Impact evaluation of intra-fractional variation on online adaptive radiotherapy for postoperative cervical and endometrial cancer

Abstract

BACKGROUND

Online adaptive radiotherapy (oART) has demonstrated improved target volume coverage and enhanced sparing of surrounding pelvic organs through daily re-optimization based on pretreatment imaging. Recently, iterative cone-beam computed tomography (iCBCT) has been integrated into oART workflows, facilitating precise daily adaptation. However, the dosimetric consequences of intra-fractional variations for clinical target volume (CTV) and organs at risk (OARs) remain insufficiently characterized.

AIM

To investigate intra-fractional CTV and OARs variation and their impact on iCBCT guided daily oART for postoperative cervical and endometrial cancer.

METHODS

Seventeen patients treated with daily postoperative iCBCT guided oART with rigorous bladder and rectal preparation protocols were enrolled. CTV and OARs were contoured on pre- and post-treatment iCBCT scans. The average surface distance (ASD), dice similarity coefficient (DSC), and 95% Hausdorff distance (HD) were utilized to evaluate the difference between pre- and post-treatment structures. Dosimetric outcomes for the pretreatment target volumes and OARs were recalculated using posttreatment contours to assess the impact of intra-fractional variation.

RESULTS

A total of 434 treatment fractions were analyzed, with an average interval time of 22 minutes between two iCBCT scans. Minimal variations were observed in the bladder, rectum, and CTV both pre- and post-treatment, with DSC exceeding 0.8. The vaginal CTV exhibited centroid deviations of 0.46 mm anteriorly, 0.11 mm laterally, and 0.58 mm superiorly, along with ASD of 1.69 mm and 95% HD of 6.42 mm. Weak correlations were observed between vaginal CTV posterior-anterior centroid deviations and rectal superior-inferior deviations (P = 0.017). Minimal dosimetric differences were observed pre- and post-treatment, with V_100% for the adapted plan of nodal CTV being 99.94% vs 99.08% and vaginal CTV being 99.97% vs 98.66%.

CONCLUSION

Daily iCBCT-guided oART with strict bladder and rectal preparation effectively compensates for intra-fractional variations, maintaining CTV coverage and OAR sparing across all treatment fractions.

Key Words: Cone-beam computed tomography; Online adaptive radiotherapy; Intra-fractional variation; Endometrial cancer; Cervical cancer

Core Tip: By analyzing 434 treatment fractions with rigorous bladder and rectal preparation protocols, the study demonstrated that intra-fractional motion of clinical target volumes and organs at risk was minimal and had negligible impact on dosimetric outcomes of postoperative cervical and endometrial cancer. These findings provide robust clinical evidence supporting the reliability and precision of iterative cone-beam computed tomography-guided online adaptive radiotherapy in the postoperative pelvic setting.

INTRODUCTION

Postoperative radiotherapy for endometrial cancer and early-stage cervical cancer treated with surgery, consisting of hysterectomy, could increase locoregional control rates and improve survival for patients with adverse prognostic factors[1,2]. The clinical target volume (CTV) that includes the proximal vagina and any paravaginal tissue (CTV-V)[3], and the inter-fractional motion of the vagina could be up to 27.9 mm induced by variation in adjacent bladder and rectum filling[4,5]. Of note, larger planning target volume (PTV) margins and volumes may increase the risk of irradiation toxicity, as higher intestinal doses have been associated with the fall of small intestine into the space vacated by the true pelvis after hysterectomy[6,7]. Currently, online adaptive radiotherapy (oART), which adapts to pelvic anatomical changes by daily re-contouring of target volumes and organs at risk (OARs), could reduce the PTV margin[8,9] and has significant advantages in postoperative radiotherapy for endometrial and cervical cancer.

Preliminary studies suggested that oART techniques achieved superior dosimetric coverage for the target volume and significantly improved dosimetry of surrounding OARs in pelvic malignancies[10,11], and in clinical practice the oART is entirely based on pretreatment imaging for per-fraction treatment plan re-optimization and reports the dosimetric results. However, it is reasonable to expect that intra-fractional variation of the filling status of the rectum and bladder has an impact on the position and shape of the adjacent CTV-V and novel CTV (CTV-N), resulting in dosimetry differences. Ensuring rigorous bladder filling and rectal emptying before treatment may help mitigate this movement to some extent. Currently, there are no studies which have precisely evaluated the impact of intra-fractional variation.

The purpose of this study was to investigate the intra-fractional position variations of the CTVs and OARs for postoperative cervical and endometrial cancer and their impact on daily iterative cone-beam computed tomography (CT)-guided oART dosimetry.

MATERIALS AND METHODS

Patient eligibility

This study included a cohort of 17 patients from October 2022 to March 2023 who underwent hysterectomy for cervical or endometrial cancer and required postoperative radiotherapy due to high-risk factors. Patients who had indications for radiotherapy in the inguinofemoral region and extended-field pelvic radiotherapy were not enrolled in this study. The study was reviewed and approved by the Institutional Review Board of Peking Union Medical College Hospital, and all patients gave informed consent for participation in the study.

Target delineation and treatment

Separate CTV-N and CTV-V were contoured according to the Radiation Therapy Oncology Group Consensus Guidelines[3]. A PTV was expanded by an isotropic margin of 0.5 cm from the CTV, and a prescribed dose of 45 Gy or 50.4 Gy in 25 or 28 fractions was applied to the PTV.

All patients underwent rigorous bladder and rectal preparation before the simulation and each fraction, including informing the patients of the treatment time in advance, emptying the rectum in advance, and emptying the bladder 1 hour and 40 minutes before the appointment time, and then drinking 450-500 mL of water within 10 minutes according to height and weight, to ensure repeatability and avoid intra-fraction errors.

Iterative cone-beam CT acquisition and daily adaptive workflow

All patients received daily postoperative iterative cone-beam CT (iCBCT)-guided oART (Ethos, Varian Medical Systems, Palo Alto, United States). Each pretreatment iCBCT scan (iCBCT1) was acquired with a tube voltage of 125 kVp, tube current of 1080 mAs, and contiguous 2-mm slice thickness. The typical scan dose parameters were CTDIVOL 21.60 mGy and DLP 324.0 mGy·cm, with an average scan time of 36.7 seconds. All scans were performed under the standard Pelvis clinical protocol of the Varian Ethos system and iCBCT reconstruction algorithm was used. The “influencer” (bladder, rectum and bowel) and CTV structures were automatically generated by the Ethos adaptive platform in iCBCT₁. The physician then reviewed and manually edited the above structures, the adapted plan (adaptive treatment plan re-optimized based on the anatomical structure of the day) and scheduled plan (reference plan recalculated after matching the PTV modified on that day to ensure the maximum dose coverage of the PTV) was generated and selected. When the treatment was completed, the posttreatment iCBCT scan (iCBCT₂) was acquired immediately.

Intra-fractional variation and dosimetry evaluation

For all adaptive fractions from each patient, the posttreatment “influencer” and CTV structures were contoured on iCBCT₂ by the same physician and verified by a second expert radiation oncologist. The centroid deviation of the above structures in three translation directions, posterior-anterior, left-right, and superior-inferior was measured to evaluate intra-fractional motion. In addition, the average surface distance (ASD), dice similarity coefficient (DSC), and 95% Hausdorff distance (HD) were used to evaluate the difference between pre- and posttreatment CTVs and "influencer". The ASD describes the difference between the boundaries of two structures and is used to compare the similarity of the contours of two structures[12]. The following formula was used:

Where A and B represent the pre- and posttreatment CTVs, respectively; the h (a, B) represents the minimum distance from a point a on contour A to contour B; the h (A, b) represents the minimum distance from a point b on contour B to contour A.

The DSC evaluates the similarity between the pre- and post-treatment CTVs by calculating the degree of overlap[13], using the following formula: DSC (A, B) = 2 |A∩B|/(|A| + |B|), where |A| represents the volume of pretreatment CTV contour, |B| represents the volume of posttreatment CTV contour, and |A∩B| represents the volume where the two contours overlap.

HD evaluates the similarity between the pre- and post-treatment CTV contours by calculating the distance between them[12,14], while 95% HD represents the distance value at the 95^th percentile after all the aforementioned distances are sorted from small to large. And 95% HD could eliminate the influence of abnormal outliers and is more robust than HD, as shown in the following formula: HD (A, B) = max [h (A, B), h (B, A)], where h (A, B) represents the maximum value of the distances from all points on the pretreatment CTV contour A to the posttreatment CTV contour B. Similarly, h (B, A) represents the maximum value of the distances from all points on the posttreatment CTV contour B after treatment to the pretreatment CTV contour A.

The V_100% (volume of CTV-N and CTV-V receiving 100% of the prescribed dose), D_99% (the dose to 99% volume of CTV-N and CTV-V), and the dose of OARs were recorded for each fraction. A comparison was made between the adapted and scheduled plans. To assess the impact of intra-fractional variation on target volumes and OARs, dosimetric outcomes for the pretreatment target volumes and OARs were recalculated using posttreatment contours. This was performed via a Python-based automated analysis program (Python 3.9), comparing two conditions: The adapted plan with posttreatment contours (adapted plan + iCBCT₂), and the scheduled plan with posttreatment contours (scheduled plan + iCBCT₂). These results were further compared to the pretreatment dosimetry, which included the adapted plan with pretreatment contours (adapted plan + iCBCT₁) and the scheduled plan with pretreatment contours (scheduled plan + iCBCT₁). The workflow of this study is shown in Figure 1.

Figure 1 Workflow of this study. CT: Computed tomography; oART: Online adaptive radiotherapy; iCBCT: Iterative cone-beam computed tomography; CTV: Clinical target volume; OARs: Organs at risk; ASD: Average surface distance; DSC: Dice similarity coefficient; HD: Hausdorff distance.

Statistical analysis

Statistical analyses were performed using SPSS 27.0 (IBM. Corp., New York, NY, United States). Given that the CTV-V is susceptible to daily change due to the pushing boundaries of the adjacent bladder and rectum, a least-square linear regression was used to analyze the correlation between CTV-V deviation and factors including average time, volume difference, volume increase rate, and centroid deviation of surrounding organs. r was calculated to assess the strength of these correlations. Additionally, the Mann-Whitney U test was applied to compare the intra-fractional CTV-V centroid deviations between fractions with average treatment times less than 25 minutes and those exceeding 25 minutes. A P value of less than 0.05 was considered statistically significant for all tests.

RESULTS

Adaption information

All patients completed the anticipated daily oART, with a total of 434 fractions. The adapted plan was selected for all fractions, and the average time between iCBCT₁ and iCBCT₂ was 22 minutes (range: 17-32 minutes).

Influencer and target volume variations

The mean bladder volumes before and after treatment measured from iCBCT₁ and iCBCT₂ were 353.9 ± 130.2 cm³ and 397.9 ± 138.7 cm³, respectively. The mean rectal volumes before and after treatment measured from iCBCT₁ and iCBCT₂ were 62.9 ± 33.5 cm³ and 61.9 ± 25.9 cm³, respectively. For the difference between pre- and post-treatment CTVs and "influencer", the overlap rates of CTV-N, CTV-V, bladder and rectum were high before and after treatment, with DSC values greater than 0.8, while the DSC value of bowel was 0.6. The centroid deviation of CTV-V was 0.46 ± 2.58 mm that indicated motion in the anterior direction, -0.11 ± 1.42 mm that indicated motion in the left direction, and -0.58 ± 7.27 mm that indicated motion in the superior direction. The detailed values for ASD, DSC, 95% HD and centroid deviation are shown in Table 1.

Table 1 Comparison of target volume and organs at risk contours before and after treatment.

Region	ASD (mm)	DSC	95% HD (mm)	Posterior-anterior (mm)	Left-right (mm)	Superior-inferior (mm)
CTV-N	0.83 ± 0.33	0.9 ± 0.09	3.94 ± 1.17	0.75 ± 1.05	-0.14 ± 1.31	-0.71 ± 6.69
CTV-V	1.69 ± 0.74	0.8 ± 0.11	6.42 ± 2.45	0.46 ± 2.58	-0.11 ± 1.42	-0.58 ± 7.27
Bladder	1.75 ± 1.02	0.91 ± 0.04	7.24 ± 3.92	0.34 ± 2.01	-0.20 ± 1.09	-1.02 ± 5.57
Rectum	1.71 ± 1.09	0.81 ± 0.08	7.67 ± 4.54	0.17 ± 2.09	0.27 ± 1.59	1.38 ± 8.12
Bowel	3.56 ± 1.79	0.6 ± 0.25	18.8 ± 14	-3.99 ± 4.17	0.29 ± 6.07	1.47 ± 31.08

Positive values of centroid deviation indicate shifts towards the inferior, right, and anterior directions, while negative values reflect superior, left, and posterior shifts. ASD: Average surface distance; DSC: Dice similarity coefficient; HD: Hausdorff distance; CTV-N: Novel clinical target volume; CTV-V: Clinical target volume that includes the proximal vagina and any paravaginal tissue.

CTV-V intra-fractional variation analysis

There was no significant correlation between the time from iCBCT₁ to iCBCT₂ and the centroid deviation of CTV-V in the posterior-anterior, left-right, and superior-inferior directions (P = 0.612, 0.092 and 0.092, respectively). Additionally, no significant difference in centroid deviation of CTV-V was observed between fractions with the treatment time of less than 25 minutes and those exceeding 25 minutes, across the above three directions (P = 0.773, 0.392 and 0.392, respectively; Supplementary Figure 1).

Organ motion and volume variation had a weak impact on CTV-V motion, as shown in Figure 2. A weak but statistically significant correlation was observed between the centroid deviation of CTV-V in the posterior-anterior direction (Figure 2A) and the rectal centroid deviation in the superior-inferior direction (P = 0.017, r = 0.11). Additionally, weak correlations were found with bladder volume differences (P < 0.001, r = 0.35), bladder volume increase rate (P < 0.001, r = 0.18), and rectum volume increase rate (P < 0.001, r = -0.55). The anterior shifts in the centroid deviations of the bladder and rectum, along with the posterior shift in the centroid deviation of the bowel (P < 0.001, P < 0.001, and P = 0.021, respectively; r = 0.59, 0.64, and 0.12, respectively), combined with variations in bladder and rectum volumes, collectively contributed to the anterior displacement of the CTV. In addition, the centroid deviations of the bladder and bowel in the posterior-anterior direction were weakly correlated with the CTV-V centroid deviations in the superior-inferior direction (both P < 0.001, r = 0.23 and -0.24, respectively). No significant correlation was identified between rectum or bladder volume difference or increased rate and the centroid deviations of CTV-V in the superior-inferior direction.

Figure 2 Correlation between intra-fractional clinical target volume that includes the proximal vagina and any paravaginal tissue motion and the motion and volume variation of surrounding organs. The r values and P values are indicated for each comparison. Point 0 represents the clinical target volume that includes the proximal vagina and any paravaginal tissue (CTV-V) position and the positions of surrounding organs on the pre-treatment iterative cone-beam computed tomography, along with zero volume difference and volume increase rate. A: Correlation between the posterior-anterior shift of CTV-V and four factors with significant correlation coefficient: Superior-inferior shift of rectum, bladder volume difference, bladder volume increase rate, and rectum volume increase rate; B: Correlation between the superior-inferior shift of CTV-V and two factors with significant correlation coefficient: Posterior-anterior shift of bladder and bowel. CTV-V: Clinical target volume that includes the proximal vagina and any paravaginal tissue.

Dosimetric outcomes

Pre- and post-treatment dosimetric outcomes for all 434 fractions were analyzed. The results of the pre- and post-treatment dosimetric outcomes showed that there was no significant difference in CTV-N coverage, with the adapted plan (99.94% vs 99.08% V_100%; 100.87% vs 100.04% D_99%) and scheduled plan (97.02% vs 96.40% V_100%; 100.03% vs 99.36% D_99%). Similar findings were observed for the pre- and post-treatment CTV-V, with the adapted plan (99.97% vs 98.66% V_100%; 100.84% vs 99.07% D_99%) and scheduled plan (96.11% vs 96.41% V_100%; 92.97% vs 95.65% D_99%). Additionally, minimal differences were noted in the pre- and post-treatment dosimetric outcomes for the OARs, including bladder, rectum, and bowel. The complete dosimetry comparisons are presented in Table 2.

Table 2 Dosimetric outcomes of target volume and organs at risks.

ROIs	Dosimetric metrics	Adapted plan + iCBCT1	Adapted plan + iCBCT2	Scheduled plan + iCBCT1	Scheduled plan + iCBCT2	P1 value	P2 value
CTV-N	V100% (%)	99.94 ± 0.09	99.08 ± 8.88	97.02 ± 6.47	96.40 ± 10.72	0.45	0.34
CTV-N	D99% (%)	100.87 ± 0.20	100.04 ± 8.96	100.03 ± 1.41	99.36 ± 9.07	0.99	0.16
CTV-V	V100% (%)	99.97 ± 0.05	98.66 ± 9.02	96.11 ± 3.71	96.41 ± 9.80	< 0.05	0.58
CTV-V	D99% (%)	100.84 ± 0.19	99.07 ± 10.57	92.97 ± 9.00	95.65 ± 11.15	< 0.05	< 0.05
Bladder	V40Gy (%)	24.62 ± 4.54	26.22 ± 5.32	26.16 ± 5.80	27.35 ± 6.00	< 0.05	< 0.05
	V30Gy (%)	38.49 ± 5.93	40.25 ± 6.28	40.62 ± 6.96	41.80 ± 6.83	< 0.05	< 0.05
	V20Gy (%)	59.72 ± 7.05	61.54 ± 7.01	63.13 ± 8.99	64.19 ± 8.37	< 0.05	0.09
	V10Gy (%)	92.40 ± 3.78	92.91 ± 3.40	94.10 ± 3.91	94.19 ± 3.55	< 0.05	0.75
	Dmean (Gy)	26.77 ± 1.95	27.38 ± 2.06	27.71 ± 2.55	28.10 ± 2.48	< 0.05	< 0.05
Rectum	V40Gy (%)	38.18 ± 9.26	37.36 ± 11.01	39.39 ± 14.46	37.74 ± 13.82	0.47	0.11
	V30Gy (%)	56.53 ± 10.42	56.49 ± 11.36	59.77 ± 14.15	58.79 ± 13.53	0.06	0.33
	V20Gy (%)	74.13 ± 9.64	74.70 ± 9.79	79.11 ± 10.51	78.84 ± 10.04	< 0.05	0.71
	V10Gy (%)	94.27 ± 4.61	94.42 ± 4.67	95.54 ± 4.01	95.71 ± 3.75	< 0.05	0.55
	Dmean (Gy)	31.67 ± 3.39	31.61 ± 3.60	32.82 ± 4.29	32.48 ± 4.00	0.06	0.26
Bowel	V40Gy (%)	12.48 ± 6.39	9.92 ± 5.44	12.93 ± 6.73	10.24 ± 5.83	< 0.05	< 0.05
	V30Gy (%)	23.83 ± 10.91	21.11 ± 10.38	25.38 ± 11.99	22.11 ± 11.36	< 0.05	< 0.05
	V20Gy (%)	41.30 ± 17.02	39.45 ± 17.04	47.04 ± 19.78	44.79 ± 19.83	< 0.05	0.11
	V10Gy (%)	60.24 ± 24.25	59.60 ± 24.74	69.08 ± 27.92	68.75 ± 28.37	0.36	0.88
	Dmean (Gy)	18.05 ± 7.33	17.23 ± 7.17	19.83 ± 8.12	18.93 ± 7.91	0.07	0.12

¹The comparison between adapted plan + iterative cone-beam computed tomography (iCBCT) pretreatment iCBCT scan and adapted plan + posttreatment iCBCT scan.

²The comparison between scheduled plan + pretreatment iCBCT scan and scheduled plan + posttreatment iCBCT scan.

iCBCT₁: Pretreatment iterative cone-beam computed tomography scan; iCBCT₂: Posttreatment iterative cone-beam computed tomography scan; CTV-N: Novel clinical target volume; CTV-V: Clinical target volume that includes the proximal vagina and any paravaginal tissue.

DISCUSSION

Currently, reported outcomes of adaptive radiotherapy have been based on pretreatment adjustments[15,16]. However, intra-fractional motion, including the dynamic changes in the position and shape of adjacent organs such as the bladder, rectum, and bowel, may influence the accuracy of dose delivery and target coverage. We designed the current prospective study in which iCBCT scans were obtained before and after treatment during daily pelvic oART to determine the impact of intra-fractional anatomical variations on target volume (CTV-N and CTV-V) positioning and dosimetry in postoperative cervical and endometrial cancer patients who were carefully instructed to fill bladders and empty rectums before simulation and intra-fraction iCBCT scans.

Previous research has predominantly focused on inter-fractional variations, which are driven by changes in bladder and rectal filling between treatment sessions. Jhingran et al[4] measured the vaginal cuff motion during pelvic intensity-modulated radiation therapy after hysterectomy, with patients instructed to fill their bladders in a standardized way. Their results indicated that the median maximum marker motion was 0.59 cm in the left-right direction, 1.46 cm in the anterior-posterior direction, and 1.2 cm in the superior-inferior direction. Jürgenliemk-Schulz et al[17] found that to achieve complete coverage of 95% of the inter-fractional vaginal volumes, an expansion of 2.3 cm was required in the anterior-posterior direction, 1.8 cm in the left-right direction, and 1.5 cm in the cranial direction. In contrast, daily oART has the advantage of adapting to the deforming postoperative bed and minimizing inter-fractional motion for each treatment session. In this study, the target volume and OARs were contoured both before and after treatment to accurately assess intra-fractional variation. The results demonstrated that the centroid of CTV-V shifted by 0.46 mm in the anterior-posterior direction, 0.11 mm in the left-right direction, and 0.58 mm in the superior-inferior direction. These shifts were significantly smaller than the previously reported inter-fractional variations, underscoring the importance of real-time adaptation in radiotherapy. In addition, image verification and registration were performed before each treatment fraction to correct deviations, and the results further confirmed that the 5 mm PTV margin used in this study was sufficient.

Intra-fractional variations are influenced by treatment time, as longer durations allow more time for organ motion, which affects contour accuracy. This is an important factor to consider due to its relevance to patient compliance and treatment efficacy[18]. Compared to the treatment times of up to 1 hour required for magnetic resonance-guided linear accelerator to adapt the CTV and OARs[19,20], iCBCT-guided oART allows for target contouring and plan re-optimization in a shorter time, with an average treatment time of approximately 22 minutes in this study. However, no significant correlation was found between treatment time and CTV-V centroid deviation in the three directions. Previous studies have shown that intra-fractional variations in rectal gas, compared to the entire rectum volume, are relatively small and remain stable for 20-25 minutes in cervical cancer patients[21]. Consequently, we compared CTV-V centroid deviation between fractions with treatment time shorter or longer than 25 minutes. However, our analysis revealed no significant difference in CTV-V centroid deviation between these two groups. This implied that no certain benefit of limiting the treatment time within 25 minutes was confirmed in this study. This may be attributed to the fact that the longest treatment time in this study was only 32 minutes, and relatively short differences in treatment durations likely resulted in reduced intra-fractional variations caused by bladder and rectal filling, as reflected by the DSC values being greater than 0.8.

In contrast, bowel motion is more unpredictable and may significantly impact dosimetry in pelvic radiotherapy[22-24]. Our findings showed that the lower DSC value for the bowel, at 0.6, indicated that bowel motion during treatment is more variable compared to that of the bladder and rectum. Importantly, a low DSC for the bowel does not necessarily indicate inadequate contouring. The bowel is highly mobile even within the same fraction due to peristalsis, which is a well-recognized phenomenon in the literature, especially when a bowel bag is included in the contouring. Although iCBCT provides slightly lower soft-tissue resolution than diagnostic CT, all contours in this study were delineated by experienced radiation oncologists and verified independently, ensuring contouring consistency. To complement the DSC, we also evaluated 95% HD, which is more sensitive to local deformation and boundary shifts. The combined use of DSC and 95% HD therefore supports that the observed variability reflects genuine bowel motion rather than technical inaccuracy. Several approaches have been explored to reduce intestinal inter-fractional changes, including pretreatment enemas and the use of oral antidiarrheal medications after chemotherapy to minimize intestinal movement. However, these approaches are primarily employed in brachytherapy with fewer fractions and are challenging to implement in external radiotherapy, which typically involves 25 to 28 fractions. Despite the large geometric variation of the bowel, the dosimetric impact in our study was minimal. This apparent discrepancy can be explained by the fact that a low DSC primarily reflects shape variability and random redistribution of bowel, rather than a systematic shift toward the high-dose region. Centroid analysis and bowel dosimetry (V40 Gy) also indicated nonspecific boundary fluctuations rather than persistent displacement into target volumes. In addition, anatomically, postoperative bowel loops are dispersed within the pelvis and less likely to remain adjacent to the high-dose isodose surfaces, with the sigmoid colon contributing more to dose proximity than the small bowel[25]. Furthermore, compared to conventional treatment planning, daily adaptive radiotherapy re-optimizes the dose distribution at each fraction to spare bowel, effectively mitigating the impact of unpredictable motion on dosimetry[26]. These findings are consistent with emerging evidence that adaptive workflows can improve the dosimetric results.

Previous studies have indicated a weak correlation between target volume shifts in the cervical region along the anterior-posterior axis and changes in rectal volume[27,28]. Our findings demonstrated a weak negative correlation between CTV-V centroid shifts and the rectum volume increase rate (P < 0.001, r = -0.55). This suggests that while bladder and rectal motion could affect CTV positioning, the extent of movement was limited, likely due to the bladder filling and rectal emptying protocols employed before treatment, which helped to reduce variability. In our study, the mean rectal volumes before and after treatment were nearly identical under conditions of advanced rectal emptying.

A notable observation from this study is that intra-fractional variations in bladder and rectum positioning were minimal and had no significant impact on dosimetric outcomes. This finding indicated that, with rigorous bladder-rectum preparation and within a specified timeframe, the structures contoured on the adapted images could be directly utilized for dosimetric representation. The overlap rates for CTV-N, CTV-V, bladder, and rectum remained high both before and after treatment, with DSC values exceeding 0.8 in most cases, indicating strong concordance between pre- and posttreatment volumes. Despite small centroid deviations in CTV-V, with 0.46 mm in the anterior direction and -0.58 mm along the superior–inferior axis that indicated a minor superior shift, these shifts did not result in clinically significant differences in CTV dosimetric coverage or OARs sparing. The adapted plan consistently maintained high CTV coverage before and after treatment, with V_100% and D_99% values for CTV-N and CTV-V differing by less than 1% between pre- and post-treatment dosimetric evaluations. This suggests that the daily adaptive radiotherapy workflow used in this study is robust, effectively compensating for minor anatomical changes within each treatment fraction.

The limitations of this study include the relatively small sample size, primarily due to the significant clinic resources required for its implementation. Although our study only included 17 patients, a total of 434 daily oART fractions were analyzed, with each posttreatment structure being re-contoured. Additionally, this study focused solely on iCBCT-guided oART. The limited image quality of iCBCT compared to diagnostic CT or magnetic resonance (MR) imaging may affect contouring accuracy, particularly for soft-tissue targets such as CTV-V. Also, the intra-fractional variation may have a different impact on dosimetric outcomes of MR-guided adaptive radiotherapy, which typically involves longer treatment durations. In addition, our cohort included only postoperative patients, for whom inguinal irradiation was not indicated. Incorporating patients requiring inguinal or para-aortic irradiation in future studies could enhance representativeness and provide further insights into intra-fractional variations. Future multi-center studies with larger cohorts are warranted to validate and generalize these findings.

CONCLUSION

This study demonstrated that intra-fractional position variations of the CTVs and OARs in daily iCBCT-based oART with a 5 mm PTV margin for postoperative endometrial and cervical cancer have minimal impact on the dosimetric outcomes. The bladder and rectum showed limited intra-fractional motion with rigorous preparation and rapid adaption, and the adaptive workflow effectively compensated for anatomical changes, ensuring adequate CTV coverage and sparing of OARs across all fractions.

ACKNOWLEDGEMENTS

The authors sincerely thank the radiation oncologists, therapists and medical physicists at our department for their technical support and assistance in data acquisition and patient care.