Knee OA Injections: What Lasts at 12 Months?

Corticosteroid shots can calm knee osteoarthritis pain quickly. They do not look like the durable winner. A 2025 Bayesian network meta-analysis by Gupta et al., restricted to trials with at least 12 months of follow-up, ranked platelet-rich plasma plus hyaluronic acid (PRP+HA) first for pain and function. Corticosteroid alone ranked last among the evaluated injectables for 12-month pain—and below placebo in its probability of ranking, measured by SUCRA.

That finding does not prove that a steroid injection makes pain worse than saline at 12 months. SUCRA ranks the probability that treatments occupy each place; it is not the same as a direct mean-difference estimate. Still, it exposes a practical mismatch. Knee OA is chronic, while many injection decisions lean heavily on relief measured over weeks.

The 12-month ranking

Gupta and colleagues pooled 37 randomized controlled trials involving 5,089 patients with knee OA. Every included trial followed participants for at least a year. Using Bayesian random-effects network meta-analysis, they compared visual analog scale pain and WOMAC or IKDC function outcomes across injection types.

At 12 months, PRP+HA ranked first for both pain and function; PRP alone ranked second, HA alone third, HA plus corticosteroid fourth, and corticosteroid alone last. The analysis excluded studies at high risk of bias from its primary analysis. Its long follow-up requirement makes it unusually useful for the question patients usually mean when they ask whether an injection "works": will it still help next year?

The ranking is not an effect size. The published summary does not provide the mean differences and credible intervals needed to say how much better PRP+HA was than PRP alone, or whether every adjacent rank differed by a clinically meaningful amount. A first-place finish can reflect a modest edge. The direction of the ranking is informative; the size of the advantage remains essential clinical context.

Why guidelines and long-term rankings diverge

The apparent disagreement is mostly about time horizon, not a mystery hidden in a centrifuge. OARSI conditionally recommends corticosteroids for short-term pain control and recognizes HA benefits beyond 12 weeks. The American College of Rheumatology conditionally recommends corticosteroids and conditionally recommends against HA; both bodies described PRP as investigational because evidence was very low quality and preparations were not standardized, as summarized by Gupta et al.

That short-term role for corticosteroids remains real. The long-term analysis does not erase early analgesia. It says the benefit did not hold up well at 12 months relative to the other injectable options in that network.

Other pooled comparisons point in the same direction. A 2021 systematic review and meta-analysis by McLarnon et al. found PRP superior to corticosteroid for functional outcomes, and a 2024 meta-analysis of randomized trials by Wang et al. reported greater pain and function improvements with PRP than with corticosteroid. These reviews do not make PRP a cure for OA. They make repeated steroid injections a weak choice when the goal is durable symptom control.

PRP has a placebo-controlled signal

Injection procedures themselves can produce expectation effects, so placebo-controlled trials matter. In a 2022 randomized trial by Chu et al., PRP outperformed sham saline injection for pain and functional outcomes in people with knee OA. The trial reported statistically and clinically meaningful improvements with PRP versus the inert comparator.

One trial cannot settle the question on its own, but it answers an important objection: PRP's apparent benefit cannot be dismissed entirely as the theater of getting a needle in the knee. A broader 2025 systematic review and meta-analysis by Wang et al. also supported PRP's efficacy and safety across randomized trials.

Clinical improvement is more useful than a pretty biomarker graph. A 2025 meta-analysis by Bensa et al. found that PRP improvements met minimum clinically important difference thresholds and varied with platelet concentration. That supports clinical relevance, while also creating a problem for anyone trying to order "the best PRP" as though it were a single standardized product.

PRP is a category, not a recipe

PRP preparations differ in leukocyte content, platelet concentration, activation method, injection volume, and number of injections. A 2025 network meta-analysis by Yu et al. found that preparation protocols produced meaningfully different therapeutic profiles. Pooling these products under one PRP label can blur both the average benefit and the best protocol.

Some evidence favors leukocyte-poor PRP and more than one injection. A 2023 meta-analysis by Kim et al. found leukocyte-poor PRP and multiple-injection regimens superior to HA at follow-up of at least six months. A 2022 network meta-analysis by Abbas et al. likewise placed leukocyte-poor PRP favorably.

But the strongest direct test of leukocyte content did not confirm that preference. In a 2024 double-blind randomized trial by Romandini et al., leukocyte-poor and leukocyte-rich PRP showed no statistically significant difference in safety or efficacy for knee OA. The pooled analyses compare protocols across different studies; the blinded trial compared the two formulations directly. That design difference may help explain the disagreement, but it does not settle it. Patients should not treat "leukocyte-poor" as a proven upgrade.

More platelets do not automatically mean better treatment either. A 2024 clinical outcome study by Boffa et al. found a non-linear relationship between platelet concentration and outcome; excessively high concentrations may impair repair. The optimal concentration range remains undefined. Any clinic offering a simple "higher concentration equals better results" pitch has skipped the inconvenient part of the evidence.

Why PRP plus HA ranks first

PRP+HA led the 12-month network ranking, and a 2024 prospective, randomized, double-blind trial by Fossati et al. provides direct trial-level support for the combination in degenerative knee joint disease. The trial enrolled people with early-stage OA, a major limit on generalizability.

It is biologically plausible that HA could contribute lubrication and a scaffold-like joint environment while PRP supplies growth factors and alters inflammatory signaling. Those are hypotheses, not proof that either pathway caused patients' symptom changes. Human trials establish patient-reported pain and function outcomes; they do not yet establish a confirmed repair mechanism inside an osteoarthritic joint.

The early-disease restriction matters. A 2023 network meta-analysis by Xue et al. also focused on mild-to-moderate OA, Kellgren-Lawrence grades I–III. Current evidence therefore supports more confidence for early-to-moderate disease than for advanced grade III–IV degeneration. It does not establish that PRP+HA can substitute for a surgical discussion when structural disease is severe.

What this evidence can—and cannot—tell you

• For durable symptom control: The 12-month evidence direction favors PRP, especially PRP+HA, over corticosteroid alone.
• For a quick flare: Corticosteroids retain evidence for short-term pain relief, which explains their guideline role.
• For choosing a PRP protocol: Evidence does not identify a validated platelet concentration, leukocyte profile, activation method, injection volume, or treatment schedule.
• For advanced OA: Do not assume results from early-to-moderate disease apply to severe joint degeneration.
• For value: No cited study supplies formal cost-effectiveness comparisons among PRP, HA, corticosteroids, and downstream outcomes such as delayed knee replacement.

Industry sponsorship across the 37 trials in the long-term network analysis was not assessed in the available evidence, so it cannot be ruled out as a source of bias. Nor can rankings answer whether an expensive, usually out-of-pocket PRP series is worth its price for a particular person.

At 12 months, PRP+HA has the best ranking signal and corticosteroid has the worst. The evidence is moderately strong on that direction, but weak on the exact PRP recipe a patient should seek.

If a clinician proposes repeated corticosteroid injections as the default long-term plan, ask a narrower question: what evidence supports that strategy at one year? The current long-term network evidence favors PRP-based treatment for pain and function, while leaving protocol details and value for money unresolved.