Bone-on-Bone Shoulder Arthritis: PRP, Stem Cells, and Shoulder Replacement

Marc Darrow, MD, JD

When someone tells me, “My shoulder is bone-on-bone,” I want to know what that means in that person’s life. Is the pain keeping you awake? Can you put on a shirt, wash your hair, or reach a shelf? Have you stopped playing golf, or are you having difficulty with the basic movements that let you live independently?

Many people who come to our office have already tried medication, physical therapy, or cortisone injections and have been told that shoulder replacement is the next step. They want to understand whether there is another reasonable option. My goal is to help them understand the source of their pain and what improvement is realistically possible.

PRP or a bone marrow injection may be considered for symptom relief in selected arthritic shoulders. Neither has been established as a reliable way to rebuild a bone-on-bone joint or produce results equivalent to shoulder replacement. The decision depends on the condition of the joint, the rotator cuff, overall health, and what the patient needs the shoulder to do. (1) (2)

What does bone-on-bone shoulder arthritis mean?

The glenohumeral joint is where the ball of the upper arm meets the shoulder socket. Osteoarthritis involves loss of its cartilage surface, with changes that may also include bone spurs, stiffness, and altered joint shape. “Bone-on-bone” usually describes severe joint-space narrowing on an X-ray.

That finding deserves attention, but a shoulder is more than its cartilage. The rotator cuff, biceps tendon, capsule, ligaments, and shoulder-blade muscles contribute to comfortable movement. The small acromioclavicular joint on top of the shoulder can also be painful. Some pain felt around the shoulder comes from the neck.

I want the examination to explain the symptoms, not simply repeat the wording of an imaging report. At the same time, when severe arthritis matches deep joint pain, grinding, and substantial loss of movement, it may be the principal problem. An injection aimed at a different structure will not change that mechanical limitation.

Why I take the question about avoiding surgery seriously

I had arthroscopic shoulder surgery years ago. It did not relieve my symptoms, and rehabilitation took about a year to bring the shoulder back to how it had felt before the operation.

Another experience stayed with me. An elderly man came in for right shoulder pain. When I asked him to raise his arms, his left arm barely moved. He explained that a previous left shoulder replacement had dislocated and could not be put back into position. He wanted to avoid a similar outcome on the right.

These are personal and clinical experiences, not estimates of how often surgery fails. They explain why I listen carefully when someone wants to discuss alternatives. They also reinforce why treatment discussions must include both the possible benefit and the possibility of a disappointing result.

There are patients we can help substantially, patients we can help only a little, and patients whose structural damage is too advanced for an injection to provide a useful result.

What can be tried without shoulder replacement?

Activity changes and physical therapy

A nonsurgical plan can include reducing repeatedly painful loads, adjusting work or household tasks, and maintaining tolerable movement. Therapy can address shoulder-blade control, remaining rotator cuff strength, and stiffness. Exercises need to fit the joint rather than force a severely restricted shoulder through painful motion.

The aim is better use of the arm. Exercise does not replace missing cartilage, and increasing pain despite a suitable program is a reason to reassess the plan.

Medication and cortisone

Pain medication and corticosteroid injections may provide symptom relief, but their benefits, risks, and duration need review. Repeated injections should not become an automatic response to a shoulder that continues to deteriorate. Medication choices also depend on other conditions and medicines, particularly in older adults.

A temporary reduction in pain can be valuable. It does not establish that arthritis has healed, and a poor response to cortisone does not, by itself, predict whether PRP or bone marrow will help.

Massage, acupuncture, and care directed at the neck

Some patients also ask about massage, acupuncture, or chiropractic care. These may be discussed for associated muscle discomfort or a separately identified neck problem. They should not be described as methods that realign an arthritic shoulder or restore its lost cartilage. Any hands-on treatment should account for bone health, instability, and the underlying diagnosis.

Hyaluronic acid

Hyaluronic acid is sometimes offered as a joint injection. However, the AAOS’s 2020 shoulder osteoarthritis guideline found no benefit for this treatment based on the evidence it reviewed. It should not be presented as a proven cartilage-restoring option. (3)

Does PRP help shoulder arthritis?

Platelet-rich plasma is made from a patient’s blood. It contains concentrated platelets and their signaling proteins; it does not contain stem cells.

In a double-blind randomized trial of 70 people with glenohumeral osteoarthritis, a single ultrasound-guided leukocyte-poor PRP injection was compared with hyaluronic acid. Pain and function improved in both groups over 12 months, with no significant difference between treatments. (1)

This offers a reason to study PRP further, but it does not prove that PRP works better than placebo, because the trial did not include a placebo group. It also did not establish cartilage restoration, equivalence to replacement, or a predictable result in someone near age 90.

When I discuss an injection, the useful question is whether a possible improvement in pain and daily function would be worthwhile for that person, given the cost, risks, and uncertainty. A promise to eliminate the need for replacement would go beyond these findings.

What about stem cells or bone marrow injections?

In this discussion, bone marrow treatments mean preparations from the patient’s own marrow. Whole bone marrow or nonconcentrated bone marrow aspirate is different from bone marrow aspirate concentrate, often called BMC or BMAC. These contain mixed populations of cells and other components. They are not purified stem-cell products, and one preparation should not be substituted for another when explaining a study.

A pilot randomized shoulder arthritis trial compared nonconcentrated marrow aspirate with cortisone. Twenty-two patients, representing 25 shoulders, completed baseline and 12-month assessments. Some secondary measures favored marrow, but the primary shoulder-specific outcome did not show a statistically significant between-group difference. Enrollment ended early, and the study was too small to settle the question. Participants were eligible only between ages 18 and 75. (2)

Marrow harvesting is an additional procedure, with its own discomfort and risks. More processing or a higher cell count should not be assumed to mean a better result. This study does not demonstrate that BMAC is better than PRP or that marrow injections reliably help an 89-year-old.

What our published shoulder study contributed

Our 2019 study included 47 patients and 50 shoulders with osteoarthritis or rotator cuff tears. Patients reported improvements in pain and daily function after one or two treatments with BMC or whole bone marrow. Mean overall improvement was 42.25% in the one-treatment group and 50.17% in the two-treatment group. (4)

These were observations in treated patients without a randomized control group. They do not establish superiority to surgery, cartilage regrowth, or the likelihood of success in an individual patient. Our detailed shoulder bone marrow research summary explains the measurements and limitations.

How this fits with clinical guidance

The AAOS’s 2020 guideline could not recommend injectable biologics, including PRP and stem-cell preparations, for glenohumeral osteoarthritis because reliable evidence was lacking. That statement predates the 2021 marrow pilot and 2022 PRP trial discussed here. Those later studies contribute information, but they do not establish an injection as an equivalent substitute for replacement. (3) (2) (1)

Where does prolotherapy fit?

Prolotherapy uses an injectant, commonly dextrose, with the intention of stimulating a local tissue response. It is a separate treatment from PRP and marrow injections. A randomized study evaluated dextrose injections for painful rotator cuff tendinopathy. That is a tendon condition, not evidence that prolotherapy restores the cartilage of an end-stage arthritic shoulder. (5)

When instability or a painful tendon is part of the clinical picture, it deserves its own assessment. It should not be used to explain away major joint destruction. For a broader discussion of tendon injuries, see PRP and bone marrow treatments for shoulder pain and rotator cuff tears.

What is the difference between anatomic and reverse replacement?

An anatomic replacement recreates the usual ball-and-socket arrangement and generally relies on a functioning rotator cuff. A reverse replacement changes the arrangement so the deltoid muscle can contribute more effectively to raising the arm.

Reverse replacement may be considered for cuff-tear arthropathy, deficient rotator cuff function, or certain patterns of bone loss and deformity. It is also used in selected patients with primary osteoarthritis. A recommendation for a reverse implant does not automatically mean the cuff is torn; the surgeon should explain why that design fits the particular shoulder. (6)

An injection does not reproduce the mechanical reconstruction of either implant. In a severely deformed joint, improving the biologic environment and replacing the joint are different proposed treatments with different capabilities.

What should patients know about replacement outcomes?

Meaningful improvement is different from complete pain relief

A study of 326 arthroplasty patients estimated that a 1.4-point change on a 0–10 pain scale represented a minimum clinically important difference. This is a research threshold for a noticeable change, not the expected average benefit of replacement and not a universal definition of a successful operation. (7)

For me, the practical question remains: will the improvement let you do the things that matter to you? A score should be interpreted alongside sleep, dressing, independence, and your satisfaction with the result.

Strength, rotation, and return to activity

A comparative study of 41 patients after reverse arthroplasty found improved function, but lower rotational strength and endurance than in the patients’ unaffected shoulders. This helps explain why a shoulder can feel better without functioning like a normal shoulder. (8)

Research on activity after reverse replacement has also emphasized uncertainty about long-term loading and sport recommendations. If your goals include golf, swimming, resistance exercise, or lifting for work, ask about those specific activities rather than assuming that better pain scores mean unrestricted use. (9)

Revision risk depends on the procedure and patient

Older hemiarthroplasty research remains useful when its scope is clear. In a Swedish register study, younger age was associated with greater revision risk; differences between resurfacing and stemmed implants were explained by age after adjustment. This does not mean that every replacement design carries the same risk. (10)

In another long-term study of 44 humeral head replacements, pain relief was maintained overall, but 11 patients had persistent moderate or severe pain, and 10 shoulders underwent revision, usually for arthritis of the socket. These findings describe humeral head replacement, not a universal failure rate for contemporary anatomic or reverse arthroplasty. (11)

Implant wear, loosening, instability, and bone loss can complicate revision. A retrieved-implant study examined fractured glenoid components, while separate revision research addressed difficulties associated with the humeral stem. Those studies help explain specific failure mechanisms; they do not establish that every implant will fail or that injections correct a failed prosthesis. (12) (13)

Overall health and recovery support matter

Research has associated complications after anatomic replacement with overall health and nicotine use. Another study found an association between resilience and patient-reported outcomes. These observations support addressing health, expectations, and support needs; they do not justify blaming a patient or predicting failure from personality. (14) (15)

An 89-year-old with bone-on-bone arthritis: is there a reasonable alternative?

A family asks about an 89-year-old mother with severe right shoulder arthritis who takes tramadol and acetaminophen. Her surgeon has proposed reverse replacement. Could PRP or stem-cell treatment help instead?

I would begin with her goals and the reason for the surgical recommendation. Does she need better sleep and easier dressing? Does she use that arm to push up from a chair or support a walker? Is the joint severely deformed, and how well does the rotator cuff work?

An injection may be worth discussing if a possible reduction in symptoms would meet her goals and she understands that benefit is uncertain. I would not tell this family that an injection can reliably rebuild her joint or that marrow must be better because she is older.

I also would not assume that replacement is unreasonable solely because of age. A small series published in a 2026 journal issue followed 13 patients aged at least 90 after reverse replacement. Movement improved, and all ultimately returned to their previous living arrangements. However, seven initially needed a rehabilitation facility, and the series included both fractures and cuff-tear arthropathy. It was not a trial of 89-year-olds with primary arthritis. (16)

A larger inpatient database study included 555 patients aged 90–99 and found more medical complications and longer hospital stays than in younger patients. Elective cases had fewer complications than nonelective cases. Together, these studies support an individualized discussion rather than a decision based on a birthday alone. (17)

Tramadol and acetaminophen deserve attention

The prescribing clinician should review pain control, drowsiness, constipation, interactions, and fall risk. Check whether any combination medicine already contains acetaminophen. Tramadol should not be abruptly stopped without medical advice. Blood thinners and other medicines also need review before any procedure; patients should not stop them independently. (18)

Are there operations that preserve the joint?

Sometimes a surgeon can address a specific tendon or other structural problem without replacing the joint. Superior capsular reconstruction, for example, has been studied in selected patients with irreparable cuff tears and minimal or no glenohumeral arthritis. It should not be presented as a general solution for bone-on-bone cartilage loss. (19)

When shoulder replacement deserves serious consideration

Persistent disabling pain, major stiffness, loss of independence, and structural damage that matches the symptoms can make replacement a reasonable option. An injection should not become a reason to postpone a useful surgical evaluation indefinitely.

The questions I want answered are straightforward:

• Which structure is causing most of the pain?
• Is the proposed goal symptom relief, improved movement, or mechanical reconstruction?
• What evidence applies to this diagnosis and age group?
• What would count as enough improvement to make the treatment worthwhile?
• What are the next options if it does not help?

My experience with unsuccessful surgery makes me take alternatives seriously. It also makes me take honest expectations seriously. The right recommendation may be nonsurgical care, an injection with clearly explained limitations, or replacement. The patient needs enough information to understand why.

Regenerative medicine therapies have not been FDA-approved to treat shoulder arthritis. This article is educational and does not replace an individualized evaluation. (20)

References