Recognising the point of diminishing returns

A useful rule of thumb sits at 12 to 18 months. If a knee injection is still delivering meaningful relief a year or more after it was given, repeating it is a reasonable clinical decision. If relief lasted only a few weeks — or arrived only partially — repeating the same injection is unlikely to perform better next time. That gap in response is the clearest signal that escalation, rather than repetition, is the appropriate next move.

Corticosteroid (cortisone) injections are designed for short-term symptom control. Pooled evidence places typical benefit somewhere between a few weeks and a few months; the anti-inflammatory effect fades as the drug clears, and nothing in the mechanism addresses the structural changes driving late-stage osteoarthritis. Cartilage lost to erosion, or a joint where bone is already contacting bone, does not respond to chemical dampening the way an acutely inflamed joint might. No intra-articular injection can restore structurally lost cartilage — the limit is the joint's condition, not the injection itself.

Continuing to inject a knee that is no longer responding also carries its own risk. In a randomised controlled trial, patients who received a corticosteroid injection every three months for two years showed greater cartilage loss on MRI than those given a placebo — with no additional pain benefit. Stacking up steroid injections in a non-responsive knee is not a neutral holding strategy; it may accelerate the structural decline it was intended to manage.

Recognising this point is not a sign that all options are exhausted. It is the appropriate moment to ask what comes next.

PRP: the strongest evidence for a next-step injection

Among the injection options available after corticosteroid failure, platelet-rich plasma (PRP) has the most substantial evidence base. A 2025 meta-analysis of 18 randomised controlled trials covering 1,995 patients found that PRP produced statistically superior improvements in both pain (VAS) and functional outcomes (WOMAC) compared with placebo, exceeding the minimal clinically important difference on both measures at three to six months — meaning the average patient in those trials noticed a real-world change, not merely a statistical one.

Two modifiable factors meaningfully influence outcomes. The first is platelet dose: across 29 RCTs, treatment arms achieving positive results at six months averaged approximately 5,500 ×10⁶ platelets, compared with roughly 2,300 ×10⁶ in arms that fell short (p<0.01). Protocol standardisation across providers remains unsettled, so the dose a patient receives in practice can vary. The second factor is combination therapy: pairing PRP with hyaluronic acid outperforms PRP alone across 11 RCTs (n=1,023) on pain, function, and Lequesne scores, while also reducing adverse events (relative risk 0.41). Where the combination is available, it represents a reasonable default over PRP monotherapy.

Patient selection substantially determines the likely ceiling of benefit. High BMI and KL grade 3–4 radiographic severity — KL grade is a standard radiographic scale of arthritis severity running from 0 (normal joint) to 4 (bone-on-bone contact) — are independent predictors of PRP failure. In a study with a mean follow-up of 51 months, patients with lower BMI, younger age, and mild-to-moderate OA had significantly longer therapeutic benefit and an overall failure rate of only 15.3%. For KL grade 3–4 disease or high BMI, the evidence for PRP becomes considerably less certain, and other escalation pathways deserve earlier consideration.

PRP offers a longer duration of effect and a different mechanism from corticosteroid, making it a genuine step up rather than a lateral substitution — provided the patient's disease stage and profile are appropriate.

BMAC and scaffold injections for PRP non-responders

For patients who have not responded adequately to PRP, two quite different options occupy the next tier of the escalation pathway: bone marrow aspirate concentrate (BMAC) and polyacrylamide hydrogel. They work through entirely distinct mechanisms, which is why distinguishing them matters clinically.

BMAC is an orthobiologic — concentrated cells and growth factors drawn from the patient's own bone marrow — and its broad aim is regenerative signalling to the joint rather than mechanical support. The evidence base, however, is considerably thinner than for PRP. Published case-level data includes a 65-year-old patient who had already failed corticosteroid injection and was facing total knee replacement; following two BMAC injections, he reported 90% pain improvement within three weeks and, at four months, was hiking for several hours daily without pain. Cases of this kind are clinically instructive, but they do not constitute the RCT-level evidence available for PRP. BMAC is appropriately treated as a plausible next step when PRP has not achieved adequate relief — particularly in patients who wish to delay or avoid surgery — rather than as an established standard pathway.

Polyacrylamide hydrogel (Arthrosamid/ipaag) sits in a different category entirely. It is not a biologic and does not act on inflammation; it is a structural scaffold that integrates into the synovial membrane, providing mechanical cushioning within the joint space. Open-label studies confirm safety and effectiveness at 12 months. Its most clinically distinctive finding, however, is preliminary evidence that a single injection may reduce bone marrow lesions in advanced arthritis — a structural effect that neither hyaluronic acid, corticosteroid, nor PRP has consistently demonstrated, and one that positions it as a different kind of option rather than simply the next injection on the list.

Genicular artery embolization as a procedural bridge

Unlike the options covered so far, genicular artery embolization (GAE) is not a knee injection at all. It is a minimally invasive interventional radiology procedure in which tiny particles are delivered via catheter to selectively block abnormal blood vessel growth in the inflamed synovial lining — a process called neovascularisation. The target is the pain-driving inflammatory tissue itself, rather than the joint space.

Open-label data are encouraging. A meta-analysis of 14 studies covering 510 patients found a pooled pre-to-post pain reduction of approximately 30 points on a 0–100 scale at six to twelve months, with 78–92% of patients reaching a clinically meaningful response by 12 months. A prospective single-arm trial reinforced both the magnitude and the mechanism: VAS pain fell by 55% at 12 months alongside measurable reductions in serum VEGF and IL-1Ra — inflammatory markers — suggesting the effect goes beyond symptom masking.

The important caveat is that three sham-controlled RCTs have so far produced mixed results. How much of the benefit seen in open-label series reflects a genuine placebo-subtracted treatment effect, and how much reflects natural disease fluctuation or expectation, has not been resolved. This uncertainty is relevant to shared decision-making and worth discussing openly with any clinician recommending the procedure.

Patient selection substantially shapes the likelihood of benefit. Across 236 patients in one retrospective series, clinical success at one year was associated with younger age, lower KL grade (mean 2.8 in responders versus 3.2 in non-responders), and preserved joint-space narrowing — findings consistent with the broader pattern that GAE performs best in mild-to-moderate OA. Patients with KL grade 4 or bone-on-bone contact are considerably less likely to respond, and the evidence supports earlier referral rather than treating GAE as a last resort before arthroplasty.

Physical and structural options that sit alongside escalation

Every injection discussed in this article acts on a joint that still has to bear weight, transmit force, and absorb load with each step. How well that mechanical environment is managed determines, in part, how long any injection's benefit lasts — which is why physical and structural strategies become more relevant at the escalation stage, not less.

Quadriceps and hamstring strengthening is the most consistently underutilised of these adjuncts. Stronger periarticular musculature reduces compressive load across the joint surface; combined with an injection that has quietened the inflammatory environment, the benefit on pain and function is often additive rather than merely complementary.

Unloader bracing works through a different but equally direct mechanical principle: redistributing force away from the affected compartment. For unicompartmental OA, a well-fitted unloader brace can extend the functional window that a PRP or scaffold injection opens, without adding any pharmacological burden.

Body weight compounds across thousands of daily steps. Even modest reductions are enough to materially improve injection response and functional outcomes — which frames weight management not as background advice but as a clinical lever at this stage of the pathway.

For patients with measurable malalignment and predominantly unicompartmental disease, knee osteotomy is worth considering before arthroplasty enters the picture. It corrects the biomechanical cause of accelerated compartment wear rather than managing its symptoms — and in appropriately selected younger patients, may defer or avoid joint replacement altogether. That makes it structurally distinct from the injectables and from replacement surgery.

Arthroscopic debridement, by contrast, has limited evidence for general OA pain and is not recommended as a routine escalation step. It may have a specific role when a structural finding such as a loose body is identified, but it is not a substitute for the options above.

Why escalation timing changes the options available

Timing shapes what remains possible. An orthopaedic planning constraint that often goes unmentioned until it matters — a mandatory wait of at least three months after a corticosteroid injection before knee replacement surgery can proceed, to reduce periprosthetic infection risk — means that allowing repeated failed injections to run on unchecked can silently compress the pre-operative window. Waiting for injections to completely fail before beginning an escalation conversation is not the cautious choice it might feel like; it is a decision that forecloses other options by default.

The evidence across every modality covered in this article points in the same direction: earlier-stage disease consistently predicts better outcomes. That pattern holds for PRP, for polyacrylamide hydrogel, and for GAE alike. Referral at a point of partial response — rather than at the point of total failure — preserves access to options that are progressively less effective as structural deterioration advances.

It is also worth being clear about what these options are. None of them reverses established structural damage. They are symptom-delaying strategies, some with meaningful durability, but not disease-modifying treatments in the established sense. Setting that expectation clearly is part of an honest clinical conversation.

An early specialist assessment does not commit a patient to any procedure. It maps the remaining options, clarifies which modalities still apply, and allows decisions to be made unhurriedly — before the available choices narrow.

1. [1] A Prospective Single-Arm Trial of Genicular Artery Embolization for Symptomatic Knee Osteoarthritis: Clinical and Biomarker Outcomes. (2025). https://doi.org/10.1016/j.jvir.2025.08.009 https://doi.org/10.1016/j.jvir.2025.08.009
2. [2] BMAC Injection Improves 65-Year Old's Knee Osteoarthritis Pain. (2023). https://doi.org/10.33425/2639-9512.1080 https://doi.org/10.33425/2639-9512.1080
3. [3] Bridging the Gap between Injections and Surgery: Meta-Analysis of Genicular Artery Embolization in Knee Osteoarthritis. (2025). https://doi.org/10.1016/j.acra.2025.05.011 https://doi.org/10.1016/j.acra.2025.05.011
4. [4] PRP Injections for the Treatment of Knee Osteoarthritis: The Improvement Is Clinically Significant and Influenced by Platelet Concentration: A Meta-analysis of Randomized Controlled Trials. (2025). https://doi.org/10.1177/03635465241246524 https://doi.org/10.1177/03635465241246524
5. [5] A Higher Platelet Dose May Yield Better Clinical Outcomes for PRP in the Treatment of Knee Osteoarthritis: A Systematic Review. (2024). https://doi.org/10.1016/j.arthro.2024.03.018 https://doi.org/10.1016/j.arthro.2024.03.018
6. [6] Meta-analysis: PRP + HA vs PRP monotherapy for knee osteoarthritis (KOA). (2025). https://doi.org/10.1186/s13018-024-05429-w https://doi.org/10.1186/s13018-024-05429-w
7. [7] Clinical outcome and risk factor predictive for failure of autologous PRP injections for low-to-moderate knee osteoarthritis. (2021). https://doi.org/10.1177/23094990211021922 https://doi.org/10.1177/23094990211021922
8. [8] Early intervention in knee osteoarthritis with genicular artery embolization is associated with improved clinical outcomes. (2025). https://doi.org/10.1007/s00330-025-11702-1 https://doi.org/10.1007/s00330-025-11702-1