Cross-Sectional Areas of Calf Soft Tissue Measured with MRI as a New Method for Staging Gynecologic Oncology-Related Extremity Lymphedema

Objectives: To determine whether the cross-sectional area (CSA) of the soft tissue of the calf measured with magnetic resonance imaging (MRI) could stage unilateral lower extremity lymphedema (LEL) secondary to gynecological cancer treatments. Materials and Methods: One hundred and twenty females with unilateral LEL and Thirty-two without LEL after gynecological cancer treatments underwent calf MRI and water displacement. Total soft tissue CSA (T), muscle CSA (M) and subcutaneous tissue CSA (S) of affected calf, and difference of T (DT), M (DM) and S (DS) between calves were obtained on MRI at mid-calf level. Volume of affected calf and difference of volume (DV) between calves were obtained by water displacement. Statistical analysis was performed to determine feasibility of MRI measurements for accessing LEL. Results: There were close correlations between volume and T or S of affected calf, and between DV and DT or DS of calves. The correlations of stages of LEL with T and S of affected calf as well as DT and DS were stronger than with the volume of affected calf and DV (all p < 0.01). Multivariate analysis showed more significant differences between LEL stages in T and S than in volume of affected calf, and in DS than in DV (all p < 0.05). No difference was found between stage 0 and 1 in volume of affected calf and in DV. For staging LEL, DS showed the most profound discrimination ability among all measurable parameters. Conclusions: DS of calves could be the most reliable parameter recommended for staging unilateral LEL.


Introduction
Lower extremity lymphedema (LEL) is a severe complication following gynecological cancer surgery with pelvic lymphadenectomy, occurring in 1% to 49% of cancer survivors (1,2).LEL can be defined as a chronic swelling of the lower extremity and is primarily caused by an impairment of lymph drainage.Since LEL is progressive, the diagnosis at the earliest stage (stage 1) is essential for preventing the progression of the disease and its complications because late stages of lymphedema may cause severe physical and psychological problems for patients owing to chronic swelling, impaired physical function, recurrent infections and disfiguring skin changes (3)(4)(5).
Although LEL is typically determined by clinical history and physical examination, however, objective assessment of LEL is complicated in current clinical settings due to a lack of reliable and sensitive methods (6).Volume measurement acquired by water displacement is the widely accepted gold standard method to assess LEL (7,8).Due to no accepted standard of the definition of LEL regarding changes in volume, the previously reported incidence of LEL varies widely.In order to detect lymphedema in patients, physicians rely on the widely accepted value of a 20% increase in volume of an affected limb as compared with that of an unaffected limb as a probable positive test for lymphedema, thus the early stage cannot be recognized in some patients with the disease (9,10).Since water displacement cannot distinguish the changes of extremity volume caused by subcutaneous tissue or muscle tissue, researchers are searching for a more feasible and informative method for monitoring and identifying LEL in early stage.
It has recently been reported that the increase of volume of lower extremity with lymphedema is significantly associated with thickened subcutaneous tissue and muscle, and can be confirmed through imaging modalities, including CT scans, ultrasound and MRI (11)(12)(13)(14)(15)(16)(17)(18).High quality soft tissue contrast on MRI imaging allows characterizing the pathological changes within subcutaneous tissue including circumferential edema, accumulated fluid, and a honeycomb pattern caused by soft tissues fibrosis and adipose hypertrophy (17,19,20).Previously, we reported that subcutaneous tissue thickness of lower extremity measured on T2-weighted MRI is a valid instrument for the assessment of LEL and could be used for staging LEL (21,22).Considering the shape of the lower extremities is irregular in patients with LEL, a soft tissue thickness measurement in one direction could not represent the size change of whole cross section of the extremity.Currently, there has not been any report that focus on using MRI to measure soft tissue cross-sectional area (CSA) of lower extremities to assess LEL secondary to cervical and endometrial cancer treatments.Considering the accepted pathophysiology of lymphedema asserting that LEL initially affects the calf and then extends to the thigh due to gravity, we speculate if assessing calves with an MRI might be helpful in early detection of this disease (23).Therefore, the purpose of this current study is to evaluate whether and how soft tissue CSA of calf measured at MRI could be used as a new indicator for assessing patients with unilateral LEL.

Participants
Our institutional review board approved this prospective study.Written informed consent was obtained from each participant prior to the study.

Clinical document
Stage 0 A latent or sub-clinical condition is not evident despite impaired lymph transport.

Stage 1
Early accumulation of fluid with relatively high protein content subsides with extremity elevation.Pitting may occur.

Stage 2
Extremity elevation alone rarely reduces tissue swelling, pitting is manifested, and extremity may or may not pit as excess fat and fibrosis supervenes.

Stage 3
Lymphostatic elephantiasis where pitting can be absent, trophic skin changes including acanthosis, further deposition of fat and fibrosis and warty overgrowths have developed.※ represents that the 32 volunteers without lower extremity lymphedema after cervical and endometrial cancer treatments have been classified as stage 0, and 16 participants have been randomly selected as left extremity involved and the remaining as right extremity involved.
summarized in Table 2.In total, 120 extremities were affected by LEL and 184 extremities were unaffected.The 120 cases with LEL were classified as stage 1, 2 or 3 according to the clinical staging standard of ISL, and the 32 cases without LEL was classified as stage 0 (Table 2).

Volume Measurements
The volume measurement of calves by water displacement was performed by one of the authors (LM.W.) using a standardized technique as described by Kettle et al (24).Volume of each calf was obtained by subtracting the foot volume recorded from the total foot and calf volume recorded.For obtaining both the foot volume and the calf volume, each lower extremity was immersed in a tank of warm water (at a temperature of 28 to 31°C, average 28°C) to the marked level of lateral malleolus and to the level of fibular head, respectively.Then the displaced volume of the foot and of the total foot and calf was measured.
In addition, the landmarks of lateral malleolus and of fibular head were used as the anatomical reference points to determine the boundary of calf for the volume measurements similar with the subsequent MRI measurements.The difference of volume (DV) between the affected calf and contralateral calf was calculated by volume of affected calf minus that of contralateral unaffected calf.

MRI Protocols
Lower extremities MRI was performed in each participant by using a 3.0-T scanner (Achieva, Phillips Medical systems, Best, the Netherlands) equipped with an eight-channel torso array coil.Each participant was placed in the supine position with feet first into the magnet.The scanning anatomic coverage was between ankle and knee.After localizing images, we obtained bilateral calves of coronal T2-weighted images (3500 msec / 120 msec of repetition time / echo time, 320  304 matrix, 5-mm thick sections, 24 to 32 slices, and 142 to 196 sec acquisition time), axial fat-suppressed T2-weighted images (3500 msec / 120 msec of repetition time/echo time, 320 × 304 matrix, 5-mm thick sections, 32 slices, and 196-sec acquisition time), and axial T1-weighted images (650 msec / 15 msec of repetition time/echo time, 320 × 304 matrix, 5-mm thick sections, 32 slices, and 132-sec acquisition time).The coronal plane images were performed along the long-axis of bilateral fibulas.The axial plane images of calves were obtained perpendicularly to the long-axis of bilateral fibulas on coronal plane images.A field of view of 400 × 400 mm 2 was used to cover the calf.The coronal T2-weighted images were included to plan the T1-and T2 weighted acquisitions while axial T1-and T2-weighted images were included for the conventional diagnostic images.

Image interpretation
Post-processed images were reviewed by two independent readers (LM.W. and JR.X. with 10 and 26 years of experience in musculoskeletal MR imaging, respectively) blinded to the clinical data.By using the uniform window at a Philips MRI workstation, (ViewForum, R4.1) total CSA in square millimeter of soft tissue (T), muscle CSA (M) and subcutaneous tissue CSA (S) of affected calf were obtained for depicting the size changes in this lower extremity.To eliminate the impact of high signal of fat tissue on CSA measurements of edematous soft tissue, the crosssectional fat-suppressed T2-weighted images of midcalf level were used.The mid-calf level corresponded to the midpoint of fibula is defined from lateral malleolus to fibula head.On the previous crosssectional images of calf, the boundary of calf, muscle and bones (fibula and tibia) were manually traced using the computer's mouse.The CSA between the boundary of calf and bones, between the boundary of muscle and bones, as well as between the boundary of calf and muscle represented T, M and S, respectively (Fig. 1).The boundary of calf was defined as the margin of calf skin.The superficial fascia, which showed a slight linear structure with low signal intensity between the low signal of muscle and high signal of subcutaneous tissue on fat-suppressed T2weighted image was defined as the boundary of muscle.The boundary of fibula and tibia was respectively defined as the outer edge of the bones cortex of fibular and tibia.In order to demonstrate the changes in size of affected calf, the previous three parameters of contralateral unaffected calf were obtained by the same methods as the affected calf.The difference of T (DT), M (DM) or S (DS) between the affected calf and contralateral calf was calculated by the corresponding parameter of affected calf minus that of contralateral unaffected calf.

Statistical Analysis
Data is reported as means ± standard deviations.The reliability of measurement of T, M and S obtained by two independent readers was assessed with the intrasubject correlation coefficient (ICC).An ICC of 0-0.20 indicated no agreement; an ICC of 0.21-0.40,poor agreement; an ICC of 0.41-0.60,moderate agreement; an ICC of 0.61-0.80,good agreement; and an ICC greater than 0.80, excellent agreement (25).Cronbach's alpha was calculated as a measure of the internal consistency of the parameters measured by MRI and by water displacement.
Spearman's rank correlation analysis was used to assess the correlation between volume and each value of CSA of calves, and the correlation of the stages of LEL with volume, T, M, and S of affected calf as well as DV, DT, DM and DS of calves.These parameters were compared between patients stratified by stages using multivariate analysis together with Bonferroni correction for multi-comparisons.If there were significant positive findings on multivariate analysis, the cutoff values of aforementioned parameters were then determined with receiver-operating characteristic (ROC) analysis for classifying stages.All statistical analysis was carried out with SPSS (version 16.0, SPSS, Chicago IL, USA).A P value of less than 0.05 was considered indicative of a significant difference.

Interobserver Measurements Agreements and Internal Consistency
There was excellent inter-observer agreement between both readers regarding measurements of T in the affected (ICC = 0.97) and contralateral (ICC = 0.95) calf, of M in the affected (ICC = 0.95) and contralateral (ICC = 0.95) calf, and of S in the affected (ICC = 0.92) and contralateral (ICC = 0.91) calf.The mean measurement value of two readers was used as the final results.The Cronbach α coefficient for the volume and MRI measurements was 0.95 and 0.97, respectively.Removing any of the parameters did not improve upon the internal consistency.The mean volumes and values of soft tissue CSA including T, M and S of affected and unaffected calf were summarized in Table 3. DV, DT, DM and DS between affected and unaffected calf were also showed in this table.Volume, T and S of affected calf showed significantly greater than of unaffected calf (all P < 0.05).No significant difference was found in M between affected and unaffected calf.

Correlations of Volumes with Soft Tissue CSA of Calves
There was a close correlation between the volume obtained by water displacement and soft tissue CSA measured by MRI for T (r = 0.943) or S (r = 0.926) of affected calf, and for T (r = 0.825), M (r = 0.632) or S (r = 0.686) of unaffected calf (all p < 0.01).There was a weak correlation between volume and M for affected calf (r = 0.349, p < 0.01).A strong correlation of DV of calves with DT (r = 0.956, p < 0.01) or DS of calves (r = 0.934, p < 0.01) was found, and no significant correlation was between DV and DM of calves (p = 0.16).

Correlations of LEL Stages with Soft Tissue CSA of Calves vs. Volume
There was a correlation of LEL stages with T (r = 0.840) or S (r = 0.838) of affected calf, and with DT (r = 0.849) or DS (r = 0.864) of calves, which was stronger than with volume of affected calf (r = 0.781) or DV (r = 0.823) of calves (all p < 0.01).There was a weak correlation between stages of LEL and M of affected calf (r = 0.254, p = 0.013), and no significant correlation was found between LEL stages and DM of calves (p = 0.764).Consequently, all these parameters except M of affected calf and DM of calves could be used as indicators for assessing the LEL.

Staging Unilateral LEL with Soft Tissue CSA of Calves vs. Volume
With increasing stages of LEL, there was a more obvious trend toward an increase in T, S than in volume of the affected calf while the trend toward an increase or decrease could not be found in M of the affected calf (Table 4, and Fig. 2).Multivariate analysis for pairwise comparisons with Bonferroni correction showed that more significant differences in T and S than in volume of affected calf, and in DT and DS than in DV of calves between LEL stages (all p < 0.05) except between stage 0 and 1. Between stage 0 and 1, significant difference was found in T and S of affected calf, and in DS of calves whereas there was no difference in volume of affected calf (p = 0.117), and in DV (p = 0.541) or DT (p = 0.061) of calves.

ROC Analysis of Soft Tissue CSA of Calves vs. Volume for Staging LEL
The cutoff values of T, S and volume of affected calf as well as DT, DS and DV of calves according to different LEL stages are shown in Table 5.Among these six parameters, area under the ROC curve (AUC) of DS for classifying LEL stage 0 vs. 1, 1 vs. 2, and 2 vs. 3 was greater than of any other parameters (Fig. 3).

Discussion
Valid, reliable and sensitive measurement of lower extremity size is required for monitoring, diagnosing and staging LEL in cervical and endometrial cancer survivors.Although volume measurement obtained by water displacement is the gold standard for assessing LEL, it is not an informative method in routine clinical practice.Pathologically, LEL primarily presents as thickened subcutaneous tissue and muscle (26).MRI allows not only characterizing the lesions within subcutaneous tissue, but also distinguishing this increasing thickened subcutaneous tissue from muscle.
Since the fat tissue within subcutaneous tissue appears as high signal on T1-and T2-weighted imaging, edema could not be well depicted on these two sequences (17,20).T2-weighted imaging with fat suppression can overcome the limitation of T1-and T2weighted imaging.As shown on fat-suppressed T2weithted imaging, measuring CSA of soft tissue of calves could be reproducible because excellent interobserver agreements of the T, M and S of calves were achieved for the repetitive measurements.
In this study, soft tissue CSA of bilateral calves except M of affected calf and DM of calves measured at MRI demonstrated a close correlation with the volume of  calves obtained by water displacement.Particularly, T and S of affected calf showed a strong correlation with volume of this calf, and DT and DS had a strong correlation with DV of calves.These correlations with excellent internal consistency in all measurements suggest that it is not M but S or T of affected calf that contribute significant changes to the volume increase of affected calf.Our findings can be explained by the pathophysiological changes of lymphedema that accumulation of lymph fluid, increasing deposition of adipose tissue, and a marked increase in the content of fibroblasts are absolutely associated with the thickened subcutaneous tissue of extremity (27).These findings are similar to previous study on arm lymphedema after breast cancer surgery that showed arm volume acquired by water displacement closely correlated with the total CSA of arm soft tissue (subcutaneous tissue and muscle tissue) and of subcutaneous tissue at CT (28).Although most of previous studies correspond with our findings that LEL was associated with thickened soft tissue, our study initially highlighted the role of thickened subcutaneous tissue on the volume increase of extremity with LEL (13,20,29,30).Thus, the methods of obtaining soft tissue CSA of lower extremity at MRI provide an alternatively valid and informative way to evaluate the size changes of extremity with LEL in a clinical environment.
For staging unilateral LEL, the more excellent correlation of soft tissue CSA with LEL stages than of the volume suggested that T and S of affected calf Notes: T = total CSA of soft tissue, M = muscle CSA, S = subcutaneous tissue CSA, Affected = affected calf, Difference = difference between affected calf and unaffected calf, AUC = area under the curve.
patients most concerned about is how and whether LEL could be detected in the early stage.The International Society of Lymphology recommends a 20% increase of volume difference between limbs to define LEL.This may result in losing opportunity of patients with this disease to receive timely treatment (31).Baas et al (32) used volume displacement to quantify frequency of lymphedema after 151 consecutive cases of groin dissection and reported that in 7% of their cases there was a > 20% volume difference between limbs whereas 75% had < 6.5% volume difference between limbs.Spillane et al (33) used a difference of 15% in volume between extremities in predicting moderate or severe LEL, which could result in misclassification rates of 16% in patients.In our study, by using DS of calves, performance of this parameter for differentiating LEL stage 1 from stage 0 achieved with 100% of sensitivity and specificity.In contrast, the parameter of DV did not indicate significant difference between stage 0 and 1.Our findings suggest DS might be more sensitive than DV in early detecting LEL.Thus, we recommend that DS of calves could be a potential indicator for classifying different LEL stages especially in its early stage.
Our study had several limitations.First, the 120 patients with LEL and the 32 participants without LEL may have had pre-clinical lymphedema in their unaffected extremity; hence, a subsequent study should include a group of un-operated, healthy controls.Second, validation of this quantitative approach in patients with LEL in bilateral extremities was not carried out.Despite this limitation, our results are suitable for the patients with unilateral LEL.Third, this quantitative approach has not been validated for the determination of therapeutic effectiveness; further studies should focus on quantification of lymphedema for prognostic purposes and determination of therapeutic responses.
In conclusion, fat-suppressed T2-weighted MRI can be a feasible and sensitive method for quantitative assessment of unilateral LEL secondary to cervical and endometrial cancer treatments.DS of calves obtained at MRI could be recommended for staging LEL and might have an advantage over volume measurement for early detection of LEL.

Fig. 1 .
Fig. 1.These images illustrate how total cross-sectional area (CSA) of soft tissue (T), muscle CSA (M) and subcutaneous tissue CSA (S) of calves are acquired.CSA between the boundary of calf and bones, between the boundary of muscle and bones, and between the boundary of calf and muscle represented T, M and S, respectively.The boundary of calf was defined as the margin of calf skin (short coarse arrow).The superficial fascia (long coarse arrow) was defined as the boundary of muscle.The boundary of fibula and tibia was respectively defined as the outer edge of the bones cortex of fibular and tibia.

Fig. 2 .
Fig. 2.These histograms show the correlations of mean volume (a), total cross-sectional area (CSA) of soft tissue (T, b), subcutaneous tissues CSA (S, c), and muscle CSA (M, d) of affected calf versus contralateral unaffected calf with stages of lower extremity lymphedema.

Fig. 3 .
Fig. 3. Fat-suppressed T2-weighted images of mid-calf level show the changes of subcutaneous tissues thickness with lower extremity lymphedema identified as stage 0 in bilateral calves (a) in a 52-year-old female, stage 1 in the right calf (b, arrows) in 48-year-old female, stage 2 in the left calf (c, arrows) in a 54-year-old woman, and stage 3 in the right calf (d, arrows) in a 62-year-old woman.

Fig. 4 .
Fig. 4. Receiver operating characteristic curves show volume, total cross-sectional area (CSA) of soft tissue (T), and subcutaneous tissues CSA (S) of affected calf as well as the difference of volume (DV), difference of T (DT) and difference of S (DS) between affected calf and contralateral unaffected calf for classifying lower extremity lymphedema stage 0 vs. 1 (a), 1 vs. 2 (b), and 2 vs. 3 (c).
± 0.24 b, c, d 0.78 ± 0.54 a, c, d 2.77 ± 1.31 a, b, d 5.66 ± 1.77 a, b, c Notes: a Different from LEL Stage 0, b Different from Stage 1, c Different from Stage 2, and d Different from Stage 3.All the comparisons denote significance after Bonferroni correction (P < 0.05).T = total CSA of soft tissue, M = muscle CSA, S = subcutaneous tissue CSA, Affected = affected calf, Difference = difference between affected calf and unaffected calf.

Table 3 . Volume (mean ± SD) and soft tissue cross-sectional area (CSA, mean ± SD) of calves
a Different from unaffected calf (all p < 0.01), b Different from unaffected calf (p = 0.945).T = total CSA of soft tissue, M = muscle CSA, S = subcutaneous tissue CSA, Difference = difference between affected calf and unaffected calf.

Table 5 . P values for statistical comparisons in volume and in soft tissue cross-sectional area (CSA) of calves between stages of lower extremity lymphedema
could be more accurate than volume of affected calf, or DT and DS of calves could be more accurate than DV of calves.Our findings highlighted that DS of calves was the best parameter for classifying the stages of LEL because this parameter had a higher AUC than any other parameters.Since the diagnosis of LEL at the stage 1 is essential for preventing the progression of the disease and its complications, what physicians and