It depends which part is damaged. Reviews describe losing aggrecan, the water-holding molecule in cartilage, as an early and reversible step, and losing type II collagen as irreversible, because the collagen network renews so slowly in adults. Some researchers argue cartilage can repair more than that, pointing to faster turnover in the ankle and to repair after knee joint distraction.
Cartilage is built from a fibre network of type II collagen holding a dense gel of aggrecan. Collagen is two-thirds of the dry mass (Eyre 2002). Aggrecan is the molecule that lets cartilage resist being compressed (Glasson 2005).
A 2011 review set out the difference plainly. The loss of aggrecan is thought to be an early and reversible process, while breakdown of type II collagen is considered irreversible and a key step in cartilage losing its structural and functional integrity (Li 2011). The same distinction runs through the earlier research. In 1996, Mitchell and colleagues called the breakdown of triple-helical collagen an important step towards irreversible tissue damage (Mitchell 1996).
Cartilage collagen is barely replaced in adult life. Radiocarbon dating of knee cartilage from 23 people found virtually no collagen replacement after skeletal maturity, and joint damage didn't speed it up (Heinemeier 2016). Its half-life has been estimated at 117 years (Verzijl 2000). If the network is cut faster than the cells can mend it, and they mend it very slowly, the loss builds up.
Studies that measure collagen fragments in blood and urine have added the repair side to that picture.
What's been a more recent and exciting development is that we've learned through studies of collagen and collagen biomarkers that not only is the disease a problem with breakdown, but it's a problem with inadequate repair.
Mouse experiments let researchers switch single enzymes off and see which damage follows.
The MMP13 result is the telling one. The authors wrote that MMP13 deficiency held back cartilage erosion even when aggrecan was depleted (Little 2009). In that model, losing the gel didn't decide the outcome. Cutting the collagen did.
The irreversible label rests on how slowly the collagen network turns over, and some researchers think the case has been overstated. Virginia Kraus's group at Duke found that protein turnover in human cartilage ran on a gradient, fastest in the ankle and slowest in the hip, which they read as a limited capacity for repair (Hsueh 2019). In a small Utrecht study, knees held slightly apart in a frame showed what the authors called an initial boost of cartilage-like tissue repair, and five years later had lost less cartilage than comparison data on untreated osteoarthritis predicted (van der Woude 2017). That study had no control group of its own, and its gain in mean cartilage thickness had faded by five years.
Tonia Vincent thinks the load on the joint decides whether that repair can happen.
The skin repairs extremely well, but if you've got a blister on your foot because you're wearing a shoe that is new and doesn't fit very well, you know perfectly well that skin will not heal if you carry on wearing the shoe … I think exactly the same analogy applies for the articular cartilage.
None of this has a neat human equivalent yet. A 2021 review said that, so far, no interventions were available to restore degraded cartilage (Hu 2021). The drug research that followed has focused on slowing breakdown, especially by blocking MMP13 selectively (Li 2011), rather than on regrowing collagen. Broad MMP blockers have done badly so far: one tested for 12 months in 401 people showed no benefit over placebo and caused musculoskeletal side effects (Krzeski 2007). No drug that blocks MMP13 has reached patients either.
Some trials of boswellia extracts in people with knee osteoarthritis have reported lower markers of type II collagen breakdown, and one reported MRI increases in cartilage volume and thickness against placebo (Kumar 2025; Vaidya 2025). Each tested one product at one dose, over 90 to 180 days, in fewer than 100 people.
Not according to the one study here that looked at healthy ageing cartilage. In ankle cartilage from 30 donors aged 16 to 75, the type II collagen and proteoglycan content didn't change after age 30, and the authors found no molecular signs of the degeneration seen in diseased joints (Aurich 2002). Age changes how cartilage cells respond to provocation (Forsyth 2005), which is covered in the question on whether MMP13 increases with age.
Aggrecan, the water-holding molecule, on the standard account. Its loss is described as early and reversible, while type II collagen loss is considered irreversible. Some researchers argue cartilage collagen has more capacity to repair than that.
Not so far for broad MMP blockers. One 12-month trial of a broad MMP-blocking drug in 401 people showed no benefit over placebo and caused joint and muscle side effects.
In healthy ankle cartilage from donors aged 16 to 75, collagen and proteoglycan content held steady after age 30, with no molecular signs of degeneration.
Its collagen barely does, on most measurements. Radiocarbon dating of knee cartilage from 23 people found virtually no replacement of the collagen network after the...
They attack different parts of cartilage. MMP13 is the main enzyme that cuts type II collagen, the fibre network. ADAMTS5 is the main aggrecanase, cutting aggrecan, the...
MMP13 cuts type II collagen, the rope-like fibre that gives cartilage its strength. It cleaves each three-stranded collagen molecule at one spot, leaving a three-quarter...
Not on its own. In healthy ankle cartilage from 30 donors aged 16 to 75, all four collagen-cutting enzymes were present but did not change with age. What age changed was...