Knee osteoarthritis
Why steroid injections wear off, what the trials say about repeating them, and where genicular artery embolization honestly fits for Los Angeles and Orange County patients.
Editorially updated October 13, 2026. This educational article was prepared by SoCal Interventional using the cited primary and authoritative sources. It does not replace an individualized medical evaluation.
Quick answer: A cortisone (corticosteroid) injection into an arthritic knee usually helps for a few weeks. In the pooled trials, the benefit is moderate at one to two weeks, smaller by four to six weeks, and no longer measurable at six months. When the shots have stopped helping, the first step is to find out why, because “cortisone stopped working” can mean the arthritis has progressed, the pain is coming from something other than an inflamed joint lining, or the injections were only ever masking the problem. Genicular artery embolization (GAE) is an outpatient, image-guided option for selected patients with mild-to-moderate knee osteoarthritis whose pain persists despite conservative care. It is not a cure, it does not rebuild cartilage, and the blinded trial evidence is mixed. This article explains both sides so you can ask better questions at your next visit.
Patients in Orange County are evaluated at our Fountain Valley office; patients in the San Fernando Valley and Los Angeles are seen in Van Nuys and Tarzana. Call (310) 929-0026 or request a consultation.
Cortisone wears off because it only calms inflammation in the joint lining; it does nothing to the cartilage, bone or alignment that drive the arthritis. In the 2015 Cochrane review of 27 trials, the benefit over sham was moderate at one to two weeks, small by four to six weeks and undetectable by six months.
Corticosteroid injections reduce inflammation in the synovium, the lining of the joint. They do not change the cartilage, the bone, or the alignment of the knee, so once the drug is cleared the underlying arthritis is still there. The 2015 Cochrane review of intra-articular corticosteroid for knee osteoarthritis pooled 27 randomized or quasi-randomized trials with 1,767 participants and reported the benefit over sham or no treatment at each follow-up point. The authors rated the evidence as low quality for every outcome and wrote that whether there are clinically important benefits after one to six weeks “remains unclear”; they found no evidence of an effect at six months.
| Time after a cortisone injection | Pain benefit vs sham (standardized mean difference) | Function benefit vs sham |
|---|---|---|
| 1–2 weeks | Moderate (−0.48) | Small to moderate (−0.43) |
| 4–6 weeks | Small to moderate (−0.41) | Small to moderate (−0.36) |
| 13 weeks | Small (−0.22; confidence interval includes no effect) | No evidence of effect (−0.13) |
| 26 weeks | No evidence of effect (−0.07) | No evidence of effect (0.06) |
Source: Jüni P et al., Intra-articular corticosteroid for knee osteoarthritis, Cochrane Database of Systematic Reviews 2015 (CD005328); negative values favor corticosteroid. In plain language: a shot that helped for six weeks and then faded did exactly what the trials predict. That is not a sign that something went wrong with the injection, and it is not a sign that the next one will last longer.
Repeating cortisone every three months has been tested and did not hold up. In a two-year randomized trial published in JAMA in 2017, patients injected every 12 weeks lost more cartilage thickness than those given saline and had no significant pain advantage. A single injection for a flare is still reasonable; a standing schedule is not.
Many patients are told they can have a cortisone shot every three months. A randomized trial from Tufts Medical Center, published in JAMA in 2017, tested exactly that schedule. One hundred forty patients with symptomatic knee osteoarthritis and signs of synovitis received either triamcinolone or saline into the knee every 12 weeks for two years. Over those two years the triamcinolone group lost significantly more cartilage thickness (an average change of −0.21 mm versus −0.10 mm for saline) and reported no significant difference in knee pain compared with the placebo group. The authors concluded that the findings “do not support this treatment” for symptomatic knee osteoarthritis.
That trial is one study, and a single well-timed injection for a flare is still a reasonable tool. What the data argue against is using repeated cortisone as a long-term strategy once it has stopped giving meaningful relief. Each additional shot carries the same short time horizon and may add a small cost to the cartilage you are trying to protect.
“Cortisone stopped working” usually means one of four things: the arthritis has progressed to bone-on-bone, the pain is coming from something other than the joint lining, inflammation is still the driver but cortisone’s window is too short, or conservative care was never completed. Only the third group is the one GAE is studied in, so the cause needs sorting out first.
Before any new procedure is discussed, the pain itself needs a fresh look. In our experience, the phrase covers several different situations, and the right next step differs for each:
GAE can reduce knee pain from an inflamed joint lining for a year or more in most responders, a different time horizon from a steroid shot. It is an outpatient, image-guided procedure in which tiny particles reduce blood flow in the abnormal vessels feeding the synovium; relief typically builds over two to six weeks rather than immediately.
Genicular artery embolization targets the same inflamed lining that cortisone treats, but by a different route and for a different duration. An interventional radiologist enters an artery through a small puncture, guides a microcatheter to the genicular branches that supply the painful synovium, and injects tiny microspheres into only those abnormal branches to reduce the flow. The procedure is outpatient, uses local anesthetic with light sedation, and most patients go home the same day. Pain relief typically begins over two to six weeks as inflammation settles. Our side-by-side comparison of GAE, cortisone, gel injections and knee replacement covers the practical differences in setting, downtime and repeatability.
What the published series report:
Two features of this evidence matter when you are coming off cortisone. First, relief in the open-label series lasted a year or more in most responders, which is a different time horizon from a steroid injection. Second, most trial patients had already had injections; prior cortisone does not rule GAE out. We ask for a gap of four to six weeks between a steroid injection and embolization so that the pain baseline is clear.
GAE cannot rebuild cartilage, reverse arthritis, fix mechanical problems such as instability or a locked meniscus, or guarantee that knee replacement is avoided. Roughly four in ten patients in single-arm series did not reach the success threshold at one year, and the sham-controlled trials so far are mixed, so it is an option for selected patients, not a proven replacement.
This is the part most marketing leaves out, and it is the part that decides whether the procedure is right for you.
Side by side, cortisone gives fast, short relief measured in weeks, while GAE gives slower-onset relief that most responders hold for one to two years in open-label series. Cortisone is an office injection that costs cartilage when repeated; GAE is a one-hour outpatient procedure with mixed blinded evidence. Neither changes the arthritis itself.
| Cortisone injection | Genicular artery embolization | |
|---|---|---|
| What it treats | Synovial inflammation (drug effect) | Synovial inflammation (reduced abnormal blood flow) |
| Measured duration of benefit | Moderate at 1–2 weeks, small by 4–6 weeks, none at 26 weeks (Cochrane 2015) | In open-label series, most responders hold benefit at 1–2 years (CVIR 2024; JVIR 2025) |
| Effect on cartilage | Greater cartilage loss with injections every 12 weeks for 2 years (JAMA 2017) | No cartilage benefit; arthritis continued to progress on MRI (CVIR 2024) |
| Strength of evidence | Many trials, low-quality pooled evidence, short-term benefit only | Consistent single-arm series; mixed sham-controlled trials; large RCTs pending |
| Setting | Office visit, minutes | Outpatient procedure, about 1 hour; home the same day |
| Typical candidate | Short-term flare control at any stage | Kellgren–Lawrence 1–3 with pain for 6+ months despite conservative care |
The patients usually evaluated for GAE after cortisone are adults, typically 45 to 80, with mild-to-moderate knee osteoarthritis (Kellgren–Lawrence 1–3), knee pain for six months or more despite physical therapy, weight management, anti-inflammatories or injections, and a pain pattern that points to the joint lining. Active infection, bleeding disorders, severe arterial disease or poor kidney function rule it out.
We look for the profile that matches the published trials: adults, typically 45 to 80, with knee osteoarthritis graded Kellgren–Lawrence 1–3 on a recent X-ray, pain for six months or longer despite physical therapy, weight management, anti-inflammatories or injections, and a pain pattern that points to the joint lining. Active infection, an uncontrolled bleeding disorder, severe peripheral arterial disease, or kidney function too limited for contrast are reasons not to proceed. The evaluation is a single visit with X-ray review (an MRI, if you have one, is helpful); the candidacy checklist on our GAE treatment guide lists what to bring. Coverage questions are addressed separately in the GAE cost and insurance guide; some insurers still classify GAE for osteoarthritis as investigational, so benefits must be verified before scheduling.
In the pooled trials, the benefit is moderate at one to two weeks, smaller by four to six weeks, and no longer detectable at 26 weeks. Relief that fades after a month or two is the expected course, not a failed injection.
One two-year randomized trial found that injections every 12 weeks produced more cartilage loss than placebo with no pain advantage. A single injection for a flare is different from a standing schedule; discuss the trade-off with your treating physician.
Yes. Most patients in the published GAE series had prior injections. We ask for a four-to-six-week gap after the last steroid injection so the pain baseline is clear before embolization.
Not necessarily. If cortisone never helped at all, the pain may not be coming from the inflamed joint lining, and GAE targets the same lining. GAE is studied in patients whose pain responded briefly to injections and then returned, with mild-to-moderate arthritis on X-ray.
Single-arm series consistently report relief lasting a year or more in most responders, but the three sham-controlled trials published so far give mixed results, and larger randomized trials are still enrolling. We present GAE as an option for selected patients, not as a proven replacement for injections or surgery.
Call (310) 929-0026 or request a GAE consultation. Orange County patients are seen at our Fountain Valley office at 10900 Warner Ave, Suite 101A, Fountain Valley, CA 92708; Los Angeles and San Fernando Valley patients are seen in Van Nuys and Tarzana. Bring your most recent knee X-rays or MRI, the dates of any injections, and a list of what you have tried.
Medical disclaimer: This article provides general education and cannot determine individual candidacy or outcome. Discuss any change in treatment with your treating physician.
Explore more: Read the complete genicular artery embolization guide, compare GAE vs injections vs knee replacement, and review our Medical Editorial Policy.
Medical content reviewed by Nathanael Sabbah, MD, interventional radiologist. Last reviewed: October 2026.