Ultrasonographic Evaluation of Skin Toxicity Following Radiotherapy of Breast Cancer: A Systematic Review

The present review aimed to systematically review skin toxicity changes following breast cancer radiotherapy (RT) using ultrasound (US). PubMed and Scopus databases were searched according to PRISMA guidelines. The characteristics of the selected studies, measured parameters, US skin findings, and their association with clinical assessments were extracted. Seventeen studies were included with a median sample size of 29 (range 11–166). There were significant US skin changes in the irradiated skin compared to the nonirradiated skin or baseline measurements. The most observed change is skin thickening secondary to radiation-induced oedema, except one study found skin thinning after pure postmastectomy RT. However, eight studies reported skin thickening predated RT attributed to axillary surgery. Four studies used US radiofrequency (RF) signals and found a decrease in the hypodermis’s Pearson correlation coefficient (PCC). Three studies reported decreased dermal echogenicity and poor visibility of the dermis–subcutaneous fat boundary (statistically analysed by one report). The present review revealed significant ultrasonographic skin toxicity changes in the irradiated skin most commonly skin thickening. However, further studies with large cohorts, appropriate US protocol, and baseline evaluation are needed. Measuring other US skin parameters and statistically evaluating the degree of the association with clinical assessments are also encouraged.


Introduction
Breast cancer is the most common malignancy in women [1]. The incidence rate is 10.4% of all cancers globally [2]. Annually, 2.1 million women are diagnosed with breast carcinoma [3]. Adjuvant radiotherapy (RT) is frequently indicated after breast-conserving surgery [4] and in selected cases after mastectomy [5]. In early-stage breast cancer, radiotherapy significantly reduces tumour recurrence and improves overall survival [6]. Despite advancements in RT techniques, skin toxicity or radiation dermatitis is a common and distressing side-effect. Acute or early skin toxicity (up to 3 months) affects nearly all breast cancer patients with some degree of erythema, oedema, or dry desquamation. A higher RT dose may lead to moist desquamation and ulceration [7]. This causes discomfort, restricts daily activities, and may interrupt treatment sessions [8]. Chronic or late skin toxicity symptoms such as telangiectasia, hypo-or hyperpigmentation, and fibrosis are less Exclusion of studies with no statistical comparisons (case study or case series) or consisting of fewer than 10 patients, as well as reviews, editorials, and non-English or nonhuman studies Abbreviations: Pre-RT = pre-radiotherapy.

Search Strategy and Selection Process
PubMed (National Centre for Biotechnology Information) and Scopus electronic databases were searched to identify relevant articles published between the earliest record and 1 March 2022 with weekly automatic email updates. Search terms used for both databases can be accessed via Supplementary Table S1. Research articles were reviewed via title, abstract, and then, finally, via full text by F.A.H and independently reviewed and cross-checked by H.A.M. and N.Y. Reference lists of the included studies were also screened through the Google Scholar database to capture any additional relevant records. Spreadsheet software was used to organise and assess the titles of included studies and identify duplicates, whereas the abstracts were viewed through word-processing software. The selection results were discussed in team meetings until consensus was achieved. The study search and selection were completed on 22 March 2022.

Quality Assessment
We used a quality assessment tool from the National Heart, Lung, and Blood Institute, Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies, to assess the quality of the included studies (Supplementary Table S2).

Data Review and Extraction
After finalising the selection process, data extraction was performed by F.A.H and independently reviewed by H.A.M and N.Y. Information was extracted, and the following data were included: author(s), publication year, study country, study design, patient demographic information, US machine, RT treatment protocol, type of breast surgery, type of clinical assessment, and time of evaluations. Then, further data were systematically extracted on the basis of measured skin parameters and locations, ultrasound skin findings, and main findings. Lastly, we summarised the ultrasonographic skin changes according to the time of assessments into early and late skin toxicity changes and their association with clinical assessments.

Study Selection and Quality Assessment
The database searches yielded 259 records from PubMed and 127 from Scopus. After removing duplicates, 321 articles were reviewed for inclusion via title, abstract, and full text following PICOS criteria. Sixteen studies met the inclusion criteria. In addition, reference lists of the included studies (n = 386) were screened, and only one study was included. The literature search process is detailed in the PRISMA flow diagram (Figure 1). The included studies were of moderate quality, except for one of good quality.

Included
Reports sought for retrieval (n = 4) Reports not retrieved (n = 0) Reports excluded: No ultrasound assessments (n = 3) Figure 1. Search strategy based on PRISMA flow diagram for new systematic reviews which included searches of databases, registers, and other sources. Table 2 describes the characteristics of the included studies. The selected studies were published between 1998 and 2021 and were geographically diverse, representing eight countries. A total of 6/17 studies were conducted in the same centre in the USA (Emory University School of Medicine), almost by the same group of authors [9,[31][32][33][34][35]. One of these studies [9] was a more comprehensive follow-up study with 77% of the patients from the previous one [32]. Lastly, three studies were conducted by the same author in Newcastle Mater Hospital, Waratah, NSW, Australia [36][37][38].   Regarding study design, all of the included studies were prospective. The reports included 783 breast cancer female patients who received radiotherapy. The number, however, may be overestimated due to the possible overlap of patients reported by the same authors or group of authors in different studies, especially researchers from Emory University School of Medicine. The sample size varied widely between 11 and 166 (median, 29). The age range of the reported studies was between 26 to79 years. The BMI (mean and median range 23.5-29.9) of the patients was reported in five studies [9,32,33,39,41], and smoking history was reported in three studies [9,32,41] (n = 12/66, 14/70, and 2/34 patients, respectively). In addition, breast volume (range 177-6988.3 cm 3 or mL) was reported in five studies [9,32,33,41,45].

Study Characteristics
The majority of the studies included conventional fractionation RT (CF) as a treatment modality as it was the standard protocol before introducing newer techniques such as hypofractionated (HF) RT or accelerated partial breast irradiation (APBI). One study assessed toxicity from HF [40], and two studies compared CF and HF treatments regarding skin toxicity [33,39]. Most of the patients received electron boost technique to the tumour bed. Breast-conserving surgery (BCS) is the most common surgical approach to breast cancer in the included studies. Two studies evaluated patients following BCS and mastectomy [39,44], and one study evaluated patients purely postmastectomy [46].
Most studies used various clinical assessments or scoring to correlate or compare with objective US findings. The most frequently reported were Radiation Therapy Oncology Group (RTOG) and Common Terminology Criteria for Adverse Events (CTCAE) grading scales. The timing of clinical and US assessments varied widely between studies. The studies evaluated acute (early) toxicity, chronic (late) toxicity, or both. Overall, the evaluation time ranged from during RT sessions to 135 months after RT, except for one patient [38] who was evaluated 22 years post RT. A total of 10/17 studies included baseline US evaluation for comparison with subsequent observations and to determine any skin thickening predated RT from operation-related oedema.

Ultrasound Protocol
The studies utilised various brands of ultrasound machines and probe settings, except for three studies, which did not give details about the US machine used [9,32,33]. The probe frequency used for the measurements ranged from 4 to 20 MHz. A total of 8/17 studies performed skin ultrasound with a probe of 20 MHz. Some studies reported the lateral resolution (range 150-200 µm), axial resolution (range 60-70 µm), and maximum depth (range 6-80 mm) of the US probe. The level of the experience of the sonographers was poorly reported in the studies except for four studies: a "radiologist" [40], a "specialist for breast ultrasound" [39], "two specially trained staff" [38], and "two radiation oncologists and one ultrasound expert" [34]. Four studies reported that the sonographer was blinded to the clinical toxicity grading [31,35,40,46]. One of them was also blinded to the patient's treatment characteristics [40].

US Skin Parameters and Locations of the Measurements
Skin thickness (epidermis plus dermis) was the most frequently measured, which is the distance from the anterior echogenic border of the epidermis to the posterior echogenic border of the dermis using B-mode. Four reports measured the epidermal thickness (the most superficial layer of the skin), one of them as a mean value [36] and the others as skin thickness ratio (STRA) by dividing the mean epidermal thickness of the treated breast by the mean epidermal thickness of the contralateral breast [9,32,33]. Another three studies measured the dermal thickness, which is the middle layer of the skin [41,44,45]. The total cutaneous thickness (epidermal and dermal thicknesses) was another parameter used by [38] to evaluate the presence of cutaneous oedema. Details of the heterogeneity are tabulated in Table 3. Table 3. Ultrasound skin toxicity findings of the irradiated breast compared with the nonirradiated or pre-RT breast.

Author(s), Year Measured Parameters and Locations Ultrasound Skin Findings Main Findings
(Borm et al. 2021) [39] Skin thickness Locations: at 12:00, 3:00, 6:00, and 9:00 around the mamilla   The findings indicated chronic skin reactions - The medial aspect was consistently thicker than the lateral aspect Three studies evaluated the echogenicity of the dermis and the visibility of the echogenic line between the dermis and subcutaneous fat [38,44,45]. The entry echo of the skin and the signal intensity of the dermis were only assessed by [44].
Four studies (almost by the same group of authors) utilised ultrasound radiofrequency (RF) signals to obtain skin thickness as the distance between backscattered signals from the epidermis and those of the hypodermis [31,34,35,42]. In addition, the same group assessed the hypodermal integrity by measuring the Pearson correlation coefficient (PCC) from the US RF data. PCC was obtained by measuring the correlation between two variables representing the adjacent scan lines within a region of interest (ROI) situated along the hypodermal surface. One of these studies extracted the dermal toxicity (the difference between the skin thickness of the treated breast and that of the untreated breast) and the hypodermal toxicity (the difference between 1 minus the PCC of the hypodermal surface on the treated breast and the untreated breast) [34]. All thickness measurements were in millimetres, except for one study calculating the dermal thickness in micrometres [44].
The locations of the US measurements within the breast varied between studies. Four reports measured the four quadrants of the breast [9,32,33,45]. Another four used the 12:00, 3:00, 6:00, and 9:00 positions around the nipple [31,34,35,39]. Three studies by the same author measured the parameters 4 cm medial and lateral to the nipple [36][37][38]. Two reports obtained measurements from the irradiated breast and boost region [40,41]. Other locations were measured from nine points within the medial, central, and lateral areas of the breast [46], the border between the upper quadrants 2-3 cm above the mammilla [44], and the upper medial quadrant [43]. One study measured the irradiated breast without specifying the location [42]. Three studies reported measurements of the boost region to find any difference in skin findings with an additional radiation dose relative to the irradiated non-boosted breast [32,34,40,41]. All included studies used the same locations of measurements on the nonirradiated breast for comparison.

Ultrasonographic Skin Toxicity Changes
All studies reported skin toxicity changes of the irradiated breast documented by ultrasonography relative to the nonirradiated breast regardless of the fractionation schedule of RT and timing of skin reactions (early or late) (Tables 3 and 4, Figure 2). Most studies found significant differences (p < 0.05) in irradiated skin parameters compared to nonirradiated skin. Furthermore, all studies reported skin thickening, except a study by [46] which observed skin thinning with more than 1 year follow-up after postmastectomy RT attributed to fibrosis as part of chronic skin reactions. Three studies noticed the most significant difference in skin thickening at 4-6 months post RT [9,36,43]. However, differences during RT and early skin reactions did not reach statistical significance in two studies [36,37]. A study by [44] noticed no significant difference regarding dermis thickness between the early and late skin reactions when compared to each other. of RT and timing of skin reactions (early or late) (Tables 3 and 4, Figure 2). Most studies found significant differences (p < 0.05) in irradiated skin parameters compared to nonirradiated skin. Furthermore, all studies reported skin thickening, except a study by [46] which observed skin thinning with more than 1 year follow-up after postmastectomy RT attributed to fibrosis as part of chronic skin reactions. Three studies noticed the most significant difference in skin thickening at 4-6 months post RT [9,36,43]. However, differences during RT and early skin reactions did not reach statistical significance in two studies [36,37]. A study by [44] noticed no significant difference regarding dermis thickness between the early and late skin reactions when compared to each other.
(A) (B)   Early and late radiation-induced effects on normal tissue can be reliably assessed using the quantitative US (Keskikuru et al. 2004) [43] Significant changes in skin thickness (p < 0.05, p < 0.01 at different timepoints) Significant changes (p < 0.05, p < 0.01 at different timepoints) until one year then declined No significant correlations between the skin thickness and the score of erythema or subcutaneous induration Increased collagen synthesis is associated with oedema resulting from radiation-induced damage to skin microvasculature  [37] No obvious skin thickness changes during RT (p-value NR).
The most marked cutaneous thickness was in patients with obvious visible breast oedema before RT (p-value NR) HFUS is not an ideal, sensitive, and quantitative measure of acute RD in this group of patients (Wratten et al. 2007) [36] A minor ↑ in epidermal thickness (p-value NR) Significant changes (p = 0.000 with or without level 2 nodal dissection)

NR
The utility of HFUS is in a research setting when assessing interventions that aim to reduce breast oedema (Warszawski et al. 1998) [44] Significant changes in the dermis thickness and echogenicity (p < 0.001) Nonsignificant changes in the structure of the dermis-subcutis border (p = 0.07) Significant changes in the dermis thickness (p = 0.0018) and echogenicity (p < 0.001 for lower dermis, p = 0.0027 for upper dermis) Nonsignificant changes in the structure of the dermis-subcutis border (p = 0.08) There were discrepancies between the clinical and US assessments, mainly in the late reactions (K = −0. 13  • Entry echoes-no significant differences between nonirradiated and irradiated skin for early or late reactions. • Signal intensity-significant reduction of the signal intensity of the upper and lower corium in the early and late reactions but was more distinct in the early reactions.

•
Reduction of echogenicity-no significant difference between early and late reactions for the upper corium, but, for the lower corium, differences were significant (more distinct in the early reactions).

•
Border structure-no significant difference in the border structure between the dermis and subcutaneous tissue of the irradiated skin compared to the nonirradiated skin. The dermis-subcutaneous boundary was less well defined in the treated breast than in the untreated breast. • Decreased dermal density or echogenicity (not quantified).
Hypodermal damage is another RT-related toxicity assessed by four studies by measuring the PCC of the hypodermis from RF data [31,34,35,42]. These reports found a significant decrease in PCC in the early and late reactions, except [34], which found significant differences in the early but not the late reactions. They stated that healthy skin has higher PCC, and that fibrosis following RT will reduce the hypodermal integrity and decrease the PCC.
Several studies reported skin thickening prior RT attributable to the axillary lymph node dissection (ALND) as a part of the surgical treatment of breast cancer. The authors of [9,32] conducted two consecutive studies to evaluate the impact of ALND on breast skin thickening during and up to 1 year post RT. They found a persistent increase in skin thickening from baseline until 1 year follow-up after RT due to axillary surgery. Four studies reported that the medial aspect of the breast was thicker than the lateral aspect [36][37][38]46]. The three studies by Wratten et al. also found the medial aspect of the nonirradiated skin was thicker than the lateral aspect.
Only two studies compared the US skin toxicity changes between CF and HF. Despite reporting significant differences in CTCAE score at the end of RT, the authors of [39] found no significant difference in skin thickening at the end and 6 weeks post RT. They stated that the results might be attributable to the fact that most patients developed only mild radiodermatitis, and differences between HF and CF were too small to be detected by the US, whereas Wang et al. (2020) [33] also found no significant differences in STRA during, 12 weeks post, and 1 year post RT. Both studies supported the hypofractionated approach as having better patient-reported and cosmetic outcomes. Additionally, comparing the boosted and non-boosted regions of the breast, the authors of [40] reported no significant difference, and that additional RT dose to the tumour bed would not lead to more fibrosis and increased skin thickness. The authors of [41] found the same measurements, Torres et al. (2016) [32] did not report the measurements or the differences. Lastly, Yoshida et al. (2012) [34] observed a lack of consistency at the tumour bed because of poor visualisation at this site and recommended eliminating the tumour bed location.

Variables Associated with Skin Toxicity
Some reports used statistical analysis to analyse the predictors or variables associated with skin toxicity. The older age group was a significant predictor of increased skin changes at the end of RT relative to baseline [32]. On the contrary, age < 65 years was significantly associated with more severe skin toxicity on bivariate analysis [41]. Breast volume was a common predictor of a greater increase in skin thickening reported in three studies [32,33,41]. On the other hand, none of the studies that collected the BMI of the patients reported that obesity was a predictor of skin toxicity. Previous chemotherapy and concurrent endocrine treatment did not predict more skin changes at the end or 6 weeks post-RT [32]. At the same time, there was no association between breast retraction/cosmetic outcome and previous systemic therapies [33].
Current smoking is another variable that was a predictor for higher baseline STRA [32] but was not a predictor of STRA at 1 year [9]. Moreover, the Caucasian race was found to be a predictor at 1 year [9] but was not at week 6 post-RT [32]. In addition, no association was found between African American race and breast asymmetry post RT [33]. The RT boost technique did not predict more severe skin changes, whether electron boost at 1 year [9] or photon boost at 6 weeks post RT [32]. Supraclavicular nodal irradiation [33] and the time interval between surgery and RT [9] were also predictors for more severe skin changes 1 year post RT. Interestingly, Wratten et al. (2007) [36] found that the type of node dissection, nodal irradiation, and postoperative wound infection were the most important factors that influenced cutaneous oedema over time using GEE (generalized estimating equations) analysis. They noted that patients who did not have a level 2 node dissection, infection, or regional nodal irradiation demonstrated no increase in epidermal thickness throughout the entire study period.

Association of US Skin Changes with Clinical Assessments
A total of 14/17 studies used various clinical assessments to compare the US skin measurements with clinical evaluations or scales (Table 4). Generally, depending on the time of clinical and US assessments, all of these studies reported a variable degree of association with clinical assessments except for two studies [39,43]. The US skin measurements were higher for patients with more severe visible or palpable skin reactions (higher grades) than patients with mild or no reactions. The reports that compared the early US skin changes revealed that US skin measurements were more significant with increasing clinical grading [32,41] or obvious skin changes [37] than those with less skin changes. However, the authors of [39], when comparing HF and CF groups at the end of RT, reported significant differences in the CTCAE scores but no significant difference in the US changes or symptoms measured by the Skindex-16 questionnaire.
On the other hand, reports comparing late US skin changes found differences in the association pattern between parameters. Of these studies, the authors of [34,45,46] stated that US measurements were most marked with increasing toxicity grading. Wong et al. (2011) [46] used retrospective acute toxicity grading. A study by [40] found a significant direct correlation with higher grades, while Yoshida et al. (2011) [35] found that PCC correlated with RTOG, but skin thickness did not. In addition, they reported no correlation between US measurements and erythema/melanin indices measured by spectrophotometry. Another study from the same group reported that skin thickness correlated with RTOG late subcutaneous toxicity, and PCC correlated with late skin toxicity [31].
Of the studies that compared both early and late skin changes, Wang et al. (2020) [33] documented a significant association between STRA and breast asymmetry or retraction measured by percentage breast retraction assessment (pBRA), while Wratten et al. (2000) [38] found that the most significant thickness was in patients with more prominent visible breast oedema. On the contrary, Keskikuru et al. (2004) [43] did not report any significant correlation in acute and chronic changes. Instead, they found a significant correlation between skin thickness and procollagens (PINP and PIIINP) measured from suction blister fluid of the irradiated skin. They assumed radiation-induced oedema manifested as skin thickening is associated with increased collagen synthesis. A study by [44] noted discrepancies between US changes and RTOG grading in the late reactions, but ultrasonic evaluation could record the structural changes in the early skin reactions much earlier than visible reactions by the naked eye.

US Reliability/Reproducibility
The reliability (intra-and interobserver reliability) of the US was assessed by only one study for evaluating the dermal and hypodermal toxicities from RT using the intraclass correlation coefficient (ICC) [34]. They found that the dermal toxicity parameter was highly reliable (high ICC) while the hypodermal toxicity parameter was moderately reliable. Moreover, only one study evaluated the reproducibility of the US measurements by three operators [32]. They observed no significant inter-or intra-operator differences between measurements compared to the healthy breast at all time points. This finding was a basis for the follow-up study by [9].

Discussion
Our interest in the present study is to systematically review the skin toxicity changes following breast cancer radiotherapy using ultrasonography. The reports revealed ultrasonographic skin toxicity changes in the irradiated breast compared to the nonirradiated breast. This study is the first systematic review summarising the available evidence for evaluating skin toxicity following breast radiotherapy using ultrasonography. In general, the results of this review demonstrated significant skin toxicity changes during and after radiation, even several years after treatment, relative to the untreated or pre-RT breast measurements. However, nonsignificant changes during and shortly after RT were reported in two studies by the same author.
Despite heterogeneity in the parameters tested and locations imaged, skin thickening was the consistent finding across the studies except one that reported skin thinning after 1 year of pure post-mastectomy RT. The oblique incident angle and flat chest wall may be responsible for increasing the RT dose delivered to the skin leading to thinning, which may explain the increased breast reconstruction complications in postmastectomy patients receiving RT [46]. Further studies with longer follow-ups are needed to document this unusual finding after postmastectomy RT. Radiation-induced skin thickening is somewhat attributed to radiation damage to skin microvasculature resulting in ischaemia and oedema [43]. Nevertheless, a considerable number of studies reported skin thickening before RT as a result of axillary surgery. The axillary surgery disrupts the lymphatic circulation, resulting in lymphatic fluid accumulation, oedema, and breast skin thickening before RT. At the same time, radiation-induced oedema cannot decompress in a patient with disrupted lymphatics secondary to surgery; this increases skin thickening with short [32] and long-term follow-up after RT [9]. Similarly, when the lymphatic drainage of the breast is compromised by surgery, irradiation, or even postoperative wound infection, RT will aggravate the oedematous skin changes and thickening and exert a synergistic effect [9,32,36]. With a longer duration between surgery and RT, there is more time to develop fibrosis from surgery resulting in more severe skin thickening [9]. Axillary irradiation may be a better alternative to ALND as a treatment approach to positive axillary lymph nodes to reduce skin thickening [32]. For future studies, baseline US skin assessment and optimal subgrouping between patients with or without axillary surgery are strongly recommended to enable better quantifying the magnitude of change attributed to RT and allow appropriate comparison.
Another interesting observation in this review is that the medial aspect of the irradiated and even the nonirradiated breast was thicker than the lateral aspect. This can be attributed to the lymph drainage from the medial parts is predominantly through the axilla, while some drainage of the untreated breast also occurs through the axilla of the treated breast. ALND will result in more oedema and increased thickening on the medial side [36]. However, it was reported by just four studies, three of which were by the same author. Future work should document this finding by measuring the same points to assess changes over time.
Limited studies have evaluated other US skin toxicity parameters such as echogenicity and signal intensity of the dermis, entry echo, and visibility of the dermis-subcutaneous fat interface. These studies reported decreased dermal echogenicity and were most distinct in the early skin reactions [44]. However, the echogenicity depends on several factors such as the thickness of the tissue between the transducer and measured point, the echogenicity of the tissues lying at, superficial, and deep to that point, and importantly on the gain setting of the US machine used [38]. Therefore, all these factors should be considered for accurate measurements of dermal echogenicity. Additionally, they identified poor visibility of the dermis-subcutaneous fat boundary, although nonsignificant differences between irradiated and nonirradiated skin were observed by [44]. The increasing gain setting will overcome this boundary's poor visibility to provide accurate skin thickness measurements [38]. We are undergoing a prospective cohort study to evaluate these parameters; hopefully, we can contribute further evidence for evaluating this common and distressing side-effect of RT.
Moreover, minimal studies assessed the skin toxicity by the US from RT techniques other than conventional fractionation-whole breast irradiation(CF-WBI), which has been blamed for a higher level of toxicity concerning other newer techniques. This shows that most studies were published more than 5 years ago. This issue may be explained that the CF-WBI was the standard radiation schedule at that time that is associated with high radiation doses and damage to the normal tissues in the treated field. To date, only three studies evaluated skin toxicity following HF ultrasonographically, two of which compared CF and HF [33,39]. Both reports supported that HF has better early and late patient-reported outcomes. In particular, as the application of hypofractionation increases, more new RT protocols are being tested in adjuvant WBI prospective trials, aiming for fewer side-effects and shorter treatment time to decrease the burden on breast cancer patients. Therefore, it could be essential to have a quantitative, feasible, and reproducible tool for assessing skin reactions not susceptible to intra-and interobserver variation in adjunct to the physical examination. Thus, more studies are needed to evaluate skin toxicity from emerging techniques such as hypofractionation, partial breast irradiation, and Mammosite.
There were limited observations regarding the skin finding differences between boosted and non-boosted regions of the breast. However, this review shows that a boost dose to the lumpectomy cavity does not contribute to more skin toxicity changes observed ultrasonographically. This finding supports the evidence that boost dose has no to limited impact on long-term cosmetic outcomes [15,18]. Nevertheless, we recommend considering these observations with future work to identify whether adding further RT dose leads to more toxicity changes. At the same time, we encourage assessment of the effect of the type of boost treatment, electron or photon, in separate studies on skin toxicity changes by ultrasound.
Several variables have been studied in the literature, including patient, tumour, and treatment-related factors that predict or associate with increased skin toxicity. These factors appeared to have a greater effect on aggravating or increasing skin changes. Consequently, poor cosmetic outcomes might result with more severe skin changes. A study by [36] found that even RT did not induce skin thickening measured with the US without axillary surgery, irradiation, or postoperative wound infection. This is a very significant observation that needs further investigation. In our review, limited studies assessed or controlled these variables. Across these studies, breast volume was the constant patient-related factor linked to enhanced skin toxicity from RT. Patients with large breast volume have a higher percentage of adipose tissue within the breast that will be more susceptible to RT toxic effects leading to more skin toxicity changes [33]. Other variables studied in our review such as age, smoking, BMI, race, systemic therapy, the time interval between surgery and RT, and nodal irradiation fluctuate in their association with skin toxicity. Further studies are necessary to confirm and control the effects of these predictors.
Despite the subjectivity of the clinical assessments and scoring scales, they are still the commonest toxicity evaluation during and following RT. Comparison with clinical assessments should be considered for any objective/quantitative technique [34]. Most studies reported that the US skin changes were consistent or associated with clinical assessments regardless of the time of evaluations. A significant correlation with procollagens (PINP and PIIINP) measured from suction blister fluid of the irradiated skin is further evidence of the ability of the US to measure skin toxicity changes accurately [43]. Yet, across the studies, the strength of the association (the use of p-value or Rho factor) has not been studied well to reach a definitive conclusion. US can also detect skin changes earlier than or even not detected by the naked eye [38,44], allowing for earlier detection or prediction of skin toxicity. Moreover, it may also become useful for assessing new interventions that reduce skin toxicity.
It is essential to consider some issues in the study methodology. First, there was significant variation in sample size across the studies, with six having small sample sizes (<20), which may have affected the overall significance of the skin changes. Second, there was some overlap of patients between studies, especially those assessed by the same authors or centres. This may have reduced the total number of patients evaluated ultrasonographically to reach an accurate conclusion about US skin changes. We are hopeful to see more publications with large cohorts and different centres in the future. Third, it is noteworthy that different ultrasound machines used and inadequate probe frequency (<18 MHz) for skin evaluation utilised by many studies may have contributed to some variability. In addition, the timeframe of assessments varied widely from actual RT sessions to 135 months after RT except for one patient evaluated 22 years post RT. This may have affected time-dependent changes, as noted by [44], which stated that US skin changes depend on the time interval between completion of RT and US assessment. The reliability and reproducibility of the ultrasound measurements were only investigated in two studies. Furthermore, none of the US assessors were blinded to the patient's radiation exposure, although some studies reported blinding to the clinical grading or the patient treatment characteristics. These factors may have given rise to some limitations or biases.
The ultrasound examination is generally objective, feasible, safe, inexpensive, and widely available. Ultrasound may provide useful development in the noninvasive assessment of RT-related skin toxicity in clinical practice and research settings. However, Wratten et al. (2007) [36] described the use of HFUS mainly in a research setting when assessing interventions that aim to reduce breast oedema, while Schack et al. (2016) [45] stated that HFUS evaluation of the skin is not considered part of large-scale follow-up routines in assessing radiation-induced morbidity. Emerging ultrasound-based techniques, such as elastography, may provide more accurate and objective features to ultrasound B-mode by measuring skin elasticity. It has been measured in different skin diseases [47,48] and gives valuable addition to the US evaluation.

Conclusions
Skin toxicity post radiotherapy treatment includes skin thickening, less echogenic dermis, a poorly visible dermis-subcutaneous fat boundary, and decreased PCC of the hypodermis compared to the nonirradiated skin. However, further studies with large cohorts and appropriate methodology are encouraged. In addition, future work on measuring other US toxicity parameters is warranted. Furthermore, US evaluation of skin toxicity from newer RT protocols, taking baseline measures, and further grouping patients with risk factors for skin toxicity will provide a more comprehensive assessment of the effect of RT on skin toxicity. Lastly, measuring skin elasticity by ultrasound elastography will further support the ability of the US to measure skin toxicity changes.
Supplementary Materials: The following supporting information can be downloaded at https: //www.mdpi.com/article/10.3390/ijerph192013439/s1: Table S1. Search terms used in the search strategy. Table S2. Quality check.

Conflicts of Interest:
The authors declare no conflict of interest.