OCD & Treatment-Resistant Depression

Evidence status. OCD DBS has an FDA Humanitarian Device Exemption; depression DBS and psychiatric MRgFUS remain investigational. Controlled evidence is limited, heterogeneous, and substantially less mature than open-label series.

Obsessive-compulsive disorder is the flagship DBS indication and the natural entry point for a developing program, as the only one with an FDA Humanitarian Device Exemption. The ablative alternatives for OCD are covered in Part III.

The Humanitarian Device Exemption

In 2009 the FDA granted a Humanitarian Device Exemption (HDE) for Reclaim DBS Therapy for OCD: bilateral stimulation of the anterior limb of the internal capsule as an adjunct to medications and as an alternative to anterior capsulotomy in adults with chronic, severe, treatment-resistant OCD who have failed at least three SSRIs. The HDE pathway applies to humanitarian-use devices and permits marketing on the basis of safety and probable benefit rather than a pivotal effectiveness trial; clinical use still requires local IRB or appropriate committee oversight at the implanting center. OCD DBS is thus the one FDA-recognized psychiatric DBS indication a program can offer clinically outside a research protocol, which is exactly why it is the usual starting point for a new service. Candidacy follows the refractoriness and committee framework detailed on the selection and program page: severe YBOCS, adequate trials of multiple SSRIs and clomipramine, antipsychotic augmentation, and genuine exposure-and-response-prevention therapy.

Real-world uptake has nonetheless been limited. Despite the HDE, annual OCD DBS volume is low and by some analyses declined after approval. Restrictive coverage and reimbursement, referral friction, the burden of a multidisciplinary pathway, limited experienced centers, patient preference, and residual uncertainty about efficacy and burden may all contribute. The HDE permits marketing on probable benefit; it neither establishes conventional PMA-level effectiveness nor secures payer coverage.

One corridor, several named targets

As developed on the foundations and circuits page, the principal OCD targets are points along, or just medial/inferior to, the ventral internal-capsule fiber corridor, and they converge on overlapping prefrontal networks:

• Ventral capsule / ventral striatum (VC/VS): the most widely used target and the anatomical basis of the HDE; the four-center worldwide series (26 patients) reported clinically significant improvement and functional gains in about two-thirds of patients, with adverse effects overwhelmingly transient.
• Nucleus accumbens (NAc): the ventral striatal reward hub; the Amsterdam group (Denys, 16 patients) reported a mean YBOCS reduction of about 46% in open treatment, with a smaller sham-controlled crossover phase favoring active stimulation.
• Bed nucleus of the stria terminalis (BNST): just posterosuperior to the NAc; supported by a small randomized sham-controlled crossover trial (Mosley).
• Subthalamic nucleus (anteromedial/limbic STN): supported by the small randomized STOC trial; proposed hyperdirect response pathways remain model-dependent and incompletely validated.

Because the targets converge, the practical question is less “which nucleus” than “which fiber bundle does the active contact capture,” and patient-specific tractography is increasingly used to individualize placement within the VC/VS.

Randomized and pooled data

The controlled evidence is built on small samples but points in a consistent direction. The STOC trial (Mallet, 16 patients, double-blind crossover) found active subthalamic stimulation reduced YBOCS to 19±8 versus 28±7 with sham (P=0.01); global functioning also improved, with GAF 56±14 versus 43±8 (P=0.005). Crossover and sham-controlled studies of VC/VS and NAc, and a randomized BNST trial, similarly favored active stimulation. The Alonso meta-analysis (31 studies, 116 subjects) estimated a global 45% YBOCS reduction and 60% responder rate across targets. Most recently, an individual-participant-data meta-analysis of nine sham-controlled RCTs (91 patients) found a 5.1-point YBOCS advantage for active over sham stimulation, an odds ratio for response of 4.7, and a number needed to treat of about 3.9. This is the strongest controlled summary to date, although the authors rated the quality of evidence as low and heterogeneity as high.

Programming a slow circuit

OCD DBS programming differs fundamentally from movement-disorder programming, and the difference is time. There is no immediate, objective sign (like tremor arrest) to optimize against; instead, the team titrates against a mood-and-anxiety response that unfolds over weeks to months. Practical features:

• Stimulation is often monopolar on ventral contacts, and OCD/mood targets often require higher amplitudes and longer pulse widths than Parkinson disease programming.
• Acute stimulation-induced effects (spontaneous mirth or smiling, a sudden lift in mood, increased energy, or conversely anxiety, panic, or autonomic changes) may inform safety assessment and contact testing, but do not reliably predict durable antiobsessional benefit.
• Adjustments are made gradually with structured follow-up; an apparent non-response should prompt systematic review of contacts and parameters before concluding failure, because the latency to benefit is long.
• Antiobsessional benefit may follow mood or anxiety improvement, but this sequence is not universal and mood elevation is not a validated surrogate for OCD response; continued ERP after implant is part of optimizing outcome.

What to anticipate

Adverse effects fall into three groups. Surgical/hardware risks are those of any DBS: intracerebral hemorrhage, infection, lead migration or fracture, and device malfunction (the STOC trial recorded a hemorrhage and infections among its serious events). Stimulation-induced effects are largely reversible by reprogramming and are dominated by hypomania (the most frequently reported stimulation-related event), along with anxiety, agitation, transient mood or autonomic changes, and, in the IPD analysis, cognitive complaints. Disease- and course-related concerns include the persistent suicidality risk inherent to this severely ill population, which requires ongoing psychiatric co-management irrespective of device response. Most stimulation effects resolve with parameter changes, underscoring why an experienced psychiatrist must remain embedded in the programming loop rather than consulted only at crises.

Depression carries the field's most instructive evidence story: open-label promise, negative randomized trials, durable long-term benefit, and a pivot toward individualized targeting. It remains investigational. A new pivotal trial is now running: Abbott's TRANSCEND study (NCT06423430) is a prospective, double-blind, multicenter randomized trial of subcallosal cingulate DBS for treatment-resistant depression, conducted under an FDA investigational device exemption, enrolling about 100 adults who have failed at least four antidepressant treatments, with unblinding at 12 months. Abbott has received FDA Breakthrough Device designation for this indication. Breakthrough designation accelerates FDA interaction and review; it is not approval, not a finding of effectiveness, and does not permit clinical use outside the trial. Routine OCD HDE therapy uses open-loop stimulation; personalized responsive (closed-loop) psychiatric stimulation has so far been reported only in research, including a single-patient depression study (Scangos). Chronic sensing is available on the Medtronic Percept platform, and the FDA approved BrainSense Adaptive closed-loop stimulation in February 2025, but that approval covers Parkinson disease only; adaptive psychiatric DBS remains investigational, and sensing capability alone does not establish efficacy. Ablative options for depression are covered in Part III.

Where electrodes go for depression

Depression DBS samples several entry points into an interconnected mood-and-reward network (developed on the foundations and circuits page):

• Subcallosal cingulate (SCC, Brodmann area 25): Mayberg's original imaging-derived target; benefit appears to depend on capturing a specific bundle of converging white-matter tracts.
• Ventral capsule / ventral striatum (VC/VS): shared with OCD, engaging ventral striatal reward circuitry.
• Superolateral medial forebrain bundle (slMFB): a tractography-defined reward/motivation target associated with rapid antidepressant effects in small open-label studies.
• Inferior thalamic peduncle and the lateral habenula: less-studied nodes in the same network.

From open-label promise to negative RCTs

Mayberg's 2005 open-label SCC report (antidepressant response in four of six profoundly refractory patients at 6 months) ignited the field. Two industry-sponsored randomized, sham-controlled pivotal trials followed, and both were negative on their primary endpoints:

• BROADEN (Holtzheimer 2017, SCC, 90 patients randomized to active or sham stimulation) reported acceptable safety and feasibility but did not show statistically significant antidepressant efficacy after 6 months of double-blind stimulation (response about 20% active vs 17% sham); enrollment was stopped after a futility analysis.
• RECLAIM (Dougherty 2015, VC/VS, 30 patients) found no significant difference between active and sham in the 16-week double-blind phase and was terminated for futility.

Why the randomized design struggled

The negative pivotal results are direct evidence that the tested protocols did not establish efficacy at their prespecified endpoints, and they prevent routine clinical adoption of DBS for depression. Slow response, difficult blinding, patient heterogeneity, optimization time, and anatomical rather than individualized targeting may have contributed, but these explanations are hypotheses rather than a way to neutralize negative trials. Later open-label and pooled data justify continued rigorous study; they do not substitute for adequately powered controlled evidence.

What happens over years

The longitudinal data are the counterweight to the negative RCTs. Crowell and colleagues followed an SCC cohort of 28 patients open-label for up to eight years: 75% of patients met response criteria for more than half their years in the study, response and remission rates remained at or above 50% and 30% through years 2 to 8, 21% showed continuous response from the first year onward, and stimulation was well tolerated, with no acute or chronic stimulation-related side effects and no suicides reported. Most powerfully, a pooled analysis of 172 SCC-implanted patients across five studies (including BROADEN's long-term phase) found mean MADRS reductions of 43% at 12 months and 53% at 24 months, with response rates of 46% and 55% respectively; the cohort was deeply ill (mean ~8 years in the current episode, 88% having had ECT), and safety across 523 patient-years was acceptable. Improvement accumulated during open follow-up, but attrition, concomitant care, expectancy, and natural history limit causal interpretation.

Which target is best?

No adequately powered head-to-head trial exists, so comparison rests on indirect synthesis. A network meta-analysis of 22 trials (15 sham-controlled) ranked the medial forebrain bundle as associated with the greatest reduction in depressive symptoms, with a responder rate around 86%, exceeding the SCC and VC/VS, consistent with the rapid, reward-mediated effects seen in slMFB pilots (Schlaepfer; Coenen). A separate 2024 systematic review/meta-analysis (7 RCTs with 198 patients and 8 open-label trials with 77) estimated 47% long-term improvement in depression-scale scores and long-term response/remission rates of 48%/35%, found no significant difference by stimulation target and larger effects in open-label than in randomized trials, and emphasized that limited sham-controlled data still do not cleanly separate active stimulation from placebo effects. The target-ranking result is therefore hypothesis-generating rather than definitive: the MFB studies are small and the comparison indirect, and randomized head-to-head trials are still needed before declaring an optimal target.

Connectomic and adaptive stimulation

The field's contemporary direction follows directly from the trial lessons. Tractography-guided SCC targeting places the contact to capture the responder-associated fiber bundle in each individual rather than at a fixed coordinate. Biomarker-driven and adaptive (closed-loop) approaches aim to read a neural signature of the depressed state and deliver stimulation responsively, personalizing both where and when stimulation is applied. These strategies are early and unproven; they are the main reason interest in depression DBS has revived rather than faded, and TRANSCEND will test whether network-guided SCC targeting can succeed in a sham-controlled design.

Long before stimulation, the field treated these circuits by making a lesion, and it still does. Anterior capsulotomy and cingulotomy remain in selected use. Radiofrequency is an established lesion technique; the incisionless psychiatric evidence discussed here concerns capsulotomy by Gamma Knife or MR-guided focused ultrasound, not an interchangeable incisionless cingulotomy pathway. The evidence is strongest in OCD; the same lesions are applied, more investigationally, in depression. This part covers the open and incisionless lesion procedures and closes with how to choose among a lesion, an implant, and the incisionless options.

Procedure and rationale

Anterior capsulotomy places bilateral lesions in the anterior limb of the internal capsule, interrupting prefrontal-thalamic and prefrontal-brainstem fibers, the same corridor stimulated by ventral-capsule DBS. The classical approach is a radiofrequency thermal lesion created through a stereotactically placed electrode; ventrally weighted variants ("ventral capsulotomy") target the inferior capsule near the ventral striatum, mirroring the VC/VS DBS target. Because the lesion is permanent, the procedure rests on durable interruption of the pathologic circuit rather than ongoing modulation.

Efficacy and the size/dose lesson

Across the modern literature, anterior capsulotomy produces meaningful response in roughly half or more of severely refractory patients. In the Brown systematic review, mean YBOCS reduction at last follow-up was 57% for capsulotomy and 37% for cingulotomy, with full-response rates of 54% and 41%, respectively, but the evidence was observational and capsulotomy carried a higher adverse-event burden. The defining safety lesson comes from long-term follow-up: in Rück's series of 25 patients followed an average of nearly 11 years, capsulotomy was effective but carried a substantial risk of adverse effects (apathy, executive dysfunction, disinhibition, and weight gain) that correlated with larger lesions, higher radiation doses, and repeated procedures. The contemporary discipline that followed is explicit: make the smallest effective lesion, avoid high-dose or repeated lesioning, and weigh the frontal-behavioral risk in consent. This is the modern echo of the field's founding error: benefit and harm both live in the frontal lobe, and restraint in lesion size is the safeguard.

Anterior cingulotomy

Anterior cingulotomy places bilateral lesions in the dorsal anterior cingulate cortex and cingulum bundle, a different node of the same limbic network, and has a long track record (originally Ballantine, Boston) in both OCD and chronic pain. The Brown systematic review found both cingulotomy and capsulotomy effective, with lower average response but fewer serious or permanent adverse events after cingulotomy; because the data are observational and heterogeneous, they should not be read as a definitive head-to-head target ranking. Cingulotomy is sometimes favored as a first lesion because of a comparatively favorable side-effect profile, with escalation to a second procedure reserved for non-responders.

Subcaudate tractotomy and limbic leucotomy

Two combination/extension procedures complete the classical set. Subcaudate tractotomy lesions the substantia innominata below the head of the caudate (historically for depression). Limbic leucotomy combines cingulotomy with subcaudate tractotomy, addressing two nodes at once; it is reserved for the most refractory cases and carries correspondingly greater behavioral risk. A pragmatic staged philosophy underlies contemporary lesioning: begin with the smallest reasonable single lesion, assess response over months, and escalate deliberately rather than starting with the largest intervention.

Choosing between a lesion and an implant

For a patient who meets criteria for surgical OCD treatment, lesion and DBS are genuine alternatives engaging overlapping circuitry, and the choice turns on tradeoffs rather than efficacy alone:

• DBS offers adjustable and interruptible stimulation, valuable when the optimal effect is uncertain or adverse effects must be dialed back, at the cost of implanted hardware, infection and device-failure risk, battery and programming burden, and ongoing cost.
• Lesioning offers a durable one-time effect with no hardware, no infection risk from a foreign body, and no programming, at the cost of irreversibility and the size/dose-dependent risk of frontal-behavioral effects.
• DBS stimulation can be switched off or reprogrammed, but implantation itself can cause permanent injury, and explantation does not undo every effect. Practical determinants include the patient's ability and willingness to manage a device, access to expert programming, comorbidity and bleeding risk, and patient preference after honest counseling.

The incisionless lesion modalities, Gamma Knife radiosurgical capsulotomy and MR-guided focused ultrasound capsulotomy, preserve the no-hardware advantage while avoiding a transcranial electrode pass. Availability, regulatory status, latency, skull feasibility, evidence, and late risks differ substantially. Psychiatric MRgFUS remains investigational in the United States.

Stereotactic Radiosurgery (Gamma Knife) Capsulotomy

Mechanism and technique. Gamma Knife radiosurgical capsulotomy focuses many cobalt-60 gamma beams to deposit a high, conformal dose at the ventral anterior internal capsule, typically with paired isocenters ("double-shot") bilaterally. No incision or electrode is used; published psychiatric protocols generally use frame-based stereotactic targeting for a single session. Crucially, the lesion is not immediate: radionecrosis evolves over weeks to many months, and clinical benefit tracks this delayed biology. Published Gamma Knife capsulotomy protocols have used maximum doses of roughly 140–180 Gy. Pittsburgh reported 140–150 Gy with two 4 mm isocenters per side; Charlottesville reported 140–160 Gy with a single 4 mm isocenter per side; the sham-controlled Lopes trial used 180 Gy with two isocenters per side. Earlier Swedish protocols used 180–200 Gy, often with three or more isocenters, larger treatment volumes, or repeat capsulotomy, and those series reported more edema, delayed cyst formation, and radionecrosis. Lower-dose, smaller-volume techniques have been associated with fewer adverse radiation effects, but no single dose threshold separates safe from unsafe treatment: lesion volume and shot configuration matter as much as the prescribed maximum dose, and, as with radiofrequency capsulotomy, they govern both efficacy and risk.

Evidence. Gamma ventral capsulotomy has a small sham-controlled trial as well as observational follow-up. Read the corrected Lopes report: at 12 months, 2/8 active and 0/8 sham participants met the combined response definition; the response-rate difference was not statistically significant, although some YBOCS severity comparisons favored active treatment. The 2014 article was retracted and replaced in 2015 after an error in calculating the primary outcome (the original report had counted 3/8 active responders). Larger uncontrolled series report longer-term benefit, including 31/55 responders (≥ 35% YBOCS reduction) at 3 years in the Rasmussen cohort, but these do not convert the randomized result into definitive proof. Benefit can take months, and late complications require long-term surveillance: in the Rasmussen series 5 patients (9%) needed steroids for transient edema, 3 (5%) developed late cysts, and 1 was left in a minimally conscious state after radionecrosis.

MR-Guided Focused Ultrasound Capsulotomy

Mechanism and technique. MR-guided focused ultrasound (MRgFUS) uses a hemispheric phased-array transducer to converge ultrasound through the intact skull onto the anterior limb of the internal capsule, raising tissue to ablative temperature. Real-time MR thermometry measures heating and stepwise low-energy sonications help evaluate feasibility and location; they cannot acutely demonstrate an anti-obsessional or antidepressant effect, which develops over months. Cumulative thermal injury may also begin before the final sonication, so early exposures should not be promised as fully reversible. There is no incision, no radiation, and no implanted hardware.

Evidence and the skull-density limit. Small phase I and open-label studies suggest benefit in refractory OCD. Hamani and colleagues (2025) followed 15 OCD and 12 major-depression participants: mean YBOCS fell by 23% at six months and 35% at twelve months, whereas mean depression-score reductions (about 25%) at six and twelve months were not statistically significant. No serious adverse effects were registered in that cohort, but its size cannot establish safety or exclude uncommon delayed harm. These findings are encouraging for OCD and inconclusive for depression; they are not controlled proof of efficacy in either disorder. Skull density and morphology influence whether sufficient energy can reach a deep capsular target, and sonication can cause transient headache, dizziness, or neurological symptoms. FDA Exablate Neuro approvals do not include OCD or depression; psychiatric MRgFUS remains investigational in the United States.

Choosing Among the Modalities

RF lesion vs radiosurgery vs MRgFUS vs DBS. Capsulotomy modalities and ventral-capsule DBS can engage overlapping circuitry, but cingulotomy is anatomically distinct and no modality can be assumed to deliver an identical network effect. They also differ in reversibility, latency, adjustability, evidence, and adverse-event profile.

In practice, preference is only one part of modality selection. A hardware-averse patient may consider a lesion after reviewing irreversibility, target-specific evidence, regulation, latency, skull feasibility, and late risks. MRgFUS offers immediate thermal lesion visualization without radiation when feasible but is investigational for psychiatric indications; radiosurgery has delayed radiation biology; radiofrequency provides an established stereotactic lesion workflow; DBS remains adjustable but entails lifelong hardware care. Every case requires the same multidisciplinary governance.