Lymphedema evaluation using noninvasive 3T MR lymphangiography

Abstract

Purpose: To exploit the long 3.0T relaxation times and low flow velocity of lymphatic fluid to develop a noninvasive 3.0T lymphangiography sequence and evaluate its relevance in patients with lymphedema.

Materials and methods: A 3.0T turbo-spin-echo (TSE) pulse train with long echo time (TEeffective = 600 msec; shot-duration = 13.2 msec) and TSE-factor (TSE-factor = 90) was developed and signal evolution simulated. The method was evaluated in healthy adults (n = 11) and patients with unilateral breast cancer treatment-related lymphedema (BCRL; n = 25), with a subgroup (n = 5) of BCRL participants scanned before and after manual lymphatic drainage (MLD) therapy. Maximal lymphatic vessel cross-sectional area, signal-to-noise-ratio (SNR), and results from a five-point categorical scoring system were recorded. Nonparametric tests were applied to evaluate study parameter differences between controls and patients, as well as between affected and contralateral sides in patients (significance criteria: two-sided P < 0.05).

Results: Patient volunteers demonstrated larger lymphatic cross-sectional areas in the affected (arm = 12.9 ± 6.3 mm2 ; torso = 17.2 ± 15.6 mm2 ) vs. contralateral (arm = 9.4 ± 3.9 mm2 ; torso = 9.1 ± 4.6 mm2 ) side; this difference was significant both for the arm (P = 0.014) and torso (P = 0.025). Affected (arm: P = 0.010; torso: P = 0.016) but not contralateral (arm: P = 0.42; torso: P = 0.71) vessel areas were significantly elevated compared with control values. Lymphatic cross-sectional areas reduced following MLD on the affected side (pre-MLD: arm = 8.8 ± 1.8 mm2 ; torso = 31.4 ± 26.0 mm2 ; post-MLD: arm = 6.6 ± 1.8 mm2 ; torso = 23.1 ± 24.3 mm2 ). This change was significant in the torso (P = 0.036). The categorical scoring was found to be less specific for detecting lateralizing disease compared to lymphatic-vessel areas.

Conclusion: A 3.0T lymphangiography sequence is proposed, which allows for upper extremity lymph stasis to be detected in ∼10 minutes without exogenous contrast agents.

Level of evidence: 1 Technical Efficacy: Stage 3 J. Magn. Reson. Imaging 2017;46:1349-1360.

Keywords: MRI; breast cancer; lymphangiography; lymphedema; manual lymphatic drainage; turbo spin echo.

© 2017 International Society for Magnetic Resonance in Medicine.

Figures

Figure 1

(A) Relevant components of the proposed lymphangiography sequence, which consists of spectral presaturation with inversion recovery (SPIR) and spatially-selective presaturation pulses for fat and inflow suppression, respectively. Preparation is followed by a long turbo-spin-echo (TSE) readout (TSE-factor = 90; echo spacing = 13.2 ms; effective echo time = 600 ms). (B) These parameters lead to a fast decay of blood and muscle signal over the readout, yet a slower decay of lymphatic and CSF magnetization due to longer T2 and lower velocity of these species. For a health control volunteer, (C) structural imaging of lymphatics are displayed as the maximum intensity projection (MIP) from diffusion-weighted-imaging-with-background suppression (DWIBS) and (D) the lymphangiography sequence, showing the thoracic duct (large central white arrow) and convergence of apparent lymphatic vessels near the axillary lymph nodes (white arrows). Note that since lymphatic node T1 and T2 are substantially lower than lymphatic fluid T1 and T2, the lymphatic nodes are not clearly visible in the lymphangiography sequence. (E-G) High spatial resolution structural imaging at the site of an axillary lymphatic node (white arrow).

Figure 2

Lymphatic vessel cross-sectional area measurement procedure. (A) Orthogonal representations of lymphangiography MIPs are generated to locate the lymphatic vessel of interest. This example is performed on the thoracic duct (yellow arrow) which is located anterior to the spinal cord. (B) The cross-sectional area is measured on magnified versions of the magnitude images themselves where vessel size can be confirmed in all three imaging planes.

Figure 3

A 53 year old female with left-sided Stage 1 BCRL (onset T1-weighted localizer shows the region imaged, along with (B) T2-weighted and (C) T1-weighted structural imaging of the involved side. Regions of subcutaneous edema are visible in the T2-weighted scan. A (D) coronal MIP and (E) source image from the lymphangiography scan show contrast asymmetry between affected and contralateral quadrants (white arrows). The thoracic duct (E; large central white arrow) is also clearly visible. Compared with the subcutaneous edema seen in (B), the lymphatic vessels in (D-E) can be traced on continuous slices in a complex matrix of dilated tubes and channels.

Figure 4

Lymphangiograms pre (above) and post (below) manual lymphatic drainage (MLD) therapy for a 78 year old female with left-sided Stage 2 BCRL scanned three years after neoadjuvant chemotherapy, radiation, and removal of 14 lymph nodes (8 positive for carcinoma). The patient experienced reduced stiffness of her limb following a 50-minute session of MLD along with the therapist palpating reduction in fibrosis and induration of the skin along the left inner forearm, upper medial arm and lateral chest wall. The post-MLD findings indicate a reduction but not complete elimination of contrast consistent with lymph stasis (white arrows). Future MLD sessions could focus attention on rerouting the remaining congestion along the lateral chest wall to the ipsilateral lower quadrant. Reductions in contrast consistent with mobilization of lymphatic fluid are depicted by white arrows in (B-C) on the affected side.

Figure 5

Summary of primary study findings. (A) Maximal lymphatic cross-sectional area did not differ significantly between left and right sides in controls, but was significantly elevated in the affected relative to contralateral side of patients. This was found both for arm and torso vessels and for both symptomatic (Stages 1–2) and sub-clinical patients. (B) The categorical scoring system was less sensitive to lateralizing disease. Neither left vs. right scores in controls, nor affected vs. contralateral scores in all patients, were significantly different. When only symptomatic patients were considered however, the scores in the affected side were significantly elevated relative to control scores. (C) In the subgroup of patients scanned before and after MLD, maximal lymphatic cross-sectional area was observed to reduce on the affected side only; this change was significant in the torso region and just beyond significance criteria in the arm region. (D) The categorical score changes pre- vs. post-MLD were not significant, however the effect sizes for the reductions were large on both the affected (Cohen’s d=0.81) and contralateral (Cohen’s d=0.63) side. * two-sided p<0.05; # one-sided p<0.05. Values summarized are mean ± one standard error of the mean. Standard deviations are summarized as well in Table 2.