What a Steroid Injection Actually Does
Corticosteroids are synthetic analogues of cortisol. Injected into a joint, they suppress inflammatory mediators — prostaglandins, leukotrienes and cytokines — which reduces pain and usually improves movement.
The best synthesis of how well this works is a Cochrane review of 27 randomised trials in 1,767 people with knee osteoarthritis [1]. Its findings, stated in full:
And the part that matters most for anyone deciding whether to keep going back:
Two honest qualifications. The review graded the evidence low quality for every outcome — the trials varied considerably (I [2]= 68%), pooled estimates were imprecise, and most had high or unclear risk of bias. And larger trials showed smaller effects, which usually indicates the pooled figure is optimistic.
So: a steroid injection produces a real, measurable, short-lived benefit, on evidence that is honest about its own limitations. Anyone who tells you it does nothing is misrepresenting the literature.
The Signal With Repeated Use
The two-year trial
The strongest evidence on repeated injection is a randomised, double-blind, placebo-controlled trial at Tufts Medical Center [2]. 140 patients with symptomatic knee osteoarthritis (Kellgren-Lawrence grade 2 or 3) and ultrasound-confirmed synovitis received 40 mg of intra-articular triamcinolone or saline every 12 weeks for 2 years.
- Cartilage: mean change in index-compartment thickness −0.21 mm with triamcinolone vs −0.10 mm with saline (between-group difference −0.11 mm; 95% CI −0.20 to −0.03)
- Pain: −1.2 vs −1.9 (between-group difference −0.6; 95% CI −1.6 to 0.3) — no significant difference, and numerically the saline group did slightly better
- Treatment-related adverse events: 3 with saline, 5 with triamcinolone
The authors' conclusion:
"Among patients with symptomatic knee osteoarthritis, 2 years of intra-articular triamcinolone, compared with intra-articular saline, resulted in significantly greater cartilage volume loss and no significant difference in knee pain. These findings do not support this treatment for patients with symptomatic knee osteoarthritis." [2]
One caveat the trial itself flags and that is frequently dropped: the minimal clinically important difference for cartilage volume has not been defined [2]. The 0.11 mm difference is statistically significant; whether a patient would ever notice it is not established.
The observational data
An analysis from the Osteoarthritis Initiative assembled a propensity-score-matched cohort of 684 participants (148 who began corticosteroid injections, 536 who did not) and followed radiographic progression over 4 years [3]:
- Kellgren-Lawrence grade worsening or knee replacement: HR 3.02 (95% CI 2.19–4.16); with continuous use, HR 4.67 (2.92–7.47)
- Joint-space-width worsening or knee replacement: HR 2.93 (2.13–4.02); continuous use 3.26 (1.78–5.96)
These are large effects. They are also observational, and the authors state their conclusion carefully: corticosteroids "may be associated with an increased risk of knee radiographic osteoarthritis progression" [3]. People who are offered repeated injections tend to have worse disease to begin with, and matching reduces but does not eliminate that.
What the cartilage-effects literature actually concludes
This is where the common summary — "steroids damage cartilage" — becomes too crude to be useful.
A systematic review of 40 studies of corticosteroid effects on articular cartilage concluded [4]:
"Corticosteroids have a time- and dose-dependent effect on articular cartilage, with beneficial effects occurring at low doses and durations and detrimental effects at high doses and durations. Clinically, beneficial effects are supported for IA administration, but the lowest efficacious dose should be used."
The laboratory work points the same way and is more specific than usually reported. A study of single clinical injection doses applied to human chondrocytes in a bioreactor found that betamethasone consistently and significantly reduced cell viability; triamcinolone did so in a single trial; and dexamethasone and methylprednisolone did not [5]. The finding is drug-specific, not a property of the whole class.
The reasonable reading: dose, frequency and agent matter, and an occasional well-indicated injection is a different proposition from a standing appointment every three months.
Tendons Are a Different Question
Intra-articular injection into a joint and injection around a tendon are not the same intervention, and they should not be reasoned about together.
The definitive review of injection therapy for tendinopathy pooled 41 trials and 2,672 participants [6]. For lateral epicondylalgia (tennis elbow), corticosteroid injection had a large short-term benefit — SMD 1.44 (95% CI 1.17–1.71) versus no intervention — and that effect reversed at intermediate and long term, with corticosteroid-treated patients doing worse than comparators at 26 and 52 weeks.
The authors add a warning that is routinely ignored when this paper is cited:
"Response to injection should not be generalised because of variation in effect between sites of tendinopathy." [6]
So the reversal is well documented at the elbow. Extending it to every tendon in the body goes beyond what the review supports.
How Cell Therapy Is Thought to Work
Mesenchymal stem cells (MSCs) are not proposed to work by replacing tissue. The current understanding is that they act by signalling:
- Immunomodulation — regulating rather than simply blocking the inflammatory response, including secretion of anti-inflammatory cytokines [7]
- Paracrine signalling — a secretome of growth factors and cytokines that is now regarded as the primary mechanism of action [8]
- Anti-apoptotic effects — protecting existing chondrocytes [9]
- Extracellular vesicles — exosomes carrying microRNAs and proteins; the evidence here is from preclinical animal studies [10]
That is a different mechanism from corticosteroid suppression. A different mechanism is not by itself a better outcome, and the sections below are about outcomes.
One distinction that matters for reading the trials
Trials in this field use two quite different products, and their results do not transfer to one another.
Autologous cells are taken from the patient — usually from fat or bone marrow — expanded in culture, and returned. Allogeneic cells come from a donor, most often umbilical cord tissue, and are manufactured in batches.
The rationale for donor cord cells is that they can be prepared in advance and that laboratory work has found Wharton's-jelly-derived cells to suppress stimulated lymphocyte proliferation and to be poor stimulators of an allogeneic immune response — properties described by their investigators as "permissive to allogeneic transplantation" [20]. That is an in vitro characterisation, not a demonstration that donor cells are safe or effective without matching in patients, and it should not be read as one.
The practical consequence: the largest positive trial used autologous adipose cells [11], while the repeat-dosing trial used allogeneic umbilical cord cells [15]. A result from one is not evidence for the other.
What the Cell-Therapy Evidence Shows
This literature does not speak with one voice, and a page that pretended otherwise would be doing exactly what this one is trying to avoid.
The largest randomised trial: positive
A phase III, multicentre, double-blind randomised controlled trial (Level 1) enrolled 261 patients with Kellgren-Lawrence grade 3 knee osteoarthritis and gave a single injection of autologous, culture-expanded adipose-derived MSCs or placebo [11].
At 6 months:
- 100-mm VAS pain improvement: 25.2 vs 15.5 (P = .004)
- Total WOMAC improvement: 21.7 vs 14.3 (P = .002)
- Significantly higher proportions of patients exceeded the minimal clinically important difference on both measures
- No serious treatment-related adverse events
- MRI showed no significant difference in change of cartilage defects between groups
The authors' own framing: the injection "provided significant pain relief and functional improvements", and "long-term results are needed to determine the disease-modifying effect" [11].
Note what this trial was: autologous cells from the patient's own fat, given once, followed for six months.
The pooled randomised evidence: null against placebo
A meta-analysis of 13 Level-I randomised trials reached the opposite conclusion, and its title states it plainly — Intra-Articular Mesenchymal Stromal Cell Injections Are No Different From Placebo in the Treatment of Knee Osteoarthritis [12]:
- VAS pain: MD 1.62 (95% CI −0.60 to 3.85) — not significant
- WOMAC pain: MD 1.88 (−0.21 to 3.98) — not significant
- WOMAC function: MD −0.67 (−6.54 to 5.19) — not significant
- On every outcome, the smallest treatment effect failed to exceed the minimal clinically important difference
- No difference in the proportion of patients crossing the MCID threshold (VAS RR 0.93, 0.55–1.57)
The same analysis found MSC injection significantly better than hyaluronic acid (VAS MD 2.00, 0.94–3.07).
Both findings are real. The meta-analysis searched to April 2020 and so predates the 261-patient trial; the trial is larger and better controlled than most of what the meta-analysis pooled, but ran for six months. Anyone claiming this question is settled — in either direction — is ahead of the evidence.
The broader reviews
A systematic review of 17 studies (6 randomised, 8 prospective observational, 3 retrospective) concluded that intra-articular MSCs "provide improvements in pain and function in knee osteoarthritis at short-term follow-up", at Level III evidence, with a mean methodological quality score of 55.5 ± 15.5 out of 100 [13].
The most recent systematic review, covering 19 randomised trials of adipose-derived cell therapies published between 2015 and 2025, found clinically meaningful pain and function improvements and no serious treatment-related adverse events — while reporting that although some trials showed favourable MRI findings, "consistent cartilage regeneration was not demonstrated" [14].
On repeat dosing
A controlled randomised phase I/II trial compared a single dose of umbilical-cord-derived MSCs, two doses six months apart, and hyaluronic acid in knee osteoarthritis [15]. The repeated-dose group (20 × 10⁶ cells at baseline and at 6 months) achieved significantly lower pain than hyaluronic acid at 12 months and outperformed the single dose. It is a small trial — 26 patients across three arms — and it tested that specific schedule, not any other.
Where the evidence does not support the claim
For partial-thickness rotator cuff tears, a randomised controlled trial compared stem cells in fibrin glue, a saline/fibrin glue mixture, and saline alone in 24 patients [16]. The primary outcome — pain on activity at 3 months — did not differ between groups (P = 0.35), and the authors concluded that stem cell injection "was not more effective than control injections".
For hip osteoarthritis, the most cited cell-therapy report is an uncontrolled cohort of 10 patients given autologous bone-marrow MSCs in three weekly doses [17]. It reported improved hip scores and radiographic stability in 9 of 10 patients. An uncontrolled series of ten cannot establish efficacy, and it did not use MRI.
The Comparison Nobody Has Run
There is no randomised trial comparing intra-articular stem cell therapy with corticosteroid injection for knee osteoarthritis. Any table placing them side by side — including the one below — is assembling separate literatures, not reporting a comparison.
The honest summary of the row that matters most: neither treatment has been shown to rebuild cartilage. One has a documented association with faster structural loss; the other has repeatedly failed to demonstrate structural gain.
The one direct comparison that does exist
A European consensus (ESSKA-ORBIT) of expert clinicians and scientists issued a grade-A recommendation — its highest evidence tier, with very strong agreement — stating [18]:
"PRP injections have been shown to provide a longer effect in comparison to the short-term effect of CS injections. They also seem to provide a safer use profile with less potential related complications."
This is the strongest formal statement available that an injectable orthobiologic outlasts corticosteroid. It is about platelet-rich plasma, not stem cells, and the distinction is worth holding onto: PRP and MSC therapy have separate evidence bases, and PRP's is currently the better developed for knee osteoarthritis. Even there the picture is not uniform — a 288-patient placebo-controlled trial found PRP no different from saline on either symptoms or cartilage volume at 12 months [19].
Where Each Treatment Reasonably Fits
Corticosteroid injection is a reasonable choice when:
- Rapid relief of an inflammatory flare is the goal, and a few weeks to three months of benefit is worth having
- The injection is being used diagnostically, to confirm the source of pain
- The underlying condition is inflammatory arthritis, where suppressing inflammation treats the disease rather than masking it
- Cost or access rules out other options — this is a legitimate clinical consideration, not a failure of ambition
The case for caution is about frequency, not existence: the structural signal is associated with repeated and continuous use [2,][3], and the cartilage literature points to using the lowest effective dose for the shortest period [4].
Cell therapy may be worth discussing when:
- Disease is mild to moderate rather than end-stage
- Steroid injections are giving progressively shorter relief
- The person understands they are choosing an investigational treatment with mixed randomised evidence, no demonstrated structural benefit, and a substantially higher cost
It is not a reasonable choice if the expectation is cartilage regrowth. No trial has shown it.
Questions Worth Asking
- "What outcome are we treating — pain, function, or joint structure?" These are different, and only the first two have been reliably improved by any injection.
- "What does the evidence say for my grade of disease?" The largest positive cell-therapy trial enrolled Kellgren-Lawrence grade 3 specifically [11].
- "How long is the benefit expected to last, and what happens after?"
- "How many steroid injections have I had, and over what period?" The structural concern attaches to cumulative, continuous use.
- "Is this treatment licensed for this use, or investigational?"
- "What is the evidence that this specific product and schedule works?" — cell source, autologous or allogeneic, dose and interval all vary between trials, and results from one do not transfer to another.
The Bottom Line
Corticosteroid injection does something real: a measurable reduction in pain and improvement in function, on low-quality but consistent evidence, fading to nothing by about six months [1]. Used repeatedly over years, it carries a documented association with faster structural loss and, in the one two-year controlled trial, no pain advantage at all over saline [2].
Cell therapy has produced a significant benefit in the largest randomised trial conducted [11]and no benefit over placebo in the pooled analysis of thirteen earlier ones [12]. It has a good safety record [14]. It has not been shown to rebuild cartilage, and the trials that looked for that have not found it [11][14][19].
Nobody has run the trial that would actually answer the question in this article's title. Until someone does, the choice is between a cheap, licensed, short-acting treatment with a structural caveat attached to repeated use, and an expensive, investigational one with genuinely mixed evidence and a better safety-versus-structure profile — made with a clinician who knows the specific joint, the specific grade, and the specific history.