The knee ligament that heals itself, and the 26 papers that cannot agree how to help it

There is a submission in jiu jitsu called the inside heel hook, and it works by taking hold of the foot and rotating it. The foot is only the handle. The force arrives at the knee.

Which is how Francis kept turning up with the medial side of his knee complaining. Not once. Repeatedly, over the years, the way anyone who trains in a room full of people learning leg entanglements eventually does.

Those sessions had a familiar shape by then. Treatment on one side, argument on the other. He would be face down having something distinctly unpleasant done to the inside of his knee while explaining what he thought the underlying mechanism was, and she would be disagreeing with him while doing it. The debates were not gentle and neither was the therapy. It was, in fairness, a productive arrangement.

So this article is about the ligament that argument was usually about, and about the strange discovery that when researchers finally went looking for the evidence behind everything physiotherapists do for it, there was much less there than anyone expected.

First, the ligament nobody thinks about

The medial collateral ligament runs down the inside of the knee, 8 to 10 centimetres long. It starts centrally on the medial femoral epicondyle, the bony bump on the inside of the thigh bone, and attaches below the joint line on the tibia, roughly 5 to 7 centimetres down, alongside the pes anserinus.

Its deep layer is essentially a thickening of the joint capsule itself, with parts attaching directly to the medial meniscus. It is not a rope stretched across the outside of the knee. It is partly woven into the joint.

And it is the most commonly injured ligament in the knee. The review puts a number on that which should end any argument about which injury deserves more attention: an elite male football team can expect roughly two MCL injuries in a single season, compared with about one ACL injury every two seasons.

Four times as common as the injury everyone worries about, and the researchers still call it the forgotten side of the knee.

The classic mechanism is valgus stress with the knee at about 20 degrees of flexion, the knee forced inward, often with the tibia rotating outward at the same time. A blow from the outside of the leg does it. So does a rotational force applied through the foot by somebody who knows what they are doing.

Why this one repairs itself

Here is the question worth asking, and almost nobody does. The ACL, a few centimetres away in the same joint, famously does not heal. The MCL usually does. Why?

The most interesting answer on the table involves a structure only named in 1990, called the epiligament. It is a layer of connective tissue wrapped around the outside of a ligament, distinct from the ligament proper, and for decades it was treated as packaging rather than anatomy.

Work in rat models, later supported in human studies, suggests it is nothing of the sort. The epiligament appears to actively supply the repair, acting as a source of both cells and new blood vessels for the healing ligament.

In plain terms, one of these ligaments comes wrapped in its own repair crew and the other does not. The MCL has the cells, the blood supply and the signalling to rebuild itself. The ACL has a thinner, quieter version of the same layer and cannot manage the job.

Two things stop this from being a fairy tale. The first is that the authors are honest that the MCL heals spontaneously but is not fully restored. Healed is not the same word as new. The second is that this is a commentary rather than a trial, and its own conclusion says the practical application of the theory remains uncertain and that whether it gains wide acceptance is still undetermined.

Still, it is the best available explanation for the single most useful fact about this injury: left alone sensibly, the ligament is on your side.

Now the uncomfortable part

Given that grade one and grade two sprains are treated without surgery almost everywhere in the world, you would assume the rehabilitation is well established. A Swedish and British group tested that assumption in 2024 and published the result in BMJ Open Sport & Exercise Medicine.

They searched Embase, MEDLINE and PEDro, registered the protocol in advance, followed PRISMA, assessed risk of bias with a formal tool and graded the certainty of the evidence using GRADE. This was done properly.

They found 26 reports covering 1,912 patients. And then the detail that stops you in your tracks:

Eighteen of those 26 were published before the year 2000. Only eight came after.

Two thirds of the evidence base for the most common ligament injury in the knee is older than the century.

What the 26 actually showed

• No reported difference in treatment between grade one and grade two. Despite being graded separately and discussed separately, they were managed the same way in the literature.
• Immediate weight bearing and walking were tolerated. Not delayed, not protected for weeks. When patients could take load, taking load was fine.
• Rehabilitation meant strengthening exercises, described by the reviewers as having "poorly reported details." Which is a polite way of saying nobody wrote down what they actually did.
• Bracing was all over the place. Some studies immobilised in a brace, others used no equipment at all, and both whether a brace was used and for how long were inconsistently reported.

The reviewers' own summary of the certainty of the evidence is not a paraphrase: very low quality.

How to hold that honestly

It is worth being clear about who this article is describing. Isolated grade one and grade two injuries, the ones that heal without surgery. A knee that gives way, locks, swells dramatically, or was injured with a pop and immediate instability is a different conversation and needs assessing properly rather than reading about.

What we take from it in practice

Respect the biology, and stop apologising to the knee. This ligament is built to repair, with a wrapper that supplies the cells and the blood vessels to do it. The evidence supports weight bearing and walking as tolerated rather than protective avoidance. The instinct to guard a sore knee for weeks is understandable and generally unhelpful.

Take the grading less literally than the label suggests. Across 26 studies and nearly 2,000 patients, grade one and grade two were not actually treated differently. The grade describes the injury. It does not, on this evidence, dictate a separate rehabilitation.

The brace question has no evidence-based answer. Anyone telling you confidently that a brace is essential, or that it is pointless, is going beyond what has been shown. The literature genuinely does not know, and it is more honest to say so.

And expect the injury to come back if the exposure does not change. A ligament that heals but is not fully restored, in a knee that will be put in the same position again next training session, is a conversation about load and about which positions are worth accepting. That conversation is the part of rehabilitation nobody publishes, and probably the part that matters most.

Which is more or less what those arguments on the treatment table were about, at considerable volume, for several years. He is still training. So is she.