Lymphoedema Evidence Library › Topics › Lymphatic surgery
Lymphatic surgery
Lymphatic surgery
33 sources
Evidence in one minute
- The evidence base is mostly level 4–5: of the 50 most-cited lymphatic surgery papers, 74% are level 4–5 and only two are RCTs. Al Qurashi AA 2026
- Choose by pathology: fluid-dominant (pitting) → physiological surgery (LVA, VLNT); fat/fibrosis-dominant (non-pitting) → reductive surgery (liposuction, excision). Zurbuchen 2025 · Executive Committee of the International Society of Lymphology 2023
- LVA: works where functional lymphatics are seen on ICG; complication rate ~1%; in BCRL the multicentre N-LVA RCT's primary outcome was null at 6 months (low certainty; revised 2026-10-08). Lee 2025 · Gaxiola-García 2024
- Lower limb (umbrella review): volume −34% to −47% (very low certainty); cellulitis −2.1 episodes/year after VLNT and 0.84→0.07/year after LVA (low by GRADE — uncontrolled before-after; the review's guideline scale calls it moderate; revised 2026-10-08); ILR prevents ~30 per 100 cases after iliofemoral lymphadenectomy but may worsen oncological outcome in limb skin cancers. Gloviczki 2026
- VLNT: better for advanced disease; donor-site lymphoedema is the main risk; combined with radical reduction in stage III (series). Perets 2024 · Ciudad 2019
- Liposuction: largest, most consistent volume reduction for non-pitting fat-dominant disease (82% weighted average in early series) only with lifelong flat-knit compression. International Lymphoedema Framework 2012
- Excisional (Charles, Thompson): complication rate ~46%; ISL says avoid except for elephantiasis/genital disease. Executive Committee of the International Society of Lymphology 2023
- Every review agrees: surgery does not end compression or self-care. Flores 2024
- Technique notes: Zheng 2024; ICG mapping in SA journal context: Hsu 2025.
Appraisal update 2026-10-09
- Narrative reviews over-read the trials. Lee 2025 calls the null N-LVA interim “significant proof” of benefit; Perets 2024 claims VLNT is “most cost-effective” with no cost data and misreports the Dionyssiou RCT; Park 2020 and Zurbuchen 2025 quote uncontrolled percentage reductions without design caveats. Use the RCT pages and Gloviczki 2026 for effects. Lee 2025 · Perets 2024 · Park 2020 · Zurbuchen 2025
- Technique and case series: Zheng 2024 OLA (10 patients; text age and BMI do not match its table; 7-day patency only) and Ciudad 2019 — very low certainty. The Al Qurashi audit counts the van Mulken robotic pilot as one of only two “level 2” RCTs; it compared two ways of doing LVA, not LVA v no LVA. Zheng 2024 · Al Qurashi AA 2026
- Cochrane protocol (Haas 2025) is methodologically strong; declared interests include a compression-garment consultancy and a 3M advisory board. Haas 2025
For a vascular surgeon in South Africa
LVA needs supermicrosurgery and ICG imaging (limited availability in SA — LAOSA 2018 notes not all procedures are offered). Liposuction for fat-dominant disease and ILR at the time of groin/pelvic node dissection are the most relevant to vascular and surgical oncology practice. Do not cite the anonymous "clinical guide" (Unknown unknown).
Textbooks
Cheng 2016 · Harder 2024 · Schaverien 2022 · Baumeister 2017
Update 2026-10-08 — downloaded evidence
- Randomised evidence now held (graded 2026-10-08): N-LVA (LVA v CDT, early BCRL): primary HRQoL outcome null, physical-function domain better (below MCID), volume unchanged at 6 months — low certainty; not cost-effective from a health-service perspective (moderate). ILR at axillary dissection: volume-defined BCRL 9.5% v 32% (preliminary, low certainty). Cochrane review of microsurgery vs CDT under way (protocol). Jonis 2024 · Kleeven 2026 · Coriddi 2023 · Haas 2025
How the evidence points
| High | Mod. | Low | V. low | None | |
|---|---|---|---|---|---|
| Positive | 0 | 0 | 1 | 0 | 0 |
| Null or negative | 0 | 0 | 2 | 0 | 0 |
Sources on this page with a comparison, by the result of their primary outcome and the GRADE certainty of their main finding (Mod. = Moderate, V. low = Very low, None = not graded). Context sources are not counted. How direction is decided.
All 33 sources
Loading 33 sources…