Ultrasound: Assessment of breast dermal thickness: Reliability, responsiveness to change, and relationship to patient‐reported outcomes

Abstract

Background

The current study assessed the level of reliability of ultrasound to assess dermal thickness, a clinical feature of breast lymphedema. Additionally, the relationship of dermal thickness to patient‐reported outcomes was investigated.

Methods

Women (n = 82) with unilateral breast edema secondary to treatment of breast cancer were randomized to an exercise or control group. Ultrasound measurements of the unaffected and affected breasts were taken at baseline and 12 weeks later at 3–4 cm superior, medial, inferior, and lateral to the nipple. Additionally, women completed breast‐related questions from the European Organization Research and Treatment Committee Quality of Life breast cancer module (EORTC‐BR23) and Lymphedema Symptom Intensity and Distress Questionnaire (LSIDS). Reliability of ultrasound measurements was determined on the unaffected breast.

Results

Intraclass correlation coefficients (2,1) ranged from 0.66 (95% CI: 0.52–0.77) for the lateral location to 0.84 (0.77–0.90) for the superior location. Percent close agreement (80%) on the unaffected breast ranged from 0.20 to 0.27 mm compared to 0.57 to 0.93 mm on the affected breast. The standard error of measurement (%) on the unaffected breast varied from 9% to 13% with smallest real difference 0.34–0.41 mm. Dermal thickness of the affected breast was not‐to‐poorly associated with EORTC BR23 and LSIDS scores.

Conclusion

Reliability of dermal thickness measurements of the breast was excellent for the superior, medial, and inferior locations, and fair to good for the lateral location. However, these measurements were not related to the symptom's women perceive and measured with the EORTC BR23 or LSIDS.

Lymphedema is associated with localized changes to the surrounding tissue. ¹ , ² , ³ One of these changes is the increase in the dermis and subdermis layers. ⁴ As a consequence of loss of contractile function, vessel dilation, and incompetent valves in the precollectors of the superficial lymphatics, lymph is not propelled through the lymphatics. ³ Rather, dermal backflow occurs in an attempt to drain the affected regions. Consequences of dermal backflow include skin thickening and induration in the affected area. ⁴ , ⁵ , ⁶ , ⁷ Other tissue changes include proliferation of adipose tissue in the subcutaneous fat layer, and as this disease progresses, some affected regions become hardened and are referred to as fibrotic. ² , ⁵ Lymphedema is regional and following treatment for breast surgery, can affect the upper limb and/or chest and breast region. ⁸ To capture the impact of this disease, a range of reliable tools are required.

Signs and symptoms of breast lymphedema include peau d'orange, redness of the breast, pain in the breast, positive pitting test, increased volume, heaviness of the breast, and hyperpigmented pores. ⁹ Symptoms of breast edema can be assessed using standardized questions such as those from the breast module of the European Organization Research and Treatment Quality of Life (EORTC‐QoL) questionnaire or the Lymphedema Symptom Intensity and Distress Survey (LSIDS)–Truncal and Head and Neck. ¹⁰ However, the quantification of physical changes is challenging. Typically, lymphedema in the breast occurs secondary to a wide local excision in which a varying amount of breast is removed. Volume‐based measures, derived from tape measure measurements, are available but their reliability and sensitivity to change from edema is questionable. ¹¹ , ¹² An alternative approach to assess edema arising from treatment of breast cancer is to measure the dermal thickness using ultrasound.

Ultrasound has been used in several studies to identify edematous breasts; ¹³ , ¹⁴ , ¹⁵ however, while the reliability of measuring the image has been investigated, ⁴ the level of reliability of the process in capturing the images is unknown. This is an important step as it informs minimal important difference and can be used to assess responsiveness to treatment.

The aim of this study was, therefore, to determine the level of reliability of the process in which dermal thickness of breast tissue is measured with ultrasound in women in whom one breast was affected by edema. Second, we will determine what is a clinically important change with respect to the dermis in the breast. Third, we will explore the relationship between dermal thickness and patient‐reported outcomes (PROs), including pain and swelling in the breast.

1. MATERIALS AND METHODS

This was a secondary analysis of a prospective randomized‐controlled study investigating the responsiveness of breast edema to a 12‐week exercise program comprising aerobic and resistance strengthening, in which the findings of exercise are reported elsewhere. ¹⁶ The study was registered with the Australian Clinical Trials Registry (ACTRN12612000771853). Ethics approval was obtained from Human Ethics Research committees at each institution at which the study was conducted and all participants provided written informed consent.

1.1. Participants

Participants were recruited through a range of strategies, including advertisements in cancer‐related newsletters, cancer registries, and from cancer radiography departments. Inclusion criteria included being ≥18 years old, surgery for breast cancer that included a wide local excision and axillary surgery, and stable breast edema ≥3 months with symptoms of heaviness and/or discomfort > 3 on a 10‐cm visual analog scale at the time of enrollment.

Women were excluded if they: had received intensive treatment for breast edema within the past 3 months; had experienced infection within the lymphedematous region requiring antibiotics; presented with primary lymphedema, had a pacemaker; had undergone bilaterally axillary surgeries; were already engaged in moderate‐to‐vigorous intensity exercise; were pregnant or breast‐feeding; or had any condition with which exclude them from participating in an exercise study, such as a musculoskeletal or neurological condition that would affect training or assessments.

Women who met the inclusion criteria underwent further screening, and for some, medical assessment, to ensure they were safe to participate in an exercise trial.

1.2. Protocol

Women were stratified by location of lymphedema (breast only vs. arm + breast) and then randomized 1:1 to an exercise or usual care group. Randomization was computer generated in blocks of 4, 6, or 8 by a person not in contact with participants, with the allocation concealed in an opaque numbered envelope opened following baseline assessments.

1.3. Interventions

The details of the interventions and monitoring are described elsewhere. ¹⁶ In brief, women allocated to the exercise program attended a gym convenient to them three times per week for 12 weeks where they undertook resistance and aerobic training. Sessions were approximately 1‐h duration. To facilitate adherence to the exercise program, women were provided with a 3‐month gym membership and supported by an Accredited Exercise Physiologist (AEP) trained by the trial AEP to deliver the intervention.

Women in the usual care group were not prescribed any exercises.

Women in both groups completed brief weekly questionnaires to monitor any changes to their lymphedema status. In addition, they attended monthly sessions in which their arms and breasts were physically assessed.

1.4. Measurements

All data were collected by a research assistant blinded to group allocation. At baseline, details regarding treatment for breast cancer and lymphedema were obtained, as well as age, height, and weight for determination of body mass index (BMI). In addition, physical measures and PROs related to breast edema were collected.

Dermal thickness was quantified on both the affected and unaffected breasts with ultrasound (Esaote MyLab™25Gold, Esaote, Italy) at 35–40 mm superior, inferior, medial, and lateral to the nipple. For measurements with the ultrasound, women were positioned in supine, with their arms resting above their heads. The location at which the ultrasound measures were to be obtained was marked, the distance recorded relative to the nipple, and a photograph of the locations from the front obtained. Three ultrasound images at each location were obtained.

In addition, PROs related to breast edema were captured with the EORTC Breast (BR23) module ¹⁷ , ¹⁸ and from the Lymphedema Symptom Intensity and Distress Survey (LSIDS) in which women were asked whether symptoms were present in the breast or chest wall rather than the arm. ¹⁹

The EORTC‐BR23 comprises four questions related to breast symptoms: questions specifically asked the extent to which the breast on the affected side was painful, swollen, oversensitive, or had skin problems in the past week. Responders were asked to the extent to which the breast was affected, ranging from “not at all” to “very much.”

The LSIDS asked women to indicate whether a range of symptoms were or were not present, and if present, the severity of the symptom and the distress with which the symptom caused the woman, on 1–10 scale, with 1 representing “slight” and 10 representing “severe.” Only those items in which > 50% of women rated an item ≥1 at baseline were included in the analysis. As such, only the items from the questions that asked about the intensity of heaviness, tightness, pain, warmth, achiness, swelling, and hardness in the breast were retained for the analysis.

1.5. Data analysis

Descriptive statistics were used to describe the participant characteristics and dermal thickness of the breast.

Intraclass correlation coefficients (ICC_2,1; absolute agreement) of dermal thickness were calculated for the dermal thickness for each of the four locations of the unaffected breast only. ICCs_2,1 were interpreted, based on Fleiss's recommendations, that ICC values of > 0.75 represent “excellent reliability” and values between 0.40 and 0.75 represent “fair to good reliability.” ²⁰ As the data were relatively homogenous, other analyses were undertaken including: (1) assessment of changes in the dermal thickness between measurement timepoints with graphing of data using Bland–Altman plots; (2) 80% close agreement of dermal thickness measurements at each of the four locations of the unaffected breast between measurement timepoints; and (3) measurement variability using standard error of measurement (SEM; SEM = √(σ² _observations + σ² _residual). ²¹ In addition, smallest real differences for each of the four locations were determined, using the formula (1.96 * SEM*√2). ²¹ , ²² The smallest real differences were used to examine the responsiveness of dermal thickness in the affected breast.

To assess the relationship between PROs and dermal thickness, the maximal dermal thickness on the affected breast was identified, irrespective of location. Spearman's correlation was then used to assess the relationship between the maximal dermal thickness at baseline and at 12 weeks with the EORTC scores for breast pain and swelling at each of the two timepoints. Similarly, Spearman's correlation coefficients were used to assess the relationships between the maximal dermal thickness at each timepoint with the respective scores from the LSIDS for items related to heaviness, tightness, pain, warmth, achiness, swelling, and hardness as well as the maximal intensity score from these variables. The strength of the relationships from the Spearman's correlation coefficients was interpreted as follows: < 0.10 as none; 0.10–0.29 as poor; 0.30–0.59 as fair; 0.6–0.79 as moderate; and 0.80–0.99 as very strong. ²³

Mean and SD are reported, unless otherwise specified. All calculations were undertaken with IBM SPSS Vs 26.

2. RESULTS

Eighty‐eight women enrolled in the trial, with pre and post data for 82 participants. Participants were aged 56.8 years old (Table 1). Participants’ weight did not change over the 12‐week period; the mean (95% CI) difference was 0.1 kg (−0.5 to 0.6 kg).

TABLE 1.

Participant characteristics at baseline

Variable	Participants (n = 88)
Age	56.8 ± 9.6 years
Height	1.63 ±0.6 m
Weight	77.8 ± 16.2 kg
Radiotherapy	87 (99%)
Chemotherapy	58 (66%)
Time since surgery (median [IQR])	15 (11–26)

Abbreviation: IQR, interquartile range.

Dermal thickness, assessed with ultrasound, is reported in Table 2. For each location, dermal thickness was significantly greater on the affected side compared to the unaffected side. The location at which the dermal thickness was greatest was determined for each individual. At baseline, the dermal thickness on the affected side was most commonly thickest at the inferior location (n = 43), followed by the medial location (n = 35), lateral location (n = 6), and lastly, the superior location (n = 3). At 12 weeks, dermal thickness was most commonly greatest in the inferior and medial locations (n = 38, respectively), followed by the lateral location (n = 5) and superior location (n = 2). Over the 12‐week period, the location with the greatest thickness in the affected breast changed in 33% of women.

TABLE 2.

Baseline and 12 weeks measures of dermal thickness (mm) using ultrasound

	Affected breast measures (mm)			Unaffected breast measures (mm)
	Baseline	12 weeks	Absolute difference 95% CI	Baseline	12 weeks	Absolute difference (95% CI)
Superior	2.09 (0.78)	1.93 (0.71)	0.16 (0.36) 0.08–0.24	1.29 (0.28)	1.32 (0.31)	−0.03 (0.17) −0.06 to 0.01
Medial	3.13 (1.37)	2.92 (1.34)	0.20 (0.57) 0.08–0.33	1.40 (0.29)	1.40 (0.29)	0.00 (0.21) −0.04 to 0.05
Inferior	3.22 (1.44)	2.89 (1.38)	0.33 (0.67) 0.18–0.48	1.33 (0.31)	1.34 (0.28)	−0.01 (0.21) −0.06 to 0.04
Lateral	2.32 (1.13)	2.04 (1.07)	0.28 (0.64) 0.14–0.43	1.09 (0.26)	1.06 (0.22)	0.03 (0.20) −0.01 to 0.08

Note: Mean and standard deviations reported.

2.1. Reliability and SEM of breast ultrasound

Dermal thickness did not significantly change for the unaffected breast between baseline and Week 12 (Table 2), with paired t‐tests revealing no significant differences between the two measurement occasions, and 95% confidence interval of the differences spanning zero (Table 2). In addition, there was no bias in the difference in the measurements occurring from the baseline to Week 12 measurements for the unaffected breast, with bias ranging from 0.1% to 2.9% for the four locations (Figure 1). In contrast, baseline dermal thickness measures of the affected breast were significantly higher than the Week 12 measurements for each of the four locations, with 95% confidence intervals not crossing zero (Table 2). Bias of the difference of scores between the two measurement timepoints for the affected breast ranged from 6.5% to 12.3% (Figure 1).

FIGURE 1.

Bland–Altman plots for measurements of dermal thickness using ultrasound for the unaffected (left) and affected (right) breast regions

ICCs_2,1 were used to evaluate the reliability of measurements of dermal thickness on the unaffected breast. ICCs ranged from “fair to good” for the lateral breast region to “excellent” for the other three regions (Table 3).

TABLE 3.

ICC_(2,1), SEM, SEM (%), and smallest real difference, based on the change score of measurements of dermal thickness of the unaffected breast

Location relative to areola	Participants (n)	ICC (2,1) (95% CI)	SEM (mm)	SEM (%)	Smallest real difference (mm)
Superior	81	0.84 (0.77–0.90)	0.122	9.3	0.34
Medial	82	0.76 (0.65–0.84)	0.148	10.6	0.41
Inferior	82	0.77 (0.66–0.84)	0.148	11.1	0.41
Lateral	81	0.66 (0.52–0.77)	0.141	13.1	0.39

As illustrated in Figure 1, data for the unaffected breast were highly homogenous. To further examine the extent to which data were similar at the two measurement periods, percent close agreement of the two measurement occasions was examined (Figure 2). On the unaffected side, 80% agreement of the baseline and 12 weeks measurements were within 0.20 mm superiorly, 0.23 mm medially, 0.27 mm inferiorly, and 0.20 mm laterally. In contrast, on the affected side, the 80% agreement of dermal thickness measures were within 0.47 mm superiorly, 0.71 mm medially, 0.93 mm inferiorly, and 0.57 mm laterally.

FIGURE 2.

Percent close agreement for dermal thickness measures for each breast location. Unaffected breast measurements (solid black line), affected breast measurements (dotted black line); 80% close agreement (gray short dash line)

Another indication of variability for dermal thickness on the unaffected side is the SEM which can both be expressed in absolute terms and as a percentage. The SEM ranged from 0.122 to 0.148 mm, whereas the SEM (%) indicated that the dermal thickness measurement varied from one test occasion to another by 9–13%, depending from which location the measurement is taken (Table 3).

The smallest real differences are used to ascertain whether the change is clinically important. Based on the variability of assessments of the unaffected breast, smallest real differences ranged from 0.34 mm for the superior location where variability was least, and 0.41 mm for the medial and inferior locations.

The smallest real differences were then applied to the changes identified on the affected side to determine whether the changes in dermal thickness exceeded the smallest real difference positively or negatively over 12 weeks, based on whether women were allocated to the exercise or usual care groups. No significant differences in distribution of responses were identified between the exercise and usual care groups (Figure 3).

FIGURE 3.

Responsiveness of affected breast dermal thickness for the exercise group (n = 39) and usual care group (n = 43). Change in dermal thickness categorized as increased; no change; or decreased based on whether the change exceeded the smallest real difference

2.2. Relationship between dermal thickness and breast symptoms

The relationships between the maximal dermal thickness of the affected breast with perception of swelling and heaviness, assessed with EORTC‐BR23, were investigated. Dermal thickness had none to poor associations with perceptions of pain and swelling at either baseline (r = −0.01 and r = 0.17, respectively) or 12 weeks (r = −0.19, r = 0.18, respectively).

The relationship between maximal dermal thickness of the affected breast and the symptoms assessed with the LSIDS was also assessed. At baseline, the correlation coefficients between the maximal dermal thickness with the individual participant's score ranged from −0.02 to 0.30 indicating that the relationship was nonexistent for heaviness, tightness, and swelling (r < 0.10); poor for warmth, achiness, and hardness (r = 0.11–0.22); and fair for pain (r = 0.30). Similarly, at 12 weeks, the correlation coefficients for individual symptoms ranged from r = 0.01 for tightness to r = 0.28 for pain in the breast, indicating that the relationships ranged from none to poor. Correlation of the maximal intensity score at baseline and at 12 weeks with the respective maximal dermal thickness was also poorly correlated (r = –0.13, r = 0.02).

3. DISCUSSION

Dermal thickness, an indicator of breast lymphedema, can be quantified using ultrasound. While criteria for detection of swelling using ultrasound have been reported previously, reliability of the image capture process and smallest real difference to inform treatments was lacking. The current study addressed these gaps. We established that reliability of measurement of the breast using ultrasound adhering to a standardized protocol was excellent for the superior, medial, and inferior locations, and “fair to good” for the lateral location. The measurements taken 12 weeks apart on the unaffected breast varied by 9–13%. The smallest real difference was 0.34 mm for the superior location, 0.39 mm for the lateral location, and 0.41 for the medial and inferior locations of the breast. However, dermal thickness was not‐to‐poorly associated with PROs, captured with EORTC‐BR23 and LSIDS.

Previous studies that have investigated dermal thickness of different regions of the body highlight the need to assess the region of interest as it varies, depending on site. ²⁴ To date, only the reproducibility of measurements of the ultrasonic images of the breast has been determined. ⁴ The current study added to the body of evidence by evaluating the reproducibility of the process, which includes measurement of the image. Ultrasound measurements of the four locations of the unaffected breast were used to determine to what extent these measures are reliable. Two issues that may impact reproducibility are the impact of radiotherapy and potential impact of change in body weight. Women enrolled in the current study were 3–293 months following breast surgery, with the median (IQR) time since surgery 15 (11–26) months. Although all women underwent radiotherapy, it is highly unlikely that this affected dermal thickness on the unaffected side. Previous studies have demonstrated no change in dermal thickness from preradiotherapy measurements to postradiotherapy measurements on the unaffected side following radiotherapy. ²⁵ The other issue that could potentially affect dermal thickness is body weight. ²⁶ Dermal thickness of the ventral and dorsal forearm increased significantly with increase in body weight and concomitant BMI. ²⁶ However, the change in body weight was less than 0.1 kg in the current study and so would not have impacted the dermal thickness of the unaffected breast. Thus, while 12 weeks is not typically used as the interim to assess reliability of measurements, there was no reason to assume that the time period would affect the measurements.

The ICC_2,1 ranged from fair to excellent, although we previously demonstrated high reliability for measurement of the ultrasound images. ⁴ A couple of reasons may explain these differences. First, as shown in the Bland–Altman plots, data on the unaffected side were quite homogeneous. Other measures of variability, such as the absolute and relative standard error of measurement, are therefore required. ²² Findings from the current study indicate that there was approximately 10% variation between two measurement occasions. Ideally, inclusion of measurements of the unaffected and affected breast would reduce the homogeneity of the data and improve the intraclass coefficient. Indeed, the ICC_2,1 for the lateral location was 0.87 (95% CI 0.82–0.91) and 0.94 (0.92–0.96) for the medial location when data from the unaffected breast and affected breast in those in the control group were included. However, this approach was not used as the dermal thickness of the affected breast may have changed in the 12‐week time period between the two‐measurement occasions. Second, previous studies have marked the skin and replaced the probe on the marked skin in order to ensure that measurements were undertaken at the identical location. ²⁷ Others have examined the reliability of dermal thickness measurements taken by different examiners within the same session ²⁸ or same day. ²⁷ While marks on the skin were not retained for the 12 weeks, we did use photographs and recorded distance to the areola to reduce error in measurements. Third, the probe which we used had a frequency of 18–6 MHz. While it is acceptable and considered high resolution, new ultrasonic machines with higher frequency probes will provide higher resolution, particularly in the superficial tissue layers. ²⁹

To date, many studies on breast edema using ultrasound are cross‐sectional, describing the presence of increased dermal thickness. ¹⁴ , ³⁰ , ³¹ , ³² , ³³ , ³⁴ The dermal thickness on the unaffected side is, on average, around 1.5 mm, with slight variation among studies. In contrast, for those with breast edema, the dermal thickness typically exceeds 2.0 mm. What has yet to be considered is what constitutes a clinically significant change, that is, smallest real difference, ²² in order to evaluate the efficacy of treatment. Based on our data, a change in dermal thickness exceeding 0.34 to 0.41 mm, depending on location, should be considered noteworthy. These values can now be used to classify whether a significant change has occurred, either positively or negatively. In addition, these now provide useful criteria against which responsiveness for questionnaires can be benchmarked.

While dermal thickness provides a physical measure related to lymphedema, it did not explain the symptoms which women report. Change in dermal thickness also likely does not relate strongly to change in overall volume, as highlighted by an investigation that examined the relationship between change in arm volume with change in dermal thickness in women with lymphedematous upper limbs following intensive decongestive therapy: the relationship was only fair (r = 0.37). ³⁵ Clinically, therefore, it will be important to capture both the symptoms experienced by women and measures of dermal thickness.

In conclusion, the level of reliability of using ultrasound with a standardized protocol for assessment of breast edema is fair to excellent, depending on location. Determination of the smallest real difference will facilitate evaluation of interventions that target breast lymphedema. Last, the lack of strong relationship between dermal thickness and PROs confirms the need to include PROs in the assessment of interventions.

AUTHOR CONTRIBUTIONS

SLK contributed to the concept, design, data analysis, and writing. NF contributed to the design, method, and writing review. ESD contributed to the concept, design, method, and writing review.

Kilbreath SL, Fearn NR, Dylke, ES . Ultrasound: Assessment of breast dermal thickness: Reliability, responsiveness to change, and relationship to patient‐reported outcomes. Skin Res Technol. 2022;28:111–118. 10.1111/srt.13100