Regenerative medicine and physical therapy: what the injection starts, and what loading finishes

What's actually in the syringe

Platelet-rich plasma starts as your own blood. It gets spun in a centrifuge to concentrate the platelets, and that concentrate gets injected into the injured tissue. Bone marrow aspirate concentrate (BMAC) works on a similar idea with marrow instead of blood. Prolotherapy uses a dextrose solution to provoke a healing response rather than deliver one.

The reason platelets are the target is that they are not just clotting cells. They carry storage compartments called alpha-granules, and those granules are packed with signaling proteins — PDGF, TGF-β, IGF-1, VEGF, bFGF, EGF. When platelets activate at an injury site, they release that payload into the tissue.

Platelets don't build tissue. They recruit the cells that do.

This is the part worth understanding, because it's where most explanations skip a step. Platelets do not become connective tissue. They send for the cells that make it.

PDGF and IGF-1 call fibroblasts to the area and tell them to divide — more fibroblasts, arriving where the damage is. TGF-β and IGF-1 then push those fibroblasts to lay down collagen and other matrix proteins. bFGF has its own job later on. It's a relay: platelet → growth factor → fibroblast → collagen. The injection is the first link in that chain, not the last.

One honest caveat, because it matters for how you should think about all of this: more is not automatically better. High concentrations of TGF-β1 in particular are associated with scarring and tendon adhesions, not cleaner healing. In laboratory work, raising the platelet concentration has actually reduced collagen production by tendon cells. "More growth factor" and "better tissue" are not the same claim.

New collagen shows up disorganized

Connective tissue heals in overlapping phases. The first week is inflammatory — a clot forms, immune cells clear debris, and fibroblasts begin depositing type III collagen, which is quick to produce but thin and mechanically weak. Over roughly the next two to six weeks, the proliferative phase piles up a lot of that matrix. There is plenty of new material, and it is laid down in no particular direction.

Remodeling begins around week four to six and is where the tissue either becomes useful or doesn't. Type III collagen is gradually replaced by type I — thicker, stronger, the collagen healthy tendon is mostly made of. Cells elongate, fibers align, and cross-links form. Collagen concentration approaches normal around three months, but full matrix maturation can take up to a year.

Even then, repaired tendon and ligament are not identical to the original. The scientific literature is blunt about this: healed tendon is closer to organized scar than to regenerated tendon, with measurably inferior mechanical properties. Healing is repair, not replacement. Which is exactly why the quality of that repair is worth working for.

Load is what tells collagen which way to point

Tendon and ligament are strong in one direction — along the line they're pulled. A pile of randomly oriented collagen has volume without much of that strength. Something has to give the new tissue its direction, and that something is mechanical load.

The mechanism is called mechanotransduction, and it's the reason exercise is a tissue-level intervention and not just conditioning. When you load a tendon, the deformation is sensed by the cells inside it through structures like integrins and mechanosensitive ion channels. Those cells convert a physical signal into a biochemical one, and respond by building and organizing matrix. Physical therapists have a name for using this deliberately: mechanotherapy.

In humans we can measure the front half of this directly. Loading a tendon raises collagen synthesis, with the response peaking about a day after exercise and staying elevated for roughly three days. Sustained loading increases tendon cross-sectional area and stiffness. The specific claim that fibers rearrange along lines of stress comes mostly from animal and engineered-tissue studies rather than human imaging — I'd rather tell you where the evidence gets thinner than pretend it doesn't.

What happens when nothing loads it

The unloading data is some of the most striking in this whole area, because it moves fast. In healthy adults placed in limb suspension, tendon stiffness dropped about 10% in fourteen days and roughly 29% by day twenty-three. Longer bed-rest studies show far larger losses. This is not deconditioning over months. It's measurable in two weeks.

So a tendon that gets a growth-factor signal and then sits still is being asked to build without a blueprint. You can get matrix without getting mechanically useful tissue.

One important nuance, because I don't want to oversimplify in the other direction: "more load is always better" is also wrong. After a surgical rotator cuff repair, for example, early aggressive motion can produce worse tissue in animal models, while protected healing does better. The dose and the timing are site-specific and injury-specific. That judgment is the job.

Now the uncomfortable part: what the trials show

Everything above is real biology. But biology being real does not automatically make an injection work in patients, and this site doesn't get to publish an honesty page about our own measurements and then go soft here.

The best-designed trials of PRP have largely failed to beat placebo:

• Knee osteoarthritis (RESTORE, JAMA 2021). 288 patients, PRP versus saline, twelve months. Pain improved 2.1 points with PRP and 1.8 with saline — an adjusted difference of 0.4, well below what a patient can feel. Cartilage volume on MRI: no difference.
• Achilles tendinopathy (ATM, JAMA 2021). 240 patients, PRP versus sham. Final scores 54.4 versus 53.4. No benefit.
• Achilles tendinopathy (JAMA 2010). Everyone did eccentric loading; half also got PRP, half got saline. Both groups improved about 21 points. The loading produced the improvement. The injection added nothing detectable.
• Tennis elbow (Cochrane, 2021). 32 trials, 2,337 people: 0.16 points better than placebo on a 10-point pain scale.
• Acute hamstring injury (NEJM, 2014). PRP did not shorten return to play versus saline.

There are positive findings too, and they deserve fair treatment. PRP alongside arthroscopic rotator cuff repair does appear to reduce re-tear rates. Some plantar fasciitis and tennis elbow studies favor PRP over corticosteroid at six to twelve months — though that partly reflects steroid's known tendency to relapse rather than PRP's strength. And PRP is not one product: leukocyte-rich versus leukocyte-poor, platelet concentration, spin protocol and injection schedule all vary enormously between studies, which is a legitimate reason pooled results are messy.

I'm also not the one holding the needle, and I'm not in a position to overrule the physician who is. Orthobiologics are an active research area, they're generally safe, and there are patients and presentations where a skilled physician has good reason to try one. What I can tell you is what the controlled trials currently support — and how to make sure the part that is well supported actually happens.

Where physical therapy fits

Here's the honest version of the relationship, and it's not the version usually sold: the injection supplies signal; loading supplies direction — and loading is the part with the stronger evidence behind it.

Progressive tendon loading has decades of trial support. Alfredson's heel-drop protocol, heavy slow resistance training, staged progressive loading programs — these consistently improve pain and function in tendinopathy, and current clinical practice guidelines put tendon loading first-line. Notably, head-to-head comparisons suggest the specific flavor matters less than people argue about: heavy slow resistance matched eccentric training at one year, with better adherence, because it asks for three sessions a week instead of fourteen.

I'll be straight about the limit of my own argument, too. No adequately powered trial has isolated "injection plus structured rehab" against "injection alone." The reason to load after an injection rests on the independent evidence for loading and on expert consensus — not on a trial that proved the combination. Practice is ahead of data here. In fact, the one high-quality three-arm trial in knee osteoarthritis that compared exercise, PRP, and both found no meaningful difference between them, and its authors concluded exercise alone was the recommended treatment.

If you're getting an injection, ask these

This is general education, not medical advice, and your physician's instructions for your case come first. But these are reasonable questions to raise:

• When do I start loading it? An international expert panel recommends rehabilitation begin within five to ten days after injection for chronic tendinopathy — while noting their own guidance rests mainly on expert opinion, not trials.
• How do I know how hard to push? The usual guide is pain-limited: work at a level that stays tolerable during the session and settles within 24 hours, then progress from there.
• What about anti-inflammatories? Most protocols ask you to avoid NSAIDs for two to three weeks post-injection, on the logic that blunting platelet activation blunts the payload. Worth knowing: that reasoning is mechanistic, and the one trial registered to test it clinically was withdrawn without enrolling anyone. Ask your physician — don't stop a prescribed medication on your own.
• What's the plan if this doesn't work? A good answer exists. Ask for it before you pay, especially since PRP is usually not covered by insurance.

The bottom line

Your body's repair machinery is genuinely impressive: platelets summon fibroblasts, fibroblasts build collagen, and collagen becomes strong connective tissue. Orthobiologics are an attempt to amplify the first step of that chain. Whether they succeed is, at the moment, genuinely uncertain — and the largest, best-blinded trials mostly say no.

The last step of the chain is not uncertain. Tissue organizes along the loads you give it, it degrades measurably without them, and progressively loading it is the intervention with the deepest evidence base in this entire conversation. That step is the one I'm responsible for — and it's the one that doesn't require a syringe.