Skip to the line map
Before the Needle

Can Stem Cell Injections Help Degenerative Disc Disease? What Studies Show

  • Updated November 14, 2026
  • Stop 6 of 6 on the Joints, tendons and spine line

Randomized trials of cell injections into painful discs have found small average improvements in pain and disability compared with sham or placebo. A 2026 meta-analysis of seven trial publications, two of which report the same trial, found those average differences were smaller than the thresholds its own authors cite for a change patients would notice, and two sham-controlled trials of autologous bone marrow products published in 2025 found no significant advantage over sham. The trials do not show that discs regenerate, long-term safety is not established, and no cell product is FDA-approved for degenerative disc disease.

What is degenerative disc disease, and when does it cause symptoms?

Spinal discs sit between the vertebrae and act as cushions. With age, discs lose water, height and structural integrity. “Degenerative disc disease” is the label used when those changes are thought to be the source of persistent low back pain.

The difficulty is that disc changes on MRI are common in people with no back pain at all. An abnormal scan does not, by itself, explain symptoms. The trials discussed here enrolled adults with chronic low back pain attributed to degenerative discs after clinical evaluation, not simply anyone with a worn disc on imaging. If your question is whether leg or back symptoms come from a herniated disc or from narrowing of the spinal canal, that diagnostic comparison is covered in spinal stenosis vs. herniated disc.

What are intradiscal cell injections designed to study?

An intradiscal injection places a cell preparation directly into the center of the disc through a needle, usually under imaging guidance. The hypothesis is that the cells, or signals they release, might reduce inflammation or support the disc’s remaining cells. The authors of the 2026 review describe these proposed mechanisms as largely hypothetical. The trials were designed to measure symptoms and function; demonstrated repair of disc tissue is a separate question.

The products studied were not the same thing:

  • allogeneic (donor) bone marrow-derived mesenchymal cells or mesenchymal precursor cells, in several trials
  • an allogeneic disc progenitor cell product tested at two doses, in one trial
  • autologous bone marrow concentrate, in one small trial

Pooling them answers a broad question (“do cell injections of this general kind outperform sham?”) but cannot tell you which specific product, if any, works.

What have randomized human studies found?

A recent pooled analysis of these randomized trials is a systematic review and meta-analysis in the Asian Spine Journal, published online March 17, 2026. Its literature search covered studies published up to April 30, 2025, so trials published after that date are not in it; two are described below. The review reports seven randomized controlled trials enrolling 594 patients, but two of those seven entries are not separate trials. The 2021 Noriega paper states that it presents long-term results, about 3.5 years after treatment, from the same randomized trial of 24 patients published in 2017, and both papers carry the same registration number, NCT01860417. The review lists them as two rows, so that trial’s patients appear twice, and the review’s figure of 594 patients cannot be taken as a count of distinct people. Subtracting the duplicate gives 570, but that is arithmetic on the review’s table, not a verified total of everyone enrolled. Five entries used a saline placebo; the two Noriega entries used a sham injection with a local anesthetic. Because some of the data feeding the pooled estimates below may not be statistically independent, as explained after the table, read them with caution.

Outcome Pooled difference, cells vs. control 95% confidence interval Heterogeneity Threshold the authors cite as clinically meaningful
Pain, visual analog scale (0 to 100 mm) −6.67 mm −9.31 to −4.02 I² = 29% About 15 to 20 mm
Disability, Oswestry Disability Index (0 to 100 points) −4.05 points −5.24 to −2.87 I² = 32% About 10 to 15 points

Negative numbers mean less pain or less disability in the cell groups. Both differences were statistically significant. Both were well below the clinically meaningful thresholds the authors themselves cite, and the authors state that the benefits may not reach levels considered clinically meaningful. A statistically significant average difference of 4 points on a 100-point disability scale means the groups differed little on average. A difference between group averages does not tell you how much any individual patient changed.

The same review also reports responder results, meaning the share of patients who reached a set improvement, and these look more favorable to cells. Patients in the cell groups had higher odds of a pain reduction of at least 30% (odds ratio 1.82, 95% CI 1.43 to 2.33, from 3 trials) or at least 50% (odds ratio 1.96, 95% CI 1.38 to 2.79, from 2 trials), and of a disability improvement of at least 10 points (odds ratio 2.13, 95% CI 1.61 to 2.82, from 2 trials). These rest on only two or three trials each, are subject to the same counting questions described below, and the authors did not present them as overriding the small average differences. Read together, they do not show that the average patient gains a clinically meaningful benefit, and the responder estimates are too thin to identify which patients, if any, would.

Timing matters too. In the pooled data, pain differences were not significant at 1, 3 or 6 months and appeared from 12 months onward; disability differences were significant from 3 months.

A note on reading the paper: the review’s abstract prints the pain result as “6.67” without a minus sign. The results section and Table 2 report it as −6.67. Table 2 also shows much larger “total” counts (for pain, 2,545) because it adds up measurements taken at several time points; the same people are counted more than once. The paper does not describe how it handled these repeated measurements, or how it handled trials in which two dose groups shared one control group; in one pooled responder analysis the shared control group appears to be counted twice. The Cochrane Handbook calls this kind of problem a unit-of-analysis error: reusing a shared comparator group “double-counts” its participants, and analyzing correlated data as if it were independent can make confidence intervals artificially narrow. How much it changes these particular estimates cannot be judged from the published report, so the pooled figures above are best read as a rough summary of small trials, not as precise estimates.

The review lists its own limits: modest trial sizes, with only one trial above 100 participants per group; three trials with some risk-of-bias concerns; subjective primary outcomes; too few studies to formally assess publication bias; and autologous evidence from a single small trial with three months of follow-up.

Two sham-controlled trials published after the review’s search. Both tested a person’s own (autologous) bone marrow products, which differ from the donor cell products that dominate the review:

  • The DREAM trial, a double-blind phase IIb study published in 2025, randomized 52 patients with moderate-to-advanced disc degeneration at up to three levels to injections of their own bone marrow-derived mesenchymal stromal cells or a sham procedure; 46 completed six months of follow-up. The cells were well tolerated, but the authors reported no significant clinical advantage over sham within six months, and the sham group showed slightly greater improvement in disability and physical quality-of-life scores.
  • A double-blind sham-controlled trial of bone marrow concentrate, published online July 17, 2025, included 63 patients (45 treated, 18 sham). It found no significant difference in its primary outcome, at least 50% pain relief, at 3, 6 or 12 months, and its authors concluded that the injection performed equivalently to sham; both groups had high success rates.

Neither trial evaluates every cell product, and finding no significant difference is not the same as proving two treatments identical. They do mean the 2026 review’s pooled estimates are not the full current picture, and the newer autologous evidence does not favor the injections.

Has imaging shown discs were regenerated?

The pooled estimates in the 2026 review cover pain, disability and adverse events. They are not pooled measurements of disc height, disc hydration on MRI or MRI grade, so they cannot be used to claim that injected cells rebuilt disc tissue. Imaging outcomes have to be read trial by trial, and the individual trials did not all measure them the same way.

Even when an individual trial reports an MRI change, two cautions apply. First, a change in an MRI signal is not the same as restored, functioning disc tissue. Second, a better scan does not prove less pain, and less pain does not prove a better disc. Structural and symptom outcomes need to be reported and interpreted separately.

What are the risks and unresolved questions?

In the pooled safety data, the review found no statistically significant difference between cell and control groups in treatment-emergent adverse events (odds ratio 1.07), serious adverse events (odds ratio 0.58) or discontinuation due to adverse events. Three trials reported mortality data, and no deaths were recorded in either group in those three; the other trials did not provide this information.

Those results come from relatively small trials with limited follow-up, so they cannot rule out uncommon or delayed harms. Any needle placed into a disc carries procedural risks, including disc infection, and the risks of a product depend on how it is made and handled. Open questions include which product and dose, if any, matter; whether any benefit lasts; and what happens over many years.

On regulation: FDA’s consumer alert on regenerative medicine products lists disc disease and back pain among uses that these products have not been approved for, and the FDA directory of approved cellular and gene therapy products, checked November 14, 2026, includes no product labeled for degenerative disc disease. Intradiscal cell products remain investigational in the United States.

How do investigational injections fit among established care options?

For chronic low back pain, standard evaluation looks first for causes that need specific treatment and for warning signs. New difficulty urinating or loss of bladder or bowel control, numbness around the genitals or anus, or weakness or numbness in both legs that is severe or getting worse need emergency assessment, as the NHS advises for sciatica; in the United States, go to an emergency department or call 911. Numbness or weakness that comes on suddenly in the face, an arm or a leg, especially on one side of the body, can be a sign of stroke; CDC advises calling 911 right away. Weakness in one leg that keeps getting worse over hours or days should be checked by a doctor promptly rather than waiting for a routine appointment. The full list of warning signs is in spinal stenosis vs. herniated disc. Once these have been ruled out, care is planned with a clinician; options to discuss include activity, exercise-based rehabilitation and symptom management. An investigational intradiscal injection, if considered at all, belongs in a registered clinical study rather than as a paid office procedure.

Can an intradiscal injection reverse degeneration? The current randomized evidence does not show that. It shows small average symptom differences against sham or placebo, below the authors’ cited thresholds for meaningful change, and no pooled imaging evidence of regeneration.

Does an MRI change prove reduced pain? No. Imaging and symptoms are separate outcomes, and disc findings are common in people without pain.

If you are weighing an offer, useful questions for a spine specialist include: What exactly is the product and who makes it? Is it being given under an FDA IND in a registered study? Which trial results apply to this product, and how large were the changes compared with sham? What are the procedural risks for my discs?

Every stop on the Joints, tendons and spine line

  1. Can Stem Cell Therapy Regrow Cartilage in Arthritic Knees?
  2. How Long Does a Meniscus Tear Take to Heal Without Surgery?
  3. Frozen Shoulder vs. Rotator Cuff Tear: Symptoms and Key Differences
  4. Tendinitis vs. Tendinosis: Why the Difference Matters
  5. Spinal Stenosis vs. Herniated Disc: How to Tell the Difference
  6. Can Stem Cell Injections Help Degenerative Disc Disease? What Studies Show

Reader comments

Leave a note on this stop

Your email address will not be published. Required fields are marked *