Background
Breast-conserving therapy (BCT), which involves a wide local excision followed by radiotherapy to the whole breast, is the standard treatment for early-stage breast cancer [
1]. The efficacy of BCT for the treatment of early-stage breast carcinoma has been established in multiple randomised trials [
2,
3]. External-beam partial breast irradiation (EB-PBI) has recently garnered increasing interest [
4,
5]. Several studies reported that EB-PBI, which delivers radiotherapy to the postoperative tumour bed (TB) with a margin of adjacent breast tissue, could achieve excellent results in certain patients. However, there are inherent challenges in defining accurate target volumes for partial breast irradiation (PBI). Studies demonstrate significant interobserver variability between radiation oncologists in defining the lumpectomy cavity, indicating the need to improve the accuracy and consistency in the delineation of the TB [
6,
7].
Surgical clips and seroma are important markers for delineating the TB for EB-PBI [
7,
8]. Kirby et al. [
7] reported that the number of implanted markers influences the accuracy of target delineation and that five to six surgical clips are preferable for TB delineation for PBI or breast boost radiotherapy. Landis et al. [
8] indicated that the shift of the centre of mass (COM) decreased and the percent volume overlap (PVO) increased significantly as the seroma clarity score (SCS) increased. The influence of the number of metal clips, SCS, delineation experience and contouring guidelines for the delineation of TB in CT images have been investigated [
6,
7,
9-
11]. However, the effect of interobserver variability on the delineation of the TB using seroma and surgical clips based on 4DCT scan is not clearly established. To investigate the impact of different markers on interobserver variability in the delineation of TB based on 4DCT scan for EB-PBI, we analysed the TB delineated by five observers in this study based on clips, seroma, and both clips and seroma.
Discussion
The accuracy of target volume delineation is critical for EB-PBI. The optimal target volume for EB-PBI remains to be established. Most reports define the tumour excision cavity or postoperative seroma as the target volume in treatment planning [
7,
8,
15]. However, these studies also reported significant interobserver variation in delineating post-lumpectomy cavities. van Mourik et al. [
16] investigated breast target volume delineations among thirteen observers in eight patients. They reported that the presence of clips or seroma reduced interobserver variability but that significant volumetric and spatial interobserver variability was observed in clinical target volume (CTV) even with the help of delineation guidelines. Therefore, reducing interobserver variability is a pressing issue. To improve delineation accuracy and consistency, we selected patients with SCS 3 ~ 5 and ≥5 surgical clips to mark the boundaries of the lumpectomy cavity.
Seroma clarity and volume in the lumpectomy cavity decreased as a function of time from surgery to the CT scan. The use of the CT-based seroma to guide the EB-PBI target volume is difficult due to a lack of clearly defined standards. Kader et al. [
6] selected 205 women with early-stage breast cancer to undergo planning CT after breast conserving surgery and found that the mean seroma volume decreased from 47 cc to 30 cc during postoperative weeks 3-8, stabilised during weeks 9-14 (mean 21 cc) and was involuted beyond 14 weeks (mean 9 cc). In our study, the average duration from lumpectomy to 4DCT scan was 10 weeks (range, 3-16). The volume of TB
C approached the volume of TB
S for six patients during weeks 4-8, and the TB
C < TB
S in two patients during weeks 0-3. However, after 8 weeks, the volume of TB
C was larger than that of TB
S for 12 patients. Therefore, given the magnitude and time trends of seroma volume and clarity loss, the optimal time to obtain the planning CT scan for PBI is within 8 weeks after surgery.
Surgical clips are not always consistent with the edge of seroma and the boundary of the lumpectomy cavity [
13,
17]. Ding et al [
13] measured the three dimensional displacements of the GTV
C, the GTVs and the GTV
C+S and found that in the LR, AP and SI directions, the displacements were 0.9 mm, 1.05 mm and 1.20 mm for GTV
C; 0.80 mm, 1.05 mm and 0.80 mm for GTVs; and 0.90 mm, 1.20 mm and 1.40 mm for GTV
C+S, respectively. In other words, the three dimensional displacements of GTV
C+S were greater than those of GTV
C and GTV
S. Yang et al. [
17] also measured the distance between surgical clips and the edge of the seroma in a coronal plane in women who underwent wide local excision of breast cancer and reported that the mean seroma edge extended beyond the clips by 0.3-0.5 cm. This study indicates that the volume of TB
C+S delineated by observers was significantly larger than TB
C and TB
S. These results may be due to decreased seroma clarity and volume in the lumpectomy cavity from the time of lumpectomy to the 4DCT scan, as well as the variability between the TB
C, TB
S, TB
C+S regarding treatment margin.
Interestingly, we observed no significant differences between observers for TB
S (p = 0.867) and COV
S (p = 0.061). This could be explained by the short average duration from lumpectomy to planning CT and because the SCS 3 ~ 5 in the lumpectomy cavity improved the visualisation of the surgical cavity. Landis et al. [
8] reported similar results in patients of SCS 4 and 5, as the average COM shift was 3 mm and 2 mm, respectively, and the PVO was 77% and 87%, respectively. Wong et al. [
18] also reported that after reviewing contouring guidelines, the differences in seroma target volume (STV), CTV, and planning target volume (PTV) were no longer statistically significant. Although Dzhugashvili et al. [
9] reported that the conformity index of TB delineation was significantly improved by the placement of surgical clips within the lumpectomy cavity, a higher level of interobserver concordance was observed by the five observers when the SCS was 3 ~ 5 points in the lumpectomy cavity. These data suggest that interobserver variability in the delineation of TB based on seroma was not obvious when the SCS was 3 ~ 5 points.
Although interobserver variability in the delineation of the TB based on seroma was minimal, the clinical reality could vary across regions. This was a retrospective study, and all the enrolled patients had undergone wide-local excision of breast cancer with full-thickness unstitching of the excision cavity. Full thickness closure of the excision cavity and oncoplastic surgical procedures are becoming more widely practiced in some regions, and these surgical methods could reduce the rate of seroma and its reliability as a marker of the TB. Therefore, patients should be given five or more surgical clips to reduce interobserver variability when only clips are used in clinical practice. Moreover, the placement of clips in the surgical cavity could be based on guidelines, and the duration from lumpectomy to 4DCT scan could be shortened.
Hurkmans et al. [
19] reported that intra- and, to a greater extent, interobserver variability in the delineation of breast target volume on CT scans can be large. Both Dzhugashvili et al. [
20] and Yang et al. [
21] also reported interobserver variability (similar to the results presented here) in the delineation of the TB based on clips. These results can be explained by the fact that tissue stranding from the surgical cavity, proximity to muscle, dense breast parenchyma, and benign calcifications may be mistaken for surgical clips. Moreover, limited soft-tissue contrast on CT makes it an unreliable modality for detecting a layer of the image lacking surgical clips and when distinguishing between surgically induced densities and normal glandular breast tissue. Finally, the experience of the radiation oncologist and subjective determination of the location of the post-surgical cavity contribute to interobserver variability. Using a combination of information to more precisely define the TB, such as surgical reports, clinical palpation of the surgical defect and CT-based planning, may decrease interobserver variability. Additionally, clear communication between the surgeon and radiotherapist, including diagrammatic explanations, are crucial for accurately targeting the TB.
Cover et al. [
22] reported that when the EE phase was reviewed in the sagittal plane, gating would reduce the mean tumour mobility from 6.3 ± 2.0 mm to 1.4 ± 0.5 mm. Moreover, 4DCT simulation scan can reduce motion artefacts [
23,
24]. Therefore, to investigate spatial mismatches of interobserver variability in the delineation of the TB based on different markers, the difference between the MD
C, MD
S and MD
C+S were further analysed and compared based on the end-exhalation phase. Our study found interobserver variability between the MD
C, MD
S and MD
C+S, as well as between the MDc and MDs, MD
C and MD
C+S, and MD
S and MD
C+S. Our results suggest spatial mismatch existed among observers in the delineation of the TB based on clips, the seroma, both the clips and seroma. Additionally, the MD of the seroma was larger than that of the clips as well as both the clips and seroma, and the MD of both the clips and seroma was larger than that of the clips alone (Table
4). However, the average CI reported by van Mourik et al. [
16] was 0.53, which was considerably higher than that in the present study. This difference is likely due to differing target volume and calculation methods. Landis et al. [
8] and Li et al. [
25] reported even higher CI values ranging from 0.73-0.75; however this analysis focused on the PTV instead of the TB, the CI of which increases due to the larger volumes.
Image-guided techniques can improve clip and seroma localisation during treatment, potentially enabling the use of a smaller GTV-to-PTV margin. Ultrasound image guidance has also been investigated [
26]. However, it is unclear if the margin is sufficient to account for interobserver contour variability. Further studies are needed to determine, whether contouring variability could result in an underdosing of the clips or seroma. Thus, it is imperative that future studies aiming to reduce margins from current treatment practice take interobserver contour variability into consideration.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
BG and JBL designed the study, contributed to the data collection, and drafted the manuscript. WW, MX, QS, YJZ and CQL made important contributions in delineating the TB based on clips, seroma, and both clips and seroma. WW and YLG contributed to data collection and analysis. All authors read and approved the final manuscript.