Image guidance with the use of CBCT has been used for prostate motion correction during treatment setup verification.¹⁶ Although daily CBCT-guided soft-tissue matching is routinely used for prostate radiotherapy, limited evidence exists regarding its dosimetric consequences following actual couch corrections in South African public-sector radiotherapy departments.

The image guidance observes and corrects for inter-fractional motion due to anatomical changes to ensure that treatment techniques such as VMAT are accurately delivered, and dose distribution about these changes during the daily fractional treatments is still predicted by the reference plan and do not correspond to the delivered treatment.¹⁷

This retrospective single-centre study was conducted at the Department of Radiation Oncology, Gauteng Academic Hospital, South Africa. Clinical records were reviewed to identify 30 consecutive patients with localised prostate cancer who received prostate-only, moderately hypofractionated external beam radiotherapy using VMAT. A total dose of 60 Gy in 20 fractions was prescribed. All included patients had a complete set of 20 daily CBCT images acquired before treatment delivery.

Pelvic immobilisation was achieved through the ProSTEP® Elekta (Stockholm, Sweden). All treatments were delivered using an Elekta Versa High Definition linear accelerator (Elekta AB, Stockholm, Sweden), with daily image guidance using the onboard CBCT imaging system.

A standardised bladder filling and rectal emptying protocol was implemented throughout treatment. Patients were instructed to empty their bladder and then consume 300 mL – 500 mL of water approximately 30 min before treatment. The aim was to achieve comfortable bladder distension, defined as approximately two-thirds bladder filling.

Rectal emptying was facilitated through a high fibre diet and the use of laxatives when required, with the aim of maintaining the anteroposterior rectal diameter at less than 4 cm. Bladder filling and rectal distension were assessed on each pre-treatment CBCT image before treatment delivery.

The daily pre-treatment CBCT datasets were retrieved from the online verification system and imported into the Monaco Treatment Planning System (Elekta AB, Stockholm, Sweden). Target volumes and OARs, including the bladder and rectum, were re-contoured on each CBCT dataset. Contouring was performed with input from an experienced radiation therapist and a radiation oncologist to ensure anatomical accuracy and consistency.

Each CBCT dataset was subsequently re-planned using the original treatment beam arrangement and planning parameters to maintain comparability with the corresponding reference planning CT. Dose was recalculated using the anatomy represented on each CBCT dataset.

Thirty patients with histologically confirmed localised prostate adenocarcinoma treated between 2019 and 2022 were included in this study. Eligibility criteria are discussed in the next section.

All patients underwent CT simulation using a Philips Brilliance Big Bore CT scanner (Philips Medical Systems, The Netherlands). Patients were positioned supine and immobilised using a ProSTEP® immobilisation device to ensure reproducibility.

Standardised bladder and rectal preparation protocols were implemented. Patients were instructed to empty their rectum before simulation and treatment. Bladder preparation was achieved by consuming 300 mL – 500 mL of water approximately 45 min before imaging, aiming for consistent and comfortable bladder filling (at least two-thirds full). Patients were encouraged to follow a high fibre diet during treatment, and mild laxatives were prescribed to achieve regular rectal voiding. The intent was to minimise rectal distension with an acceptable luminal calibre of ≤ 4 cm at widest extent. Optimal compliance of bladder and rectal volume was prioritised prior to CBCT corrections.

Computed tomography images were acquired with a slice thickness of 2 mm–3 mm, and the CBCT images were acquired with the preset of KV collimator S20, KV filter F0 and KV panel position at small. These datasets were then imported into the Monaco Treatment Planning System (Version 5.11.03; Elekta AB, Stockholm, Sweden) for treatment planning.

Target volumes and OARs were delineated in accordance with ICRU Report 83 recommendations. Clinical target volume (CTV): Prostate gland. Planning target volume (PTV): CTV with a uniform 7 mm all round, and 5 mm posteriorly to minimise rectal dose.^18,19

Volumetric modulated arc therapy treatment plans were generated using at least two full arcs on the Elekta Versa HD platform. The prescribed dose was 60 Gy delivered in 20 fractions (3 Gy per fraction), consistent with a moderately hypofractionated regimen. PTV and OAR constraints are detailed in Table 1. These were derived from published French Genitourinary group guidelines.²⁰

Daily pre-treatment CBCT imaging was performed using the Elekta Versa HD onboard imaging system. Soft-tissue registration was conducted before treatment delivery to align the prostate with the reference planning CT (as seen on Figure 1).

Image registration was performed using a six-degree-of-freedom (6-DoF) algorithm. However, in routine clinical practice, only translational corrections (three degrees of freedom: (1) lateral, (2) longitudinal and (3) vertical) were applied. Rotational deviations of less than 3° were considered acceptable and were not corrected.

All applied couch shifts were systematically recorded in the Elekta Mosaiq record-and-verify system for subsequent analysis.

Cone beam CT datasets were exported from the online verification system and imported into the Monaco Treatment Planning System for analysis. Each CBCT dataset was rigidly registered to the corresponding planning CT to ensure accurate spatial alignment.

Given the known limitations of CBCT Hounsfield Unit (HU) accuracy for dose calculation, anatomical information from the CBCT was mapped onto the planning CT dataset. Dose recalculation was therefore performed on the reference CT using the CBCT-derived anatomy to maintain dosimetric accuracy (as seen in Figure 2).

For each treatment fraction, a corresponding recalculated plan was generated. No modifications were made to beam geometry, optimisation parameters, normalisation or monitor units. This approach ensured direct comparability between the original planned dose distribution and the recalculated dose based on daily anatomical variation.

Dose - Volume Histogram (DVHs) were generated for both the original treatment plans and the recalculated plans based on daily CBCT anatomy.

The primary dosimetric endpoints were:

Differences between planned and recalculated dose metrics were evaluated on both a per fraction basis and cumulatively across all treatment fractions.

Dosimetric variation in OARs was recorded; however, these parameters were not included in the primary analysis.

Following the CT simulation, the acquired CT images were exported to the treatment planning system (Elekta Monaco). The radiation oncologist delineated the target volumes and OARs, after which the radiotherapist and medical physicist developed the treatment plan and performed the necessary quality assurance (QA) procedures. Once approved, the treatment plan was transferred to the linear accelerator (Linac) for treatment delivery. Before each treatment session, patient positioning was verified using CBCT imaging, with the images archived in the record-and-verify system (Elekta MOSAIQ).

For this study, the archived CBCT images were retrieved and contoured (as seen in Figure 3). The contoured CBCT datasets were subsequently imported into the treatment planning system, where treatment re-plans were generated and optimised.

The original treatment plans and the CBCT-based re-plans were then evaluated to compare the dosimetric parameters and quantify any dose deviations (as seen in Figure 4).

Step 1: Illustrates the CBCT image fused with the CT image before table correction is applied and treatment takes place.

Step 2: Illustrates the CT and CBCT images fusion after CBCT image import into the planning system.

Step 3: Illustrate the plan optimisation of the CBCT (re-plan).

Statistical analysis was performed using Statistical Package for the Social Sciences (SPSS) (Version 28, International Business Machine [IBM] Corp., Armonk, New York, United States). Descriptive statistics were used to summarise couch shifts and dosimetric parameters. Normality was assessed using skewness and kurtosis measures.²¹ The collected data in the form of couch-shift values (cm), also plans and re-plans in terms of absorbed dose, were entered into IBM, SPSS (Version 28, IBM Corporation, Armonk, New York, United States 2021) software for conducting the analysis where the p-value statistical significance or non-significance was assessed. Initially, data cleaning was executed to identify any anomalies. The subsequent analyses encompassed descriptive statistics including means, 95% confidence intervals, standard deviations, medians, IQR and measurements of skewness and kurtosis. Boxplots were employed to visually illustrate the distribution of average couch shifts. A scrutiny for normality indicated data skewness; hence, nonparametric Spearman’s Rho correlation was employed to assess the presence of significant correlations between the couch-shift measurements and the dosimetry.

A multiple linear regression model was employed to ascertain whether couch shifts could serve as predictors of dosimetry. The significance level is the probability of rejecting the null hypothesis when the null hypothesis is in fact true. The p-value is defined as the probability, assuming the null hypothesis of no effect or difference, of obtaining a result as extreme as or more extreme than the one observed. The significance level (α) was set at α < 0.05, with outcomes deemed significant if the resultant p-value was < 0.05. To compare mean shifts (standard deviation) in each direction of the couch shifts, analysis of variance (ANOVA) analysis was employed. It is a statistical formula that is most employed for hypothesis testing at present. It spans a diverse range of applications and possesses the adaptability to accommodate a wide array of experimental designs. It is used to compare variances across the means of different groups and also involves handling each treatment as a quantitative variable, assigning a value of 1 when it is present and 0 when it is absent.²¹

Spearman’s rank correlation coefficient was used to evaluate associations between couch shifts and dosimetric variation due to non-normal data distribution.

Differences between planned and delivered PTV dose metrics (D_98% and D_2%) were assessed using one-way ANOVA.

Multiple linear regression analysis was performed to determine whether couch-shift magnitude and direction predicted dosimetric variation. A p-value < 0.05 was considered statistically significant.

Ethical approval for this study was obtained from the Faculty of Health Sciences Research Ethics Committee at the University of Pretoria (Reference No. 260/2022). Given the retrospective nature of the study, a waiver of informed consent was granted by the Committee.

All patient data were anonymised during data extraction, documentation and analysis to ensure confidentiality and compliance with ethical standards.

Daily CBCT-guided soft-tissue registration resulted in measurable translational couch shifts across all axes (as seen in Figure 5).

Mean (± s.d.) shifts are described in Table 2. Shifts to the left were a mean of 0.08 ± 0.06 cm, while to the right, shifts averaged 0.06 ± 0.04 cm.

The largest displacements were observed in the superior and posterior directions. Shifts exceeding 5 mm were infrequent.

Mean PTV D_98% was 57.8 ± 1.18 Gy for planned doses and 57.4 ± 1.67 Gy for delivered doses, with no statistically significant difference (F = 1.57, p = 0.201).

Similarly, PTV D_2% values showed no significant difference between planned (62.97 ± 0.39 Gy) and delivered doses (62.94 ± 0.60 Gy; F = 1.84, p = 0.135).

All patients maintained PTV coverage above the institutional threshold (D_98% ≥ 57 Gy). Twenty-five patients (83%) demonstrated < 3% deviation between planned and delivered D_98%, while the remaining patients showed deviations > 3% that did not compromise clinical acceptability.

Spearman correlation analysis demonstrated a statistically significant association between superior shifts and reduced PTV D_2% (r = –0.37, p = 0.04). No other directional shifts showed significant correlations with PTV dose metrics.

Multiple linear regression analysis did not demonstrate a significant overall model for predicting variation in PTV D_98% (F = 1.57, p = 0.20) or D2% (F = 1.84, p = 0.135).

However, inferior shifts were independently associated with a small reduction in D98% (β = –0.46, p = 0.05), and superior shifts predicted a reduction in D_2% (β = –1.91 Gy, p = 0.01).

Despite statistical significance, all observed variations remained within clinically acceptable dose constraints.

Across all 600 fractions (datasets), planned and delivered dose–volume histograms demonstrated strong agreement. No clinically meaningful under dosage or over dosage of the PTV was observed (as seen on Figure 6).

Mean PTV D_98% was 57.8 ± 1.18 Gy for planned doses and 57.4 ± 1.67 Gy for delivered doses, with no statistically significant difference (F = 1.57, p = 0.201).

Similarly, PTV D_2% values showed no significant difference between planned (62.97 ± 0.39 Gy) and delivered doses (62.94 ± 0.60 Gy; F = 1.84, p = 0.135).

D_98% (minimum dose): Multiple linear regression evaluating all six shift directions did not show a significant mode. Overall model: F = 1.57, p = 0.20.

When tested individually, inferior shifts independently predicted a small reduction in D_98%, β = –0.46 and p = 0.05. The effect size was < 5% and did not compromise target coverage.

D_2% (maximum dose): The multivariable model was not significant. Overall model: F = 1.84, p = 0.135. Shift in the superior (SUP) direction independently predicted a reduction on D_2%. β = –1.91 Gy and p = 0.01. This effect reduced D_2% (hotspot dose) but remained within planning tolerance.

Clinical target volume to PTV margins are intuitionally dependent and are determined by the treatment technique available to ensure correct PTV position and the impact of organ motion on target localisation. In a comprehensive review, Yartsev et al. evaluated multiple clinical trials evaluating CTV to PTV margins in prostate cancer. This includes margins utilised and techniques applied for correction of organ motion. A 7-mm / 5-mm anisotropic margin used at out institute matches the overall CBCT range and is more conservative anteriorly and laterally while in keeping with most frequently used 5-mm margin posteriorly. Oehler et al calculated a 7.2-mm anterior, 5.5-mm posterior, 7.1-mm sup-inf and 5.5-mm lateral margin using IGRT or CBCT with KV/Kv fiducials. This study also concluded that IGRT margins fell within 5 mm – 8 mm range which supports a 7 mm anterior and lateral margin application. Li et al. validated soft-tissue image guidance with clinical 8/5-mm plans and found that after online soft-tissue alignment, 8/5 mm met CTV coverage criteria in 95% of fractions, while 6/4 mm did so in 91%.^19,22,23

This study evaluated the dosimetric impact of daily CBCT-guided soft-tissue alignment and couch-shift corrections during moderately hypofractionated VMAT for localised prostate cancer. Across 600 fractions in 30 patients, daily image guidance resulted in small but consistent translational corrections, predominantly in the superior–inferior and anterior–posterior directions. Despite these shifts, delivered dose distributions remained closely aligned with the planned DVHs, with no statistically significant differences in PTV D_98% or D_2%. These findings confirm that daily soft-tissue IGRT maintains dosimetric doses as prescribed robustness in hypofractionated prostate radiotherapy and support the implementation of tight PTV margins.¹⁸

Day-to-day variability in prostate position is well recognised and driven largely by changes in rectal and bladder filling.²⁴ Numerous studies have shown that prostate motion can exceed 5 mm in any direction and may compromise coverage in hypofractionated regimens, where larger doses per fraction magnify the effect of geometric inaccuracy.²⁵ Similar to previous reports, this study found that the majority of corrections occurred in the superior and posterior directions, reflecting the variability in rectal filling.^26,27 Maund et al. demonstrated that the nature of the shifts facilitated a reduction in CTV to PTV margins, with a significant reduction in OAR dose, most notable in the rectal dose.²⁷

Importantly, despite these shifts, delivered D_98% values stayed above the departmental coverage threshold (≥ 57 Gy) for all patients, with 83% showing < 3% difference between planned and delivered values. This high concordance indicates that daily CBCT-guided prostate alignment is sufficient to maintain PTV coverage even with small PTV margins of 5 mm – 7 mm.

Although overall differences between planned and delivered dose metrics were not statistically significant, directional associations were observed. Inferior shifts were associated with small reductions in D_98%, while superior shifts were linked to modest decreases in D_2%.

These findings are consistent with previous dose-tracking studies, where shifts along steep dose gradients result in measurable but clinically minor variations in dose distribution. Importantly, in this study, all observed changes remained within established planning constraints, indicating that the magnitude of these effects is unlikely to be clinically significant when daily IGRT is applied.

These effects were statistically significant but clinically negligible, with all dose variations remaining within accepted constraints. Similar directional effects have been reported in dose-tracking studies, where shifts along the prostate’s steeper dose gradients produce modest changes in minimum or maximum dose.^28,29 One of these studies included treatment plans from both pencil beam scanning and proton beam therapy as well as VMAT.²⁹ Our findings confirm that the magnitude of these changes is small when rigorous daily IGRT is applied, implying that IGRT plays an important role in treatment setup of hypofractionated prostate treatments.

The robustness of dose delivery observed in this study reinforces the critical role of daily image guidance in hypofractionated prostate radiotherapy. Larger fraction size increases the biological impact of geometric uncertainties; therefore, precise localisation is essential. The present findings demonstrate that translational couch corrections based on daily CBCT are sufficient to maintain dosimetric accuracy, even without explicit correction of rotational errors. This supports existing evidence that small rotational deviations (< 3°) have limited dosimetric impact in prostate radiotherapy when translational alignment is optimised.

These findings carry significant implications for clinical practice. The demonstrated stability of the delivered dose supports the continued use of tight PTV margins in conjunction with daily CBCT guidance. Margin reduction has the potential to decrease dose to surrounding OAR, particularly the rectum and bladder, thereby reducing treatment-related toxicity.³⁰

Furthermore, the findings support the broader implementation of hypofractionated regimens, which offer significant advantages in terms of patient convenience and resource utilisation.³¹

Moderate hypofractionation (60 Gy/20 fractions) is now endorsed in multiple international guidelines and has demonstrated non-inferiority to conventional fractionation in large Phase III trials (CHHiP, PROFIT).⁶ However, its safe implementation requires precise localisation, as reduced fraction sizes and reduced margins amplify the risk of geographic miss.³²

This study provides evidence from a state-sector institution supporting the safe execution of IGRT-based hypofractionation. Our findings demonstrate no compromise in target coverage, no significant overdosing of hotspots excellent stability of delivered dose despite daily anatomical variation effective correction of motion using 3-DoF couch shifts, and feasibility of maintaining 3 mm – 5-mm PTV margins.

These outcomes align with international experience and are particularly valuable in resource-constrained settings where adoption of hypofractionation can greatly reduce machine load, improve access and shorten waiting times.³³ Supporting our approach of mapping CBCT anatomy onto the planning CT, the OAR dose constraints used in this study aligned with guidelines published for hypofractionated radiotherapy.²⁰

Taken together, the consistency of planned and delivered doses across multiple studies and our own cohort provides strong evidence that daily CBCT is a reliable tool for maintaining dosimetric fidelity in hypofractionated prostate radiotherapy.

Several limitations must be acknowledged. Firstly, the dose calculation was performed on the planning CT with CBCT-derived anatomy transferred onto the reference datasets; therefore, direct CBCT dose calculation was not performed due to known HU-density inaccuracies in the pelvis.^34,35 Secondly, only translational couch corrections were analysed; rotational effects were not quantified and may have contributed to small geometric uncertainties.³⁶ Thirdly, OAR dose variation was not included in this analysis, although future evaluation will be important to understand rectal and bladder dose stability in hypofractionated IGRT.

Despite these limitations, the dataset was complete for all 600 fractions, and the consistency of findings across patients strengthens the validity of the conclusions. The utilisation of CBCT images to visualise the prostate for image guidance has instilled confidence in adaptive radiotherapy, specifically in terms of precisely localising cancer tumours and enhancing the distribution of absorbed dose. Addressing the anatomical changes that patients undergo during treatment, such as loss of weight and the regression of the tumour, is facilitated by daily CBCT image acquisition.³⁷ This practice ensures that prescribed absorbed dose limits remain within acceptable tolerances, minimising the risk of inadequate target volume dosing and minimising absorbed doses to OARs.

Future work should incorporate deformable image registration and direct CBCT-based dose calculation to further refine dose-tracking accuracy. In addition, evaluation of OAR dose variation will be essential to fully characterise the clinical benefits of margin reduction.