of catheter needles had moved >5 mm in the caudal direction by the second fraction and more than 35% had moved >5 mm by the third fraction, with more than 20% of catheters in total moving ≥12 mm.Without correction, these displacements would have resulted in a mean 28% decrease in the D90 to the planning target volume (PTV) in the second fraction, and a mean 32% decrease in D90 PTV in the third fraction.Holly et al 36 found an average displacement of 11 mm to result in >20% decrease in V100 and 38% decrease in prostate D90 with a corresponding increase in the mean minimum dose delivered to the most irradiated 10% of the urethra (D10%) from 118 to 125%.In a study of pre-planned LDR prostate brachytherapy, Beaulieu et al 43 found only 13 of 35 studied cases had relatively constant volumes with <5% variation with significant changes up to 30% and a resulting mean dosimetric impact on prostate V100 of -5.7%, up to -20.9%.
Several studies have systematically manually displaced the position of radioactive sources in brachytherapy plans to determine the threshold of movement for significant dosimetric impact and are summarised in Table 3. Hoskin et al 49 report a 5% decrease in prostate D90 and Table 3. Summary of identified studies of manual source displacement in pelvic brachytherapy

Pertinent dosimetric results
2019 Poder et al. 46 20 HDR prostate three catheters displaced: AP, anteroposterior; CC, craniocaudal; D1cc, the minimum dose received by the most irradiated 1 cc of the volume; D0.1cc, the minimum dose received by the most irradiated 0.1 cc of the volume; D2cc, the minimum dose received by the most irradiated 2 cc of the volume; D90, dose delivered to a minimum of 90% of the volume; HDR, high dose rate; HRCTV, high risk clinical target volume; TCP, tumour control probability; V80, volume receiving 80% of prescription dose; V100, volume receiving 100% of prescription dose; V120, volume receiving 120% of prescription dose

BJR
Houlihan et al V100 to be associated with a 10% increase in biochemical failure in prostate brachytherapy.Poder et al 46 found the proportion of plans which demonstrated at least a 5% reduction in target coverage parameters increased with displacements >2 mm.The study found that the target minimum V100 was not met in 75% of plans following a 3 mm shift in the cranial direction, in 50% of plans following a 3 mm shift in the caudal direction, and in 35% of plans following a 3 mm shift in the medial direction.In a study of HDR cervical brachytherapy plans, Tanderup et al 48  It is clearly important that the potential movement of all radioactive sources and OARs during brachytherapy is considered given the potential clinical impact geometric and dosimetric uncertainties can have. 5Displacements as small as 3 mm have been shown to have significant outcomes on brachytherapy dosimetry and target coverage in studies.

CURRENT STATUS OF IN VIVO DOSIMETRY IN BRACHYTHERAPY
In vivo dosimetry consists of real-time monitoring of radioactive source placement and dose during the delivery of radiotherapy.This involves the placement of radiation detectors in the vicinity of radioactive sources within the body, which relay the measured dose to the clinical staff and hence allows for comparison of calculated radiotherapy dose with actual dose delivered. 11In vivo dosemeters, therefore, allow for independent verification of brachytherapy delivery, comparison of institutional practice and quality assurance of radiotherapy treatment provision resulting in safer, more accurate clinical practice. 50This is of particular importance with the delivery of high brachytherapy doses, e.g. the delivery of a boost to the dominant intraprostatic lesion seen on MRI, which has been explored in recent studies. 51,52With increasing dose, the potential for adverse effects also increases and so precise accurate dose assessment is vital. 52e inclusion of in vivo dosimetry in clinical practice has been hesitant, due to a lack of affordable, efficient, commercially available dosemeters.The requirements for precision, stability and dosemeter positioning certainty are additional challenges that limit the routine adoption of in vivo dosimetry in clinical practice. 53Most studies to date focus on pre-clinical models demonstrating proof of concept, [54][55][56] although some clinical studies have been performed in pelvic brachytherapy.Dosemeters in the form of metal-oxidesemiconductor field-effect transistors (MOSFET), optical fibres and semiconductors have been tested clinically, all within HDR brachytherapy settings, with dosemeters inserted in the rectum, urinary catheter or within the brachytherapy target.While several are commercially available, [57][58][59] they are not routinely used in brachytherapy clinical practice. 11,53Limitations include angular and energy dependence of semiconductor diodes, energy dependence and limited lifespan of MOSFETs and Cerenkov light production in optical fibre dosemeters. 53lley et al 60 evaluated the feasibility and effectiveness of a nanoscintillator-based fibre-optic dosemeter (nanoFOD) compared to thermoluminescent dosemeters (TLD) in vaginal cylinder HDR brachytherapy.The dosemeter was adhered to the cylinder at a fixed distance, to which two TLDs were also attached to provide reference measurements.Real-time data were available for 27 fractions among 9 participants.The fibre-optic dosemeter readings were comparable to TLD measurements and 63% of measurements with the fibreoptic dosemeter were within 5% of the treatment planning system (TPS) (compared with 70% of TLD measurements), 26% were within 5-10% (22% of TLD measurements) and 11% were within 10-20% (7% of TLD measurements), with a median ratio of nanoFOD/TPS dose of 1.00 (IQR 0.94-1.02).The use of TLD as a reference standard demonstrated feasibility of the nanoFOD within a clinical setting.
In a study of a radioluminescent crystal dosemeter placed within a dedicated brachytherapy catheter during HDR brachytherapy Johansen et al 61 found measured compared with planned doses to differ by a mean of -4.7% (range -17 to +12%) with mean shifts of brachytherapy needles of 0.2±1.1 mm (radial) and 0.3±2.0mm (longitudinal).Limitations of the study included the measurement of displacements relative to the radioluminescent crystal rather than to patient anatomy and the use of only one dosemeter.Integration of in vivo dosemeters with imaging systems and the use of an array of dosemeters would reduce positional uncertainty. 62Additional studies of the clinical use of in vivo dosimetry in pelvic brachytherapy are summarised in Table 4.
While the magnitude of what constitutes a clinically acceptable deviation is variable and specific to each patient site, it is essential for clinically useful in vivo dosemeters to detect deviations classified as 'wrong' by the AAPM (dose distribution and delivery deviations of 5% and positioning deviations of 3 mm) and the ideal is for detection sensitivity to be as high as possible. 8Currently, the accuracy of in vivo dosimetry systems varies significantly with mean differences between calculated and measured radiation dose for MOSFET, optically stimulated dosemeters and semiconductors of up to 6.7, 4.7 and 15.5% respectively (Table 4).

CLINICAL CONSIDERATIONS
Several clinical considerations are necessary in order to overcome the current limitations associated with the integration of in vivo dosimetry into routine clinical brachytherapy.

Workflow
Service and resource pressures as well as the existing complexities of brachytherapy procedures are potential barriers to the practical implementation of in vivo dosimetry. 72In addition, the greater the time between imaging and treatment delivery, the greater the risk of internal organ motion and increased positional uncertainties. 65ntegration with the existing patient workflow, e.g.affixing the in vivo dosemeters to the afterloading device in HDR gynaecological brachytherapy, is preferable, to avoid the need for additional procedures. 53

Dosemeter placement
Important considerations during in vivo dosimetry are the accuracy, reproducibility and stability of dosemeter placement.Appropriate BJR In vivo dosimetry in pelvic brachytherapy fixation must take place to ensure no movement occurs between dosemeter insertion and delivery of radiotherapy.Waldhäusl et al report dosemeter probe shifts as small as 2.5 mm result in measured dose differences of >10%. 73Due to the steep dose fall off associated with brachytherapy, dosemeter movement of only a few millimetres can result in erroneous dose measurements, the triggering of false alarms and the failure to detect radiotherapy dose deviations.This requirement for accurate placement of in vivo dosemeters can limit their practical use. 4 Therefore, dosemeters must be used in conjunction with imaging techniques to ensure adequate localisation.In addition, the insertion process of in vivo dosemeters should be integrated with existing equipment such as urinary catheters and applicators to minimise risks of bleeding and infection. 53

Sensitivity and specificity
The sensitivity and specificity of dosemeters are other important considerations.In the context of in vivo dosimetry, sensitivity is the likelihood that a dosemeter will detect errors in dose or positioning if these errors exist. 71Use of a dosemeter with high sensitivity, therefore, should detect any dosimetric errors that occur and confirm the absence of such errors if no alarm sounds.Specificity is the likelihood that when a dosemeter signals an error in dose or positioning, that this is a true error and not a 'false alarm' . 71A balance must be struck to ensure that the vast majority of errors are detected without the expense of triggering excessive false alarms.Dosemeter susceptibility to external factors and environmental influences including humidity, temperature, direction, angular dependence and energy dependence are important to consider and these influences should be minimised where possible, or at least correction factors clearly documented. 50he atomic number of the chosen dosemeter should be similar to water to reduce energy dependence. 50st Significant costs are associated with the implementation and use of in vivo dosimetry in radiotherapy. 74Many hospitals and healthcare systems have limited budgets and it is imperative that the dosemeters are cost-effective. 75,76

FUTURE OF IN VIVO DOSIMETRY
Time-resolved, or real-time dosimetry, has the potential to significantly reduce brachytherapy errors.Triggering an alarm during the delivery of radiation in brachytherapy signifies to the clinical team that an error has occurred and prompts immediate investigation and resolution of this error.This may result in treatment interruptions, prolonging of treatment times and may cause discomfort for the patient in addition to increasing the complexity of the procedure for clinical staff who must compensate for the detected dose error. 77ntegration of in vivo dosemeters with treatment planning software to In the future, in vivo dosemeters could also facilitate uptake in radiobiology-guided brachytherapy.Hypoxia is associated with radiotherapy resistance and inferior clinical outcomes. 78It is commonly associated with solid tumours due to their immature, disorganised vascular supply which develops as a result of overexpression of proangiogenic factors. 79Movsas et al 80  Dose escalation within identified areas of tumour hypoxia is a potential method by which the negative effect of hypoxia on tumour control can be overcome. 83Hypoxic sensors have been described in the literature including the aforementioned Eppendorf oxygen electrode, 80 a fibre-optic sensor using ruthenium luminophore incorporated into a silicone rubber polymer tip, 84 fluorescent peptide probes based on the oxygen-dependent degradation domain of HIF-1α, 85 imaging such as blood-oxygen-level dependent (BOLD) functional MRI which evaluates changes in signal intensity between diamagnetic oxyhaemoglobin and paramagnetic deoxyhaemoglobin, 86 and PET/CT using hypoxia-specific tracers such as 18 F-fluoromisonidazole ( 18 F-FMISO). 87Integration of a hypoxic sensor within an in vivo dosemeter would allow for tumour hypoxia to be mapped and measured in real-time with dose escalation to these areas.
A similar approach could be taken with dosemeters which detect the presence of DNA double-strand breaks (DSBs).DSBs are a critical from of DNA damage and, if not correctly repaired, are an important mechanism by which radiation induces cell death. 88A preclinical model consisting of magnetic streptavidin beads attached to four kilobase pair DNA strands has shown promising results in the detection of DNA DSBs. 89Detecting these DSBs in real-time during brachytherapy would provide the opportunity to adapt the dose depending on their quantity and location.For example, increased dose could be delivered in areas with minimal DSBs with reduced dose in areas with many DSBs.
Brachytherapy plans adapted to tumour hypoxia and DSBs would enable dose escalation in areas of radioresistance and reduction in areas of radiosensitivity.By detecting these, in vivo dosemeters have the potential to provide for a personalised radiotherapy approach in real-time adaptive brachytherapy which should result in improved outcomes in terms of tumour control and toxicities for patients.

CONCLUSION
Prostate and gynaecological brachytherapy have increased in complexity in recent years, due to advances in imaging, techniques and software.Adaptive brachytherapy has the potential to optimise target dose distribution, reduce side-effects from treatment and introduces the potential for dose escalation.It is important that the radiation doses delivered to the brachytherapy target and OARs are accurately measured and documented, and in vivo dosimetry provides an opportunity for adaptive brachytherapy in real-time.
There are several important considerations regarding the practicalities of in vivo dosimetry, which must be addressed prior to its incorporation into routine clinical practice, and the benefits of the procedure must outweigh any potential risks to the patient.The EU Horizon 2020 Origin project 90 is working to address the current barriers to the clinical implementation of in vivo dosimetry and to develop a real-time system based on optical fibre-based sensing technology for use in prostate and gynaecological brachytherapy.

FUNDING
The ORIGIN project is an initiative of the Photonics Public Private Partnership ( www.photonics21.org), and has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No 871324.Authors are also supported by the Research and Development Division of the Public Health Agency of NI (COM/5610/20).

Table 1 .
Summary of identified studies of HRCTV and OAR movement during cervical brachytherapy

Table 2 .
Summary of identified studies of observed radioactive source and OAR movement during prostate brachytherapy CBCT, cone-beam computed tomography; D1/2/3, day 1,2,3; D2cc, the minimum dose received by the most irradiated 2 cc of the volume; D90, dose delivered to a minimum of 90% of the volume; HDR, high dose rate; HRCTV, high risk clinical target volume; OAR, organ at risk; PDR, pulsed dose rate Values in Gy reported as EQD2.6 of 15 birpublications.org/bjrBr J Radiol;95:20220046

Table 2 .
CBCT, cone-beam computed tomography; CI, confidence interval; CTV, clinical target volume; D10, dose delivered to a minimum of 10% of the volume;

Table 2 .
found the HRCTV dose-volume histogram (DVH) shifted by a mean of approximately 2% per mm shift in the lateral and longitudinal directions, and by approximately 1.5% per mm shift in the anterior and posterior directions.The D2cc of the bladder and rectum changed by approximately 5% per mm shift in the anterior and posterior directions and the D0.1cc of the same OARs changed by approximately 6% per mm shift in the same directions.

Table 4 .
Summary of identified clinical studies using in vivo dosimetry in pelvic brachytherapy allow for real-time monitoring of radiation dose delivery and distribution could allow for the brachytherapy plan to be adapted in such a manner as to compensate for any significant dose deviations.Such advances are dependent on high precision and accuracy of in vivo dosimetry and advanced software development but would minimise any additional treatment time and clinical staff workload as a result of errors in radiation dose and distribution.
82asured pO2 in human prostate carcinomas using Eppendorf microelectrodes and found these to be significantly lower than the pO2 present in normal muscle controls, with increasing hypoxia associated with increasing clinical stage.The ratio of prostate to normal muscle pO2 was the strongest predictor for biochemical control.Similar results were found in studies by Turaka et al81of prostate cancer, and by Rofstad et al82of cervical cancer, demonstrating the need for adaptive strategies to target hypoxia within tumours.