Does stem cell therapy work for knees? Modest relief, uncertain repair

Stem cell therapy for knees does something measurable, but the effect is small and the confidence around it is low. The most thorough look at the question is a Cochrane review, Cochrane being an international group that pools trial results under strict rules, which gathered 25 randomised trials and 1,341 people with knee osteoarthritis, the gradual wearing down of the smooth cartilage capping the bones inside a joint [1].

Against a placebo injection, meaning a dummy injection with nothing active in it, the gain is modest. On a 0-to-10 pain scale, people who got the dummy injection averaged about 4.5 and people who got stem cells averaged about 1.2 points lower, near 3.3 [1]. The range around that estimate ran from a 2.5-point improvement all the way down to no improvement.

Movement followed the same shape. On a 0-to-100 questionnaire about everyday activity, where 0 means no difficulty at all, the dummy group scored about 46 and the stem cell group about 14 points better, near 32 [1]. Asked plainly whether their treatment had worked, about 53 in every 100 people said yes after a dummy injection, against about 68 in every 100 after stem cells, across 4 trials and 348 people [1]. The reviewers had very little confidence in that last figure.

Not one of those 25 trials measured whether the joint itself changed. Cochrane went looking for X-ray evidence of arthritis progressing or holding steady and found that no included study had assessed it [1]. So whether these injections rebuild a worn knee is a question the randomised evidence has not answered.

Smaller studies that did scan the joint have reported changes. In a randomised comparison of 48 people with early knee osteoarthritis, damaged cartilage volume in one part of the knee fell by about 50 cubic millimetres over six months in the group injected with cells taken from their own fat, and grew by about 36 cubic millimetres in the group given hyaluronic acid, a lubricating gel injection commonly used in arthritic knees [10]. A review covering 26 clinical studies and 610 patients in total found signs of cartilage protection or repair in 18 of the 21 studies that looked, and said outright that 610 patients is too few to settle it [5].

Stem cells are also the less studied choice next to the main alternative. A review of injection treatments for knee pain rated shockwave therapy and platelet-rich plasma, a concentrate spun down from a patient's own blood, as the best researched of the options, and judged the evidence for cell-based injections beyond platelet-rich plasma to be limited [14].

Serious harm was uncommon and did not clearly differ between groups: across 7 placebo-controlled trials, 5 of 219 people given a dummy injection and 4 of 242 given stem cells had a serious adverse event [1]. Cochrane still flagged that any needle entering a joint carries a small risk of septic arthritis, an infection inside the joint, and that the long-term risks of injecting cells meant to encourage growth are unknown [1].

The cells involved are usually mesenchymal stem cells, adult cells taken from fat, bone marrow or donated umbilical cord tissue that can develop into several tissue types, cartilage among them [3].

Three routes have been proposed for how they might help, and they are not the same claim:

• the injected cells become new cartilage cells
• they release signalling molecules that change how the joint's existing cells behave
• they damp down the low-grade inflammation that pushes osteoarthritis along [3][7][9]

Only the first would count as regrowing a knee, and which of the three actually happens in a person is unsettled.

Two findings pull in different directions. Several studies using arthroscopy, a small camera passed inside the joint, reported hyaline-like cartilage, the smooth load-bearing kind, appearing after injection [9]. A randomised, double-blind, dose-escalating study went the other way and injected only exosomes, the tiny membrane packets that cells release to carry chemical messages, without the cells themselves, and reported no adverse consequences plus some improvement in clinical scores and MRI scans [4].

The joint also has cells of its own. Mouse work has identified a rare population of cartilage progenitor cells sitting in the surface layer of the joint that responds to mechanical load: forced running increased their number, suspending the animals so the joint bore no weight reduced it, and deleting the cells made arthritis progress faster [8].

Cochrane put it more plainly. The mechanism by which stem cell injections might slow osteoarthritis has not been established [1].

The trials are small. Sample sizes ran from 6 to 252 people, and only two of the 25 had more than 100 participants [1]. Small trials produce unstable estimates, which is part of why the pain result ranges from meaningful relief to none.

"Stem cell therapy" is also not one procedure. The source of the cells, how they were prepared and how many were injected varied from trial to trial [1], and a review of 26 studies reached the same conclusion about the field as a whole [5]. Pooling those trials means averaging treatments that were not the same treatment.

One detail deserves weight: up to three larger randomised trials were run and then withdrawn before their results were reported [1]. Cochrane treated this as suspected publication bias, the concern that unflattering results stay in a drawer while flattering ones get published. If that is what happened, the published picture is better than the real one.

The evidence does not speak to everyone:

• participants averaged 51 to 66 years old, with symptoms lasting one to ten years [1], so none of this describes young people with sports injuries or children
• the main measurement point was three to six months [1], so how long any benefit lasts is largely unmeasured
• the work on the joint's own progenitor cells and mechanical loading was done in mice [8], not people
• the exosome study was a dose-escalation trial designed mainly to check safety [4], not to establish how well it works

Safety is under-measured rather than established. Serious events were too rare across the trials for the comparison to mean much [1], and the theoretical risks of a therapy designed to promote cell growth, or of using cells from a donor rather than the patient, have not been followed for long enough to rule out [1].

Longevity and regenerative medicine clinics offer stem cell injections for knees, and a first visit usually starts with a consultation and imaging to grade how far the arthritis has gone, followed by cells harvested from the person's own fat or bone marrow, or supplied from donated tissue, and injected into the joint. The useful question to bring is which cells, from where, and prepared how. Because the source, preparation and dose differed across the trials that have been run [1], two clinics using the same phrase may not be performing the same procedure, and the pooled results above describe an average of methods rather than any particular one.