MRI-guided convection-enhanced delivery

The research record

This is the project's file for this landmark, as written and checked. Sources are listed at the end; anything not yet confirmed is marked "Not yet verified".

Who

Krystof Bankiewicz, Massimo Fiandaca, Jason Richardson, Martin Krauze, John Forsayeth and colleagues, University of California San Francisco. Key papers: Krauze et al., Exp Neurol 2005; Fiandaca et al., NeuroImage 2009; Richardson et al., Stereotact Funct Neurosurg 2011 and Mol Ther 2011.

What happened

Convection-enhanced delivery (CED) pushes fluid into brain tissue by gentle pressure through a thin cannula, so it spreads beyond the needle tip. Without imaging, nobody sees where it goes. A drug that leaks into spinal fluid or the wrong region is wasted. The team added a visible tracer (gadolinium) and watched the infusion on MRI.

• 2005. Real-time MRI of liposomes infused into primate putamen showed they travelled along perivascular spaces (channels around blood vessels). The authors said this may explain side effects seen in clinical trials using CED and argued that monitoring is needed.
• 2009. Real-time MRI tracked AAV vector delivery in three monkeys (corona radiata, putamen, thalamus) using gadolinium-loaded liposomes as a stand-in. Their spread matched AAV1-GFP closely but AAV2-AADC less closely.
• 2011. A skull-mounted aiming device for a planned human trial was tested in monkeys. Across 11 targets the average cannula tip error was 0.8 mm. The distribution volume of the tracer rose in a straight line with the volume infused (R² = 0.97), with no blockage, backflow or tissue damage. A second paper modelled the planned AAV2-GDNF procedure in monkey putamen, with tip error under 1 mm and GDNF expression matching the MRI tracer.

Why it moved us toward a cure

The neurturin autopsies (landmark 179) suggested that a growth-factor gene therapy can fail because it covers too little of the target. Real-time MRI lets surgeons adjust until coverage is adequate. It was used in the MRI-guided AADC trial (landmark 112). The NIH Phase 1 of AAV2-GDNF (13 people, gadolinium co-infusion; Rocco 2022, Munjal 2026 and Heiss 2026 all describe this one study) used real-time MRI to confirm placement and adjust the cannula, and reported no brain-tissue toxicity on serial MRI for up to 5 years (Rocco 2022). That is the route behind the GDNF gene therapy line (landmark 39).

Outcome

Advanced as a tool. Coverage is still limited. In the 13-person AAV2-GDNF study, 3 people (23%) had unintended spread into the caudate (Munjal 2026), and mean putamen coverage was 26% ± 10% (Heiss 2026). The later AB-1005 Phase 1b (11 people, higher dose, AskBio) reported 63% coverage (Van Laar 2025). The NIH study reported no clinical benefit (Munjal 2026).

Sources

• Krauze et al., Exp Neurol 2005, DOI 10.1016/j.expneurol.2005.07.009, PMID 16109410
• Fiandaca et al., NeuroImage 2009, DOI 10.1016/j.neuroimage.2008.11.012, PMID 19095069
• Richardson et al., Stereotact Funct Neurosurg 2011, DOI 10.1159/000323544, PMID 21494065
• Richardson et al., Mol Ther 2011, DOI 10.1038/mt.2011.11, PMID 21343917
• Rocco et al., Mol Ther 2022, DOI 10.1016/j.ymthe.2022.08.003; Munjal et al., Neurosurgery 2026, DOI 10.1227/neu.0000000000003729; Van Laar et al., Mov Disord 2025, DOI 10.1002/mds.30193 (abstracts via Europe PMC); Heiss et al., Mov Disord 2026;41:2399–2407, DOI 10.1002/mds.70393
• Not yet verified: the first-in-human use of this method and who did it; whether REGENERATE-PD uses real-time MRI; origin of CED outside the field; the Krauze, Fiandaca and Richardson papers' funders beyond NINDS grants listed in PubMed; the claim that the method was used in the AADC trial (landmark 112).