The meniscus, and what it does for cartilage

The menisci are two crescents of fibrocartilage sitting between the femur and the tibia, anchored at both horns and around the rim. Their collagen runs circumferentially, which is the whole trick: an axial load pushing down on a wedge-shaped ring gets converted into tension around the ring rather than pressure straight through it. That is what spreads a load across the cartilage instead of concentrating it. The medial meniscus is tethered at more points than the lateral one, and that is the usual explanation for why it is the one that tears.

What resection costs, measured

The mechanical bill comes due immediately. Take meniscus out and the load it was spreading lands on a smaller patch of cartilage: contact stresses rise about 100% after a medial meniscectomy and 200% to 350% after a lateral one. The relationship between how much tissue is removed and how far stress rises is roughly linear, which is why "partial" is doing real work in the name of the operation.

That linearity has a corollary worth stating plainly. A torn meniscus is still carrying load. A tear is not mechanically equivalent to an absent meniscus, and treating the two as the same thing is the assumption that makes resection look free. It is also why an intact rim matters: the hoop only carries tension if it is continuous and anchored at both ends, so a resection that breaches the rim costs more than its volume suggests.

Which tears can heal

Blood supply decides it, and it recedes with age. The developing meniscus is supplied across its whole width; by adulthood only the peripheral 10% to 30% still has a blood supply, and everything inside that depends on diffusion from joint fluid. That peripheral vascular border — thick, convex, attached to the capsule — is the red zone.

This is the anatomy behind the red-red, red-white and white-white classification a surgeon uses to decide whether a tear is worth stitching. Worth being precise about what that anatomy is: it maps where blood arrives. It is not a healing rate, and no part of the zone system was derived from measuring how often tears in each zone actually heal.

What a tear on a scan means

Nine hundred and ninety-one people in Framingham, aged 50 to 90, were scanned without being selected for knee trouble. A meniscal tear or meniscal destruction turned up in 19% of women aged 50 to 59 and 56% of men aged 70 to 90. Among people with radiographic osteoarthritis, 63% of those with knee pain on most days had a tear — and so did 60% of those without. Without radiographic osteoarthritis, the figures were 32% and 23%.

Sixty-one percent of everyone found to have a tear had reported no pain, aching or stiffness in the previous month.

That is the base rate any MRI report has to be read against. In a knee past fifty, finding a meniscal tear is close to finding a grey hair: real, visible, and not by itself the reason anything hurts.

What resecting it does over years

The European consensus on traumatic tears states the direction plainly: preservation should be the first line of treatment wherever it is possible, because the clinical and radiological long-term outcomes are worse after partial meniscectomy than after preservation. That is a consensus statement rather than a trial result, and it is graded as such by the group that wrote it.

Where the comparison has been pooled, it points the same way. Across six studies at a mean of four years, advanced knee osteoarthritis was about half as likely after repair as after resection of an acute tear (OR 0.51), with progression to knee replacement at the same estimate. In the same analysis mean joint space width did not separate, and the resected knees actually scored four points higher on Lysholm — below the threshold a patient would notice, and pointing the other way. Those results disagree with each other and the authors report all of them.

None of that is randomised, and the knees that get repaired are not the knees that get resected.

Where the imaging clock runs fastest

In the Osteoarthritis Initiative, every one of the 31 knees that had a partial meniscectomy in the year before the case-defining visit went on to radiographic osteoarthritis, and no control knee had had one. A prevalent tear or maceration predicted it too, though less absolutely: 165 of 280 such knees (58.9%) developed radiographic osteoarthritis, crude OR 2.51.

Then the two come apart, and this is the part worth carrying. Among the knees that developed radiographic osteoarthritis, having a torn meniscus did not predict losing cartilage over the following year (OR 0.88, 95% CI 0.51 to 1.51). Having had the tear operated on did (OR 4.51, 95% CI 1.53 to 13.33). Cartilage loss showed up in 80.8% of the operated knees and 37.4% of the knees with meniscal damage left alone.

The obvious objection is that the operated knees were worse to begin with, and it is a real one — this is a cohort, and the operation was chosen. What makes the result hard to dismiss entirely is the direction of the split: the tear itself stopped predicting cartilage loss at the point where the surgery started predicting it.

The one randomised structural result

Everything above is observational, and observational evidence here has a specific weakness: the knee that gets operated on is the knee that hurt more, and a knee that hurt more may have been heading somewhere worse anyway.

One trial escapes that, because it randomised against a sham. In FIDELITY at five years, 48 of 67 knees (72%) after arthroscopic partial meniscectomy and 44 of 74 (60%) after placebo surgery had progressed at least one Kellgren-Lawrence grade — an adjusted absolute risk difference of 13%, with an interval running from -2% to 28%. The OARSI sum score moved 0.7 points in the same direction, and that interval excluded zero.

Read what those instruments measure, because the trialists flag it themselves: both Kellgren-Lawrence and the OARSI atlas are osteophyte-driven grading systems, chosen partly for their resilience to the noise in measuring joint space width. So the randomised structural signal is real, it is in the expected direction, it is imprecise, and it is more about bone spurs than about measured cartilage.

What would sharpen this

The surgeon's actual decision is how much meniscus to take, and nobody has randomised it. The cadaver work brackets the answer at nothing and everything and reports the relationship between the two as roughly linear; the cohorts consistently rank resection volume among the strongest predictors they have. That is a lot of converging evidence about a variable no trial has ever allocated, and a dose-response study of resection volume would give a reader facing this operation the one number nobody can currently give them.

The second gap is cartilage itself. The randomised structural evidence is measured with instruments that count osteophytes well and cartilage poorly, and the cohort that does measure cartilage cannot randomise. A trial with a cartilage-specific imaging endpoint would resolve a question this whole shelf currently answers by triangulation.