What the 10-year data actually show

For young, active patients weighing these two procedures, the most direct answer the evidence can offer is this: at ten years, OATS mosaicplasty shows a meaningfully lower failure rate than microfracture. That is not a unanimous verdict, but it is the direction the better-powered randomised trials point.

The firmest data come from Gudas et al. (Am J Sports Med, 2012), the only prospective randomised trial comparing the two techniques in young athletes to reach a full decade of follow-up. At ten years, 14% of OATS patients had experienced treatment failure versus 38% of microfracture patients — a statistically significant difference (P<0.05). Muthu et al.'s 2024 network meta-analysis, drawing on a broader pool of studies, corroborates that finding: mosaicplasty demonstrated significantly better long-term functional outcomes at ten years compared with microfracture.

The picture is not without contradiction. Ulstein et al. (KSSTA, 2014), a Level II randomised trial with a median 9.8-year follow-up in 25 patients, found no statistically significant difference in Lysholm score, KOOS, isokinetic muscle strength, or radiographic osteoarthritis between the two techniques. Mean Lysholm scores sat at 69.7 for the microfracture group and 62.6 for the OAT group — a gap that failed to reach significance, and an absolute level that suggests neither procedure reliably restores patients to full healthy function.

Small trial populations and varying patient selection account for much of that divergence, and the evidence is better understood as directional rather than definitive.

Why short-term follow-up studies can mislead

The Gudas trial data reveal something that raw failure-rate statistics alone do not communicate: the two procedures looked broadly equivalent at short-term assessment. Divergence only became apparent between years five and ten, as microfracture outcomes declined while mosaicplasty results held stable.

That timing matters enormously when evaluating the wider literature. A trial reporting outcomes at two years — or even five — may capture the period when the two techniques are most similar, not most distinct. Any equivalence found at that stage cannot safely be extrapolated to the decade mark. Studies that stop at 24 months are not simply a shorter version of the same story; they are measuring a different phase of the biological timeline, and conclusions drawn from them will systematically underestimate the advantage that durable cartilage tissue may confer over time.

For active patients whose aim is to remain in sport or sustained weight-bearing activity into their late thirties and forties, the gradient of change between years five and ten is what actually informs prognosis. A repair that holds steady at year eight carries different clinical weight from one that peaks at year three and then declines — even when both looked identical at the first post-operative review.

Why the two techniques age differently

The divergence described in the preceding sections has a straightforward biological explanation — one that makes the late deterioration of microfracture repairs feel mechanically inevitable rather than statistically surprising.

OATS places a cylindrical plug of the patient's own mature cartilage directly into the defect. That tissue is genuine hyaline cartilage, built around type II collagen — the same structural protein as the surrounding joint surface. It is load-tolerant by design and does not need to adapt to athletic demand; it already is the tissue required for it.

Microfracture works differently. Puncturing the subchondral bone releases marrow cells that clot within the defect and gradually differentiate into fibrocartilage — a repair tissue organised around type I collagen. Fibrocartilage is softer and less mechanically resilient than hyaline cartilage; under the repeated high-impact loading of weight-bearing sport, it tends to thin and fragment over time. There is a further concern: perforating the subchondral bone plate during microfracture can disrupt the underlying bone architecture, and some evidence suggests this may narrow the options available should a second procedure ever be needed.

OATS carries its own trade-off. Harvesting the graft plugs from a lower-load region of the same knee introduces donor-site morbidity — a source of pain that can persist alongside, or independently of, the repaired area. In practice, this can equalise patient-reported pain scores even when the structural outcome at the repair site favours OATS.

Why the trials disagree — and what both tell us

Two null-finding trials deserve honest attention rather than footnote status.

Ulstein and colleagues' RCT, with a median follow-up of 9.8 years, found no statistically significant advantage for either technique across functional scores, muscle strength, or radiographic osteoarthritis. What the absolute scores reveal — already noted earlier — is arguably the more important point: neither procedure reliably restored patients to the functional level of a healthy reference population. Lim and colleagues' level-2 comparison reached a similar conclusion, finding no significant difference in Lysholm or Tegner scores across microfracture, OAT, and ACI.

The divergence across trials reflects genuine variation in patient populations rather than methodological inconsistency. Patient age, body mass index, defect size and location, concomitant pathology such as ligament instability or malalignment, and surgical execution all modify outcomes in ways that RCT stratifications have not yet systematically captured. A study enrolling older, lower-demand patients with larger or less well-contained defects is asking a materially different clinical question from one recruiting young athletes with discrete femoral condyle lesions.

Gudas enrolled precisely the latter group — which is why the biomechanical advantage of transplanted hyaline cartilage had both the time and the loading conditions to become visible across a decade. The null-finding trials reflect the messier reality of broader patient populations, where multiple competing variables can dilute a technique-specific signal.

For younger, active patients with appropriately sized focal lesions, the weight of evidence tilts toward OATS — but the mediocre absolute scores seen even in studies that favour neither technique serve as a calibrating benchmark: cartilage repair is joint preservation, and patient expectations should be set accordingly from the outset.

Which patients suit which technique

Translating trial findings into individual decisions means matching several clinical variables — none of which operates in isolation.

Lesion size is the first filter. OATS is generally favoured when a focal defect exceeds 2 cm²; mosaicplasty, which uses multiple smaller plugs arranged across the defect footprint, can extend this approach to lesions of approximately 4 cm². For defects under 2 cm², microfracture has historically been considered sufficient — though the durability concerns described in earlier sections apply regardless of size. It is worth being explicit: these thresholds are drawn from subgroup observations rather than pre-specified trial arms, which makes them useful guides rather than hard cutoffs.

Age and activity demand sit alongside size. Patients under 40 with a clear goal of returning to weight-bearing or competitive sport represent the group in whom OATS durability benefits are most evident and most clinically relevant. For older or lower-demand patients — where sustaining a decade of structural integrity is less pressing — the risk-benefit balance shifts, and microfracture's lower procedural burden may reasonably outweigh its structural limitations.

Surgical complexity deserves honest acknowledgement. OATS is technically demanding and requires approximately six months of rehabilitation — longer and more intensive than microfracture. That is not a reason to default to the simpler option when the evidence favours the more demanding one, but it is a variable patients need to factor into their decision.

Concomitant pathology also matters. Malalignment, ligament instability, or meniscal deficiency left unaddressed will undermine any cartilage repair procedure. A specialist assessment covering joint mechanics, imaging, and functional goals is the necessary starting point before a pathway is agreed.

What 'improvement' means — and what it does not

Pareek and colleagues' 2016 systematic review offers a distinction that is easy to miss when headline failure rates dominate the discussion. At ten-year follow-up, OATS patients showed statistically significant improvements from baseline in both IKDC and Lysholm scores — objective markers of knee function and symptom burden. Tegner activity-level scores, which measure whether a patient has returned to their specific pre-injury sport and loading demands, did not improve significantly.

The gap between those two findings matters in practice. A knee that performs measurably better on a clinical assessment than it did before surgery is a real and meaningful outcome. It is not, however, the same as returning to competitive sport at the pre-injury level, or resuming the skiing, running, or contact sport that the original damage interrupted. Both improvements may occur for a given patient; neither can be promised.

Two further caveats sit alongside this. RCT evidence beyond ten years is essentially absent — whether the OATS advantage over microfracture holds at twelve or fifteen years has not been established at trial level, and extrapolating from the available data carries genuine uncertainty. Separately, OATS involves approximately six months of structured rehabilitation, a timeline that is part of the procedural commitment rather than a minor footnote.

For patients working through this decision, the most productive question to put to a specialist is not only which technique performs better but what does 'better' mean for my specific activity goals, my timeline, and my tolerance of surgical risk — and, specifically, whether functional score improvement or return to a defined sport level is the more realistic and relevant benchmark for them.