Table 2.
Translational maturity of representative efficacy, biomarker, safety, and registered studies.
| Study | Design/sample | Reported or inferred target | Evidence role and principal limitation |
|---|---|---|---|
| Li X. et al. (2024) | Single case; n = 1; 5-week follow-up. | Insufficient detail to determine. | Operability and hypothesis generation only; no counterfactual, reproducible interface, durable patency, or causal inference. |
| Chen J. Y. et al. (2025) | Prospective single-arm cohort; n = 26; 1 month. | Lymphatic flap-vein. | Median MMSE 3 to 5 and ~60% caregiver-reported improvement; non-significant CSF trends and two transient arm-elevation events; no control, blinded outcomes, or durable patency. |
| Tang et al. (2025) | Prospective single-arm mechanistic cohort; n = 28; approximately 1-week cognitive assessment. | Lymph node-vein. | ~64.3% had a higher MMSE at 1 week and nodal tau signal was detected; the window is vulnerable to perioperative and practice effects and does not establish durable tau clearance. |
| Ma et al. (2025) | Retrospective single-center study; n = 41; 3 months. | Collecting lymphatic vessel-vein. | Exploratory 50% CDR-SB response definition, >67% with at least one favorable Aβ42/40 or p-tau181 measure, and two mild adverse events; multiple endpoints, no control, and target-level patency gap. |
| Fu et al. (2026) | Prospective single-arm study; n = 139 severe AD; 6 months. | Collecting lymphatic vessel-vein. | Direct lymphatic-vessel-to-EJV anastomoses with partial regional node resection; local nodal relations insufficiently reported. Single-arm clinical/biomarker signals with reported assessor blinding and a descriptive external lecanemab reference, not a randomized within-study comparison. Delirium/sleep disturbance and one bleeding/hematoma reoperation; no durable source-defined patency or causal attribution. |
| Wu et al. (2025) | Retrospective pilot; 36 source cases, 10 analyzed; 4 short-term responders; 1 month. | Lymphatic flap-vein. | Within-sample PVS thresholds (~5,150 and ~3,630 mm3) produced AUCs of 1.00 and 0.88; severe overfitting risk, no external validation, and no durable patency proof. |
| Hu et al. (2026) | Exploratory single-arm biomarker study; n = 30; 180 days. | Insufficient detail to determine. | Plasma Aβ42 plus Aβ42/40 model AUC 0.737; no control, external validation, target-level patency, source attribution, or clinical-surrogate validation. |
| Jiang et al. (2026) | Case report; n = 1; hemorrhage at 23 days and later seizure/immune findings. | Insufficient detail to determine. | High-importance delayed neurological/infectious/immune safety signal; incomplete operative detail and temporal association cannot establish incidence or causality. |
| Chen X. et al. (2026) | Retrospective case series; n = 5; short follow-up. | Collecting vessel-interposed vein graft-EJV (dynamic neck LVA). | All grafts were described as patent with no recorded postoperative complications; small mean cognitive changes and heterogeneous PET signals. Very small uncontrolled sample, multiple endpoints, measurement sensitivity, and no validation of the proposed dynamic mechanism or long-term safety. |
| Jin L. et al. (2026) | Retrospective comparative cohort; 6-month lecanemab versus dcLVA; group sizes were not stated in the PubMed abstract. | Insufficient operative detail to classify confidently. | Cognitive decline was reported in 23.5% versus 81.8%, and change in centiloid favored lecanemab; non-randomized treatment selection, unclear interface and patency, and internally derived imaging threshold prevent a causal comparative-effect estimate. |
| Li Y. et al. (2026) | Preoperative CSF proteomics; n = 90; internally defined responders and non-responders. | Interface not reported sufficiently for target classification. | NPTX2/IGSF10 AUC 0.854 and four-marker AUC 0.880; internally derived prediction without external validation, untreated comparison, or target-specific patency. Candidate stratification evidence only. |
| ClinicalTrials.gov (2026a) | Planned multicenter open-label, blinded-endpoint randomized study; approximately n = 376; 12-month core outcomes; no results. | Broadly described deep cervical lymphatic tissue-to-vein procedure. | Important randomized design and multimodal endpoint plan; target-level method, learning curve, patency, harms, and outcomes remain unavailable. |
| ClinicalTrials.gov (2026b) | Planned multicenter prospective registry; approximately n = 814 across ~40 centers; follow-up to 24 months; no results. | Insufficient detail to determine. | Can characterize heterogeneity, learning, rare harms, and external validity; no non-surgical control and therefore not a substitute for randomized evidence. |
Classification follows the available titles, abstracts, methods, operative descriptions, figures, and full-text detail. Primary interface assignment is based on the lymphatic-side structure directly connected to the venous lumen. Afferent/pre-nodal and efferent/post-nodal are treated only as local relations to a specified node, not as separate operative categories. Sources with insufficient target disclosure remain unclassified. Newly added reports were interpreted according to the verified information level available; abstract-level comparative results were not used for unreported procedural or causal claims. Temporal association in a case report does not establish causality. Registrations describe planned, not completed, evidence. AD, Alzheimer’s disease; Aβ, amyloid beta; AUC, area under the receiver operating characteristic curve; CDR-SB, Clinical Dementia Rating-Sum of Boxes; CSF, cerebrospinal fluid; dcLVA, deep cervical lymphaticovenous anastomosis; EJV, external jugular vein; IGSF10, immunoglobulin superfamily member 10; LVA, lymphaticovenous anastomosis; MMSE, Mini-Mental State Examination; NPTX2, neuronal pentraxin-2; PET, positron emission tomography; p-tau, phosphorylated tau (numeric suffix denotes the residue); PVS, perivascular space.