Selective Knee Denervation for Chronic Knee Pain: A Clinical Review

By Dr. Chris Lakhiani, MD, FACS · September 2026

1. Clinical Problem

1.1 Epidemiology

Chronic knee pain is one of the most common musculoskeletal complaints in adults over 50, with symptomatic knee osteoarthritis affecting an estimated 10–15% of older adults and rising in prevalence with obesity and longevity.⁵ Total knee arthroplasty (TKA) remains the definitive treatment for advanced structural disease, but a substantial minority — approximately 15–20% of patients — remain dissatisfied with persistent pain after a technically successful arthroplasty, and roughly 10% report persistent moderate-to-severe pain at one year.^6,7 The absolute number of patients with unresolved post-arthroplasty pain grows each year as arthroplasty volumes expand.

1.2 Limitations of the standard algorithm

The conventional stepwise algorithm — conservative care (weight loss, physical therapy, NSAIDs), intra-articular injection (corticosteroid, hyaluronic acid), and eventual arthroplasty — assumes that pain in the knee is primarily a function of joint pathology and that structural correction of the joint corrects the pain. In the majority of patients this holds. In a clinically important minority it does not. Patients whose pain has a substantial neural component — from irritation, entrapment, or neuroma of the genicular nerves innervating the joint capsule — may derive little benefit from further joint-focused intervention and may even experience persistent or worsened pain after arthroplasty.^3,8

1.3 The neuropathic contribution to knee pain

The genicular nerves are sensory afferents that transmit nociceptive information from the anterior knee capsule, retinaculum, periosteum, and adjacent soft tissues. Repetitive mechanical irritation, direct injury during surgery, entrapment in scar tissue, and stretch during arthroplasty can render these nerves themselves a primary generator of pain.^1,9 The result is a pain phenotype that is often described as sharp, burning, or lancinating rather than the deep ache of degenerative joint pain, is provoked by specific movements or pressure at predictable anatomic sites, and does not track linearly with radiographic joint findings. This subset of patients is under-recognized in orthopedic and pain practice because the standard workup emphasizes joint imaging and intra-articular response rather than nerve-directed diagnostic maneuvers.

2. Neuroanatomy of the Knee

2.1 The genicular sensory network

The anterior knee capsule is innervated by a retinacular network of sensory branches — the genicular nerves — that arise from the femoral, tibial, and common peroneal nerves and enter the capsule at predictable, anatomically defined locations. The foundational cadaveric work of Horner and Dellon (1994) mapped this network in detail and provided the anatomic basis for selective denervation as a surgical intervention.¹ Subsequent refinement by Franco and colleagues (2015) further characterized the anterior capsular innervation and informed contemporary block and ablation targets.⁹

2.2 The principal targeted branches

• Superomedial genicular nerve — a branch from the femoral nerve (nerve to vastus medialis), entering the anteromedial distal capsule adjacent to the vastus medialis muscle. One of the three targets in classical genicular RFA.
• Superolateral genicular nerve — a branch from the femoral nerve (nerve to vastus lateralis), entering the anterolateral capsule. The second classical RFA target.
• Inferomedial genicular nerve — a branch from the tibial nerve, running along the tibial epicondyle-shaft junction just below the medial joint line. The third classical RFA target.
• Vastus intermedius branch (recurrent branch to the anterior capsule) — a small motor-to-sensory branch approximately 2 cm above the patella in the midline, an important surgical target that is not reliably addressed by standard three-target genicular RFA.^1,9
• Saphenous nerve — infrapatellar branch — providing medial and anteromedial knee sensation. A common source of neuroma pain after medial arthroscopic portals, medial arthrotomy, or anteromedial hardware.¹⁰

2.3 Origins and clinical implications

The superomedial and superolateral genicular nerves derive from the femoral trunk. The inferomedial genicular derives from the tibial component of the sciatic. The inferolateral genicular (less commonly targeted because of its proximity to the common peroneal nerve and the risk of foot drop with ablation) arises from the common peroneal nerve.^1,9 These origins matter both for diagnostic block interpretation and for surgical planning: a block that captures the classical three targets but leaves the vastus intermedius branch or infrapatellar saphenous unaddressed may under-represent the peripheral contribution to a given patient's pain, and a surgical denervation that omits one of the anatomic contributors may leave residual pain from the un-denervated territory.

2.4 Landmarks used in imaging-guided intervention

For fluoroscopy-guided RFA, the classical three targets are approached at the junction of the femoral or tibial shaft and their respective epicondyles: superomedial (junction of medial femoral shaft and medial epicondyle), superolateral (junction of lateral femoral shaft and lateral epicondyle), and inferomedial (junction of medial tibial shaft and medial epicondyle).^4,11 Ultrasound-guided approaches use the periosteal reflection and adjacent vascular landmarks (the genicular arteries travel with the nerves) to improve accuracy. Contemporary anatomic work has, however, demonstrated that the actual nerves are frequently more proximal or in slightly different positions than the classical radiographic targets, which is one reason surgical exposure — where the nerve itself can be identified and treated — has renewed interest for durable denervation.^9,11

3. Patient Selection

3.1 Threshold criteria

• Chronic pain (≥3 months) that has not responded to a full course of appropriately supervised conservative management, including activity modification, physical therapy, and pharmacologic and injection-based options where indicated.
• Anatomically consistent pain distribution — pain that maps onto one or more genicular innervation territories rather than a diffuse joint-line or global joint pain pattern.
• Neuropathic pain features — burning, electric, shooting, or lancinating quality; tenderness or Tinel sign over a specific anatomic point; provocation by pressure or specific movement.
• Positive diagnostic genicular nerve block — the physiologic anchor of patient selection.

3.2 The diagnostic genicular block

A pre-operative diagnostic block of the targeted genicular nerves — typically performed under ultrasound or fluoroscopic guidance with a short-acting local anesthetic — is the single most important selection tool for surgical denervation. A meaningful reduction in pain during the block's effective duration provides a physiologic prediction of the response to durable denervation. Commonly cited thresholds are ≥50% pain reduction, with some centers using a more stringent ≥70–80% to improve the specificity of selection.^3,4 Practical guidance:

• Confirm the block is anatomically accurate (image guidance, expected sensory effect).
• Quantify pain response with a numerical rating scale before, at peak effect, and at 30–60 minute intervals.
• Ask the patient to attempt provoking maneuvers (stair climb, sit-to-stand) while the block is active; movement-evoked response is often more informative than resting pain.
• Failure of an anatomically well-executed block to reduce pain by 50% argues strongly against isolated peripheral denervation on that target and prompts consideration of central sensitization (see §3.4).

3.3 Additional selection considerations

• Recalcitrant OA in patients who are not candidates for arthroplasty, decline arthroplasty, or wish to defer it.
• Persistent post-TKA pain after infection, loosening, malalignment, and instability have been ruled out.
• Post-traumatic knee pain with a plausible anatomic nerve generator.
• Discrete post-surgical neuroma, particularly of the infrapatellar branch of the saphenous nerve.

3.4 Screening for central sensitization

A meaningful subset of patients presenting for evaluation of chronic knee pain — particularly those with prior failed procedures — carry a centralized pain component that will limit the durability of any peripheral intervention. The 2026 Raasveld-Eberlin framework provides a surgically operational construct for identifying centralized pain in the peripheral nerve population, using five criteria of which a <50% response to an anatomically appropriate diagnostic block is the operationally decisive finding.¹² Screening against this framework before offering surgical denervation is prudent; see the companion professional article, Diagnosing Centralized Pain After Peripheral Nerve Injury, for the full framework and integration guidance.

3.5 Contraindications

• Active joint or soft-tissue infection.
• Active inflammatory arthritis flare (rheumatoid arthritis, crystalline arthropathy) — treat the systemic disease first.
• Primary structural instability requiring structural correction (mechanical, not neural, problem).
• Insufficient pain reduction with an anatomically well-executed diagnostic block.
• Dominant centralized pain phenotype without a clear peripheral target.

4. Technique: Surgical Denervation

4.1 General setup

Selective knee denervation is an outpatient procedure performed under general or regional anesthesia, typically with a thigh tourniquet. The specific nerves targeted are dictated by the pre-operative pain map and the results of the diagnostic block(s); most patients undergo denervation of a subset of the anterior capsular network rather than all branches. Careful pre-operative marking of the pain distribution and the planned incisions in the awake patient improves precision.^2,3

4.2 Exposure and nerve identification

Small anatomically directed incisions are placed over each targeted nerve at the anatomic landmark established by Horner and Dellon and refined by Franco and colleagues.^1,9 The nerve is identified in the subcutaneous or subfascial plane depending on the branch and confirmed by direct visualization. Loupe magnification and, where available, a nerve stimulator can aid in identification and in distinguishing purely sensory from mixed branches.

4.3 The key technical maneuver: transection with proximal implantation

The defining technical difference between surgical denervation and radiofrequency ablation is the treatment of the transected proximal nerve end. In surgical denervation, after transection, the proximal end of the nerve is implanted into adjacent muscle — a receiving bed that discourages symptomatic neuroma formation by removing the transected end from the mobile, inflammatory subcutaneous plane and burying it in a well-vascularized, less-mechanosensitive environment.^2,3 This maneuver is the primary reason surgical denervation offers a more durable and lower neuroma-risk profile than percutaneous ablation, which does not address the proximal end at all.

4.4 Preservation of the joint capsule

Denervation is a soft-tissue, extra-articular operation. The joint capsule is not violated and the intra-articular environment is preserved. Motor function is likewise preserved because the targeted branches are sensory afferents.^1,2

4.5 Adjunctive maneuvers

For larger nerve targets — particularly the infrapatellar branch of the saphenous nerve when a symptomatic neuroma is present — targeted muscle reinnervation (TMR) or regenerative peripheral nerve interfaces (RPNI) may be added to reduce the risk of symptomatic recurrence.^13,14 TMR and RPNI have a strong evidence base for symptomatic neuroma management and are natural adjuncts when denervation is being performed for a patient whose pain is driven in part by a defined neuroma rather than pure genicular hypersensitivity.

4.6 Technical variants

Robey and colleagues (2023) described a contemporary approach to surgical genicular denervation for osteoarthritis in patients who had failed conservative care and were not TKA candidates, targeting the medial retinacular, lateral retinacular, and infrapatellar branch nerves with transection and muscle burial where feasible.³ Their published series (see §6) demonstrated meaningful pain and functional improvement with this technique. Modifications continue to be described, and the specific combination of branches denervated in any individual patient should reflect the pre-operative anatomic and physiologic workup rather than a fixed template.

5. Radiofrequency Ablation (RFA) — When and How It Fits

5.1 Technique

Genicular RFA is a percutaneous procedure performed under fluoroscopic or ultrasound guidance. A cannulated electrode is placed at each anatomic target and a lesion is created by radiofrequency energy. Three variants exist: conventional (thermal) RFA — typically 80°C for 60–120 seconds at each target site; pulsed RFA — non-thermal, with lower peak temperatures and less definitive lesioning; and cooled RFA — a larger, more spherical lesion produced by internally cooled electrodes that may improve capture of anatomically variable nerve positions.^4,15,16

5.2 Evidence base

Randomized data support superiority of genicular RFA over corticosteroid injection and hyaluronic acid injection for chronic OA-related knee pain at intermediate follow-up.^15,16,17,18 A sham-controlled RCT by Franco and colleagues demonstrated significant VAS reduction at 3–6 months in the active arm.¹⁷ The American Society of Regional Anesthesia (ASRA) has published a position on the role of genicular denervation for OA of the knee.¹⁹

5.3 Duration and role

Typical duration of RFA-mediated relief is 6–12 months, occasionally longer, with eventual regeneration and return of symptoms in most patients. This finite duration is both a limitation (need for repeat procedures) and a use case: RFA is well-suited as a bridging or trial procedure for patients considering surgical denervation, and for patients who prefer a lower-commitment intervention or wish to postpone or avoid a surgical procedure.^4,15

5.4 Comparison: Surgical Denervation vs. RFA

6. Outcome Data

6.1 Surgical denervation for osteoarthritis

Robey and colleagues (2023) reported a retrospective series of 21 patients (24 knees) undergoing surgical genicular denervation for primary knee osteoarthritis who had failed conservative care and were not TKA candidates.³ Mean VAS improved from 8.7 to 2.9, WOMAC from 69 to 32, and EQ-5D from 0.183 to 0.646. Approximately 92% of patients experienced clinically significant pain improvement and 75% reported meaningful functional improvement. The series represents the largest contemporary published experience with surgical genicular denervation specifically in the non-TKA-candidate OA population.

6.2 Selective denervation for post-arthroplasty pain

Dellon and colleagues published a series of patients undergoing selective knee denervation for persistent pain after TKA, with sustained pain relief at follow-up in the majority of appropriately selected patients — that is, patients with anatomically consistent pain, positive diagnostic block, and no confounding infection, loosening, or malalignment.^2,20 Subsequent series by Kachooei and others have reproduced these findings in the post-TKA population and further characterized the anatomy of the surgically relevant branches in this setting.²¹ Fine and colleagues have described the neuroma-pain phenotype specifically after knee arthroplasty, informing patient selection and adjunctive strategies.²²

6.3 RFA outcomes

The randomized trial by Choi and colleagues demonstrated significant VAS reduction after genicular RFA versus sham for chronic OA-related knee pain at 12 weeks and beyond.¹⁵ Iannaccone and colleagues examined cooled RFA and demonstrated efficacy for OA-related pain.¹⁶ Franco and colleagues added anatomic and clinical evidence supporting the technique.¹⁷ Nguyen and colleagues reported that cooled RFA outperformed intra-articular corticosteroid injection for chronic knee pain at 6 and 12 months.¹⁸ El-Hakeim and colleagues demonstrated efficacy of fluoroscopically guided conventional RFA for chronic OA-related knee pain.²³

6.4 Realistic framing

Not every patient responds. Careful selection — positive diagnostic block, anatomically consistent pain, absence of dominant central sensitization, appropriate joint condition — is the strongest predictor of durable benefit. Published series report clinically meaningful benefit in 70–90% of selected patients; the corollary is that 10–30% do not achieve their goals, and this figure should be shared explicitly with patients during consent. Most published data are retrospective; higher-quality prospective and randomized data on surgical denervation specifically are needed.

7. Practical Integration and Case Considerations

7.1 A sequenced approach

1. Conservative care — weight optimization, physical therapy, activity modification, pharmacologic management.
2. Intra-articular management — corticosteroid or hyaluronic acid injection where clinically appropriate.
3. Diagnostic genicular block(s) with image guidance to test the peripheral nerve contribution.
4. If positive (≥50% relief with anatomically well-executed block): trial of RFA or proceed to surgical denervation based on patient preference, expected duration required, and neuroma vs. genicular-hypersensitivity phenotype.
5. Concurrent central-focused care where indicated (behavioral pain intervention, SNRI or gabapentinoid trial, sleep and mood optimization) — particularly when partial centralization is suspected.¹²
6. Arthroplasty considered on its own merits for structural indications; denervation is a complement to, not a substitute for, appropriate joint replacement.

7.2 Illustrative composite cases

The following are illustrative composites, not real patients. They demonstrate how the framework applies in decision-making.

7.3 Case A — Non-TKA-candidate OA with strong block response

A 68-year-old with symptomatic bilateral knee osteoarthritis worse on the right. Twelve months of PT, NSAIDs, and two rounds of intra-articular corticosteroid injection have provided only transient relief. Body habitus and comorbidities make her a poor TKA candidate; she also declines arthroplasty. Pain maps to the anteromedial capsule with tenderness at the superomedial and inferomedial anatomic points. Ultrasound-guided diagnostic block of the superomedial, superolateral, and inferomedial genicular nerves produces 75% pain reduction for the expected duration of the local anesthetic. Plan: proceed with surgical genicular denervation, transecting the three targeted branches and burying the proximal ends into adjacent muscle.

7.4 Case B — Persistent post-TKA pain, moderate block response

A 61-year-old with persistent anteromedial knee pain 18 months after a technically well-executed right TKA. Serial workup rules out infection (normal inflammatory markers, negative aspirate), component loosening (normal imaging), and malalignment. Pain is anatomically consistent with the infrapatellar branch of the saphenous nerve and the superomedial genicular. Diagnostic block yields 60% pain reduction. Plan: selective surgical denervation of the involved branches; given the discrete post-surgical neuroma of the infrapatellar branch of the saphenous nerve, adjunctive TMR or RPNI is planned for that target. Patient counseled on expected 65–75% probability of clinically meaningful durable relief.

7.5 Case C — Chronic post-traumatic knee pain with signs of centralization

A 44-year-old with 3 years of pain after a work-related knee injury. Pain has spread beyond the initial injury zone to the entire anterior thigh and calf, with brush allodynia and cold hyperalgesia at examination. PHQ-9 in the moderate-severe range, fragmented sleep, and functional decline out of proportion to imaging. Diagnostic genicular block, anatomically well-executed, yields 30% pain reduction and no functional improvement during the block window. Plan: defer surgical denervation. Pivot to central-focused care: pain psychology referral (pain reprocessing therapy or CBT-CP), SNRI trial, graded PT with desensitization, sleep and mood optimization. Re-evaluate in 3–6 months; a peripheral procedure may become appropriate if the centralized component responds to treatment and a residual peripheral target is identifiable.

8. Limitations and Considerations

• Variable individual response. Even in well-selected patients, response varies; the block predicts but does not guarantee outcome.
• Anatomic variation. Not all genicular nerves are identifiable at surgery in every patient; anatomic variability in the branching pattern and courses is well-described.^1,9
• Regeneration. Long-term regeneration of transected nerves remains a theoretical concern even with muscle implantation; long-horizon data are limited.
• Data quality. Most published series on surgical denervation are retrospective or small prospective cohorts. Higher-quality RCTs on surgical denervation specifically are needed, particularly head-to-head comparisons with RFA and with structured non-operative care.
• Selection bias. Published outcomes reflect the practice patterns of nerve-focused surgeons operating on well-selected patients with positive diagnostic blocks; generalization to unselected populations is not appropriate.

9. Conclusion

Selective knee denervation is a technically established, evidence-informed option for a well-defined patient population — those with chronic knee pain of neural origin, positive diagnostic block, and appropriate joint condition. It complements rather than replaces conservative care and, where indicated, joint replacement. Appropriate patient selection, anchored by an anatomically well-executed diagnostic block and screening for central sensitization, is the strongest determinant of a durable, clinically meaningful outcome. For referring orthopedic, pain, and primary care colleagues, the operational message is straightforward: a patient with chronic knee pain, a plausible nerve generator, and inadequate response to a joint-focused algorithm deserves a nerve-focused evaluation before the algorithm is declared complete.