International Cartilage Repair Society grading of chondral lesions and their MRI correlates — from intact hyaline cartilage to exposed subchondral bone
The International Cartilage Repair Society (ICRS) grading system, introduced in 2000 as a standardized arthroscopic complement to the older Outerbridge classification, scores chondral lesions on a 0–4 scale of increasing depth and severity. Grade 0 and Grade 1 sit at the reassuring end of that scale — cartilage that is either completely normal or shows only superficial change — but they are simultaneously the grades that MRI struggles hardest to characterize with confidence.
Before ICRS, most surgeons used the Outerbridge classification (1961), originally devised for patellar chondromalacia and later applied loosely across the whole knee. ICRS refined this into a reproducible arthroscopic standard:
• Grade 0 — Normal: smooth, intact, glistening white cartilage with no surface irregularity, indentation, or discoloration.
• Grade 1 — Nearly normal: superficial lesions only. – 1A: soft indentation and/or superficial softening, without a visible surface break. – 1B: superficial fissures and cracks are present, but they remain confined to the superficial (tangential) zone and do not extend into the transitional or deep layers.
Both grades preserve the fundamental biomechanical function of cartilage: the superficial zone's tightly packed, collagen-dense lamina still bears shear load and retains proteoglycan-driven compressive resilience beneath it. Grade 1 lesions are common incidental findings — many are simply early wear-and-tear changes that never progress to a clinically significant defect.
On MRI, Grade 0 cartilage shows uniform intermediate-to-high signal on fluid-sensitive sequences with a smooth, continuous articular surface and sharply defined margins against joint fluid. Grade 1 change, when visible at all, appears as:
• A focal area of mildly increased intrasubstance signal on T2-weighted or proton-density fat-saturated sequences, without a discrete surface defect (correlating with 1A softening).
• A thin, superficial line of signal reaching only partway into the cartilage thickness, corresponding to a 1B fissure — often only a few tenths of a millimeter deep, near or below the in-plane resolution of routine clinical MRI.
At 1.5T with conventional fast spin-echo protocols, in-plane resolution is typically 0.3–0.5 mm and cartilage itself may only be 2–4 mm thick at the femoral condyle — meaning a superficial fissure can occupy a single pixel row or less. This is why even high-quality MRI is frequently read as "normal" in the face of arthroscopically confirmed Grade 1 change, and why radiologists are taught to hedge with descriptors like "chondral irregularity, cannot exclude low-grade fissuring" rather than commit to a grade.
In the classic MRI–arthroscopy correlation study by Bredella and colleagues, sensitivity for detecting Grade 1 lesions on conventional MRI was only around 25–30% — meaning roughly two-thirds of true Grade 1 lesions were missed or read as normal. Sensitivity climbed steeply with lesion depth at every subsequent grade, which is the central theme this simulator is built to illustrate.
The older Outerbridge classification (originally four grades, based on open surgical inspection of the patella) is still referenced informally in many reports, but it was never designed for arthroscopic use across the whole knee and its grade boundaries do not map cleanly onto ICRS. ICRS Grade 0–1 was deliberately defined to correspond to what Outerbridge would call normal-to-Grade-I change, preserving continuity with decades of older literature while tightening the definition with explicit subtypes (1A softening vs. 1B fissuring).
Clinically, an isolated Grade 0–1 finding — whether picked up incidentally on MRI obtained for another reason, or noted in passing during arthroscopy performed for a meniscal or ligament procedure — is essentially never, on its own, an indication for cartilage-directed surgery. It is recorded, followed if the patient is symptomatic, and used as a baseline: many Grade 1 lesions remain radiographically and arthroscopically stable for years, while a minority progress. This is precisely why grading systems exist — not to trigger intervention at every abnormal finding, but to give surgeons and radiologists a shared, reproducible vocabulary for tracking a lesion's natural history over time.
Grade 2 lesions extend through less than half of cartilage thickness but clearly breach the superficial zone into the transitional layer. Architecturally the deep zone and the calcified cartilage tidemark remain untouched, so the lesion is still, strictly speaking, a partial-thickness defect — but it is also the grade with the least consistent inter-reader agreement of the entire ICRS scale, both arthroscopically and on MRI.
Arthroscopically, a Grade 2 lesion is diagnosed when a probe can be seated into a defect that clearly extends past the superficial zone into the transitional (middle) zone, but the surgeon can still visually or tactilely confirm that healthy, structurally intact cartilage remains beneath it — the deep zone and the calcified cartilage layer are not involved. The lesion crater typically has soft, slightly frayed edges rather than the sharp, vertical walls seen in deeper grades.
Biomechanically this matters: the transitional zone's more randomly oriented collagen fibrils and higher proteoglycan content still provide meaningful compressive load distribution, so a Grade 2 lesion — while abnormal — has not yet lost the layer primarily responsible for shock absorption.
A Grade 2 lesion on fluid-sensitive MRI (fat-saturated proton density or T2-weighted) appears as a focal area of increased signal intensity extending into, but not through, the cartilage thickness — typically reaching roughly a third to just under half of the way to the subchondral bone plate. Two features distinguish it from Grade 1 and Grade 3 on imaging:
• Versus Grade 1: the signal abnormality is deeper and usually wider, often with a discrete surface irregularity or thinning visible on the articular contour, not just intrasubstance signal change.
• Versus Grade 3: a rim of preserved, normal-signal cartilage is still visible between the base of the defect and the low-signal calcified cartilage/subchondral bone plate — this residual "deep cartilage stripe" is the single most useful discriminator on MRI.
In practice this stripe can be only one or two voxels wide, which is exactly why Grade 2 versus Grade 3 distinction is one of the most common sources of disagreement between MRI readers and between MRI and arthroscopy alike.
Grade 2 sits at a genuine measurement boundary problem, not just a training issue. Several factors compound:
• Partial-volume averaging: standard clinical slice thickness (2–4 mm) means a lesion that tapers in three dimensions gets averaged with adjacent normal cartilage in the imaging plane, blurring the apparent depth.
• Motion and pulse-sequence variability: cartilage-specific protocols (3D fat-saturated spoiled gradient echo, isotropic 3D fast spin echo) resolve the deep cartilage stripe far better than routine 2D sequences, but not every scan uses them.
• Reader experience: musculoskeletal-fellowship-trained radiologists show substantially higher agreement with arthroscopic grade than general radiologists, particularly at Grade 2–3.
The practical takeaway echoed throughout the cartilage-imaging literature is that MRI-reported grade should be treated as an estimate with a built-in margin of error at this level of severity, and that a hedge of "Grade 2–3" is often more honest than a falsely precise single number.
This grade illustrates a general principle of MRI–arthroscopy correlation: agreement is worst precisely where clinical decisions are most sensitive to the exact number — a Grade 2 lesion is typically managed non-operatively or with lesion-preserving techniques, while a Grade 3–4 lesion may prompt surgical planning.
Grade 3 marks the point at which more than half of cartilage thickness is lost, with the defect approaching or reaching the calcified cartilage layer immediately above the tidemark — yet the subchondral bone plate itself remains covered. The ICRS subgrades (3A–3D) exist precisely because "deep" is not a single state: a lesion can be deep-but-not-quite-there, deep-and-there, or deep-with-blistering, and each has a distinct MRI signature.
ICRS Grade 3 is subdivided to capture how close the lesion base sits to the bone, and whether the surface has remained closed:
• 3A — Lesion extends past 50% of cartilage depth, but not down to the calcified layer.
• 3B — Lesion extends down to the calcified layer.
• 3C — Lesion extends down to, but not through, the subchondral bone plate (the calcified layer is breached, bone plate is still intact and covered).
• 3D — Blistering: the deep layers have delaminated from the subchondral bone, creating a "blister" of cartilage that may look nearly intact on the joint surface but is structurally detached underneath. This subtype is arthroscopically deceptive — probing reveals a soft, ballotable surface over a hidden cavity rather than an obvious open crater.
Across all four subtypes the shared and defining feature is that the subchondral bone plate is not yet exposed to the joint space — this is what separates the entire Grade 3 family from Grade 4.
On imaging, Grade 3 lesions produce a signal abnormality that extends through nearly the entire visible cartilage thickness, often making the lesion look "full thickness" at first glance. The key differentiator from Grade 4 is not depth alone but the status of the low-signal line at the very base of the defect:
• In Grade 3, a thin, continuous, dark (low-signal) line — the calcified cartilage/subchondral bone plate — remains intact underneath the defect, separating the fluid-bright defect from the marrow space.
• Grade 3D blistering has a distinct signature: a fluid-signal cleft or delamination plane running parallel to the joint surface beneath grossly intact-appearing superficial cartilage — this is easy to miss unless specifically sought, since the articular surface contour can look deceptively normal.
Because the defect is deep, MRI sensitivity rises substantially at this grade compared with Grade 1–2, but subgrade-level distinction (3A vs 3B vs 3C) is still imprecise on routine sequences and is not typically attempted from imaging alone.
Grade 3 lesions sit at an important treatment inflection point. Because the subchondral bone plate is still intact, chondral-surface-only techniques remain viable options: microfracture (which deliberately breaches the plate to recruit marrow-derived stem cells) becomes a rational choice precisely because the bone underneath is otherwise healthy, and cell-based approaches such as ACI/MACI can be applied onto a stable subchondral bed.
What separates management here from Grade 4 is the absence of an osseous component to the defect — there is no bone loss or exposed marrow to backfill, so the reconstructive problem is purely chondral. This is the last grade on the ICRS scale where that statement holds true.
Because 3D blistering can look nearly normal on the joint surface, it is one of the more common causes of arthroscopic under-grading — a probe test (gentle indentation with an arthroscopic hook) is required to detect the underlying delamination that MRI, by contrast, can sometimes reveal more reliably as a subtle cleft of fluid signal.
Grade 4 is reached the moment the subchondral bone plate itself is breached: cartilage loss is full-thickness and the underlying bone is directly exposed to the joint space, making this technically an osteochondral rather than a purely chondral lesion. Of every grade on the ICRS scale, this is the one most clearly visible on MRI — and the one most likely to redirect a patient toward bone-addressing surgical procedures.
A Grade 4 lesion is confirmed arthroscopically when a probe passes directly onto exposed subchondral bone with no residual cartilage cap — the calcified cartilage layer and the entire hyaline cartilage thickness above it are gone at the defect base. The exposed bone may appear:
• Eburnated: smooth, dense, ivory-like bone, typically seen in chronic, load-bearing defects where repetitive mechanical stress has sclerosed the exposed surface.
• Granular or bleeding: in more acute traumatic defects, exposed cancellous bone can appear irregular and may bleed on probing.
Because the bone plate is now open to the joint, this lesion behaves mechanically and biologically very differently from Grade 3 — synovial fluid, inflammatory mediators, and mechanical shear all have direct access to subchondral bone, which is far more pain-sensitive (richly innervated) than cartilage itself, which has no nerve supply at all. This is a major reason Grade 4 lesions are disproportionately symptomatic relative to their surface area.
Grade 4 is the most MRI-conspicuous lesion on the entire scale, characterized by:
• A focal cartilage defect extending completely through the cartilage thickness, with joint fluid (bright on T2/fluid-sensitive sequences) in direct contact with the subchondral bone surface — no intervening low-signal calcified layer remains.
• Subchondral bone marrow edema pattern: ill-defined, high-signal-intensity change on fluid-sensitive sequences (or low signal on T1) in the bone marrow immediately beneath the defect, reflecting a reactive response to altered mechanical loading and marrow vascular/inflammatory change. This is one of the most reliable secondary signs supporting a Grade 4 read even when the defect margins themselves are imperfectly seen.
• Possible subchondral cyst formation in chronic lesions, appearing as a well-circumscribed, fluid-signal focus within the marrow — a sign of long-standing altered biomechanics at the defect site.
Because of this combination of a frank surface defect plus a corroborating marrow signal change, sensitivity for Grade 4 lesions on routine MRI is consistently the highest of any ICRS grade — often exceeding 90%.
Bone marrow edema is a double-edged clue: it strongly increases confidence in identifying a Grade 4 lesion, but on its own (without a visible cartilage defect) it is nonspecific and can also reflect subchondral stress reaction, early avascular change, or a healing microfracture bed, so it should never be read as diagnostic of full-thickness chondral loss in isolation.
A focal area of subchondral bone exposure with overlying marrow edema can arise from more than one underlying process, and distinguishing them changes management even when the ICRS grade at the joint surface is identical:
• Degenerative/traumatic Grade 4 chondral defect: focal, usually at a site of known mechanical loading or prior trauma, with a well-defined cartilage defect margin.
• Osteochondritis dissecans (OCD): a subchondral fragment of bone (with or without attached cartilage) that may be partially or fully separated from the parent bone, typically in a younger patient at a characteristic location (classically the lateral aspect of the medial femoral condyle); MRI staging of OCD (fluid rim completely or incompletely surrounding the fragment) is a related but distinct question from ICRS chondral grading.
• Osteonecrosis (spontaneous or secondary): a geographic area of subchondral marrow signal change with a serpiginous low-signal rim, which can progress to subchondral collapse and secondary cartilage loss — here the bone pathology precedes and drives the cartilage injury, the reverse sequence from a primary chondral defect.
Recognizing which process is present matters because OCD and osteonecrosis have their own staging systems and treatment algorithms distinct from — though sometimes ultimately converging with — osteochondral grafting once a Grade 4-equivalent surface defect is present.
The presence of an osseous component fundamentally shifts treatment logic. Pure chondral-surface techniques designed to resurface an intact bony bed — microfracture, ACI/MACI — are less well suited to a defect that also has bone loss to replace, because they do not restore lost subchondral bone volume or contour.
Grade 4 lesions, particularly when deep or associated with subchondral cysts, more often point toward osteochondral procedures that replace both the cartilage and the underlying bone in a single graft: osteochondral autograft transfer (OATS/mosaicplasty) for smaller defects, or osteochondral allograft transplantation for larger lesions where donor-site morbidity from autograft harvest becomes limiting. The decision is never based on grade alone — lesion size, containment, location, and patient factors (age, activity level, alignment) are weighed together — but ICRS Grade 4 status is one of the clearest single triggers for considering a bone-addressing rather than a purely chondral-surface procedure.
Every grade examined so far has one behavior in common: MRI consistently correlates with, but tends to underestimate, arthroscopically confirmed lesion severity — and that gap is largest for the intermediate grades where treatment decisions are most sensitive to getting the number right. Understanding both the known limitations of MRI grading and the sequences that narrow the gap is what turns a grade on a report into a defensible surgical plan.
Across the cartilage-imaging literature, a consistent pattern holds: MRI sensitivity for detecting a chondral lesion, and for correctly matching its ICRS grade, rises with the severity of the lesion itself. Grade 1 lesions are frequently missed entirely; Grade 4 lesions are rarely missed. The practical consequence is a systematic bias toward under-grading — a true Grade 2 lesion is more likely to be read as normal or Grade 1 than a true Grade 3 lesion is to be read as Grade 2, and so on up the scale.
This is not simply a matter of reader skill. It reflects genuine physical limits: partial-volume averaging across finite slice thickness, insufficient contrast between adjacent cartilage zones on routine sequences, and the sub-millimeter scale of early lesions relative to standard in-plane resolution. Surgeons who plan procedures purely from an MRI-reported grade risk being surprised — usually in the direction of finding a worse lesion at arthroscopy than the imaging suggested.
A meaningful portion of the MRI–arthroscopy gap can be closed with acquisition choices tailored specifically to cartilage rather than general-purpose knee protocols:
• 3D fat-saturated spoiled gradient-echo (FS-GRE): acquires thin, often sub-millimeter isotropic slices with high cartilage-to-fluid contrast, dramatically improving delineation of the deep cartilage stripe that separates Grade 2 from Grade 3, and reducing partial-volume blurring of small surface fissures.
• Isotropic 3D fast spin-echo (e.g., CUBE/SPACE/VISTA-type sequences): allows multiplanar reformatting from a single high-resolution acquisition, useful for confirming lesion depth and extent from more than one plane without rescanning.
• T2 mapping: quantifies collagen fiber organization and water content zone-by-zone through cartilage thickness, potentially flagging Grade 1-level matrix disruption before any morphologic defect is visible at all.
• dGEMRIC (delayed gadolinium-enhanced MRI of cartilage): exploits the negatively charged contrast agent's inverse distribution relative to glycosaminoglycan (GAG) content, providing a compositional readout of proteoglycan loss — again capable of detecting biochemical degeneration that precedes any structural, morphology-based ICRS-style grade change.
These sequences do not replace arthroscopy as the reference standard, but at 3T with an optimized cartilage protocol, overall grading accuracy against arthroscopic ground truth can approach the high 80s to low 90s percent range — a substantial improvement over routine clinical imaging, particularly at the intermediate grades.
Compositional techniques like T2 mapping and dGEMRIC represent a conceptual shift: instead of asking "how deep is the visible defect," they ask "how healthy is the collagen-proteoglycan matrix," which can flag at-risk cartilage before it ever reaches a scoreable ICRS grade at all.
The table below condenses every stage of this simulator into a single reference: how deep the lesion runs, what it looks like on a well-optimized MRI protocol, and what that typically means for surgical candidacy. It is the "cheat sheet" a surgeon mentally reconstructs when reading a cartilage MRI report before ever picking up an arthroscope.
ICRS grade is never used in isolation to select a treatment; it is one input alongside lesion size, containment, location, and patient-specific factors (age, activity demands, limb alignment, and the health of surrounding cartilage) — but it is consistently the first branch point in the decision tree:
• Grade 0–1: typically managed non-operatively; incidental finding, not a surgical target unless clearly symptomatic and progressive.
• Grade 2: conservative management or activity modification is often favored; surgical intervention is considered selectively, usually only when a lesion is symptomatic and progressive despite conservative care.
• Grade 3 (bone plate intact): candidate for chondral-surface restoration — microfracture for smaller, lower-demand defects, or cell-based cartilage repair (ACI/MACI) for larger defects in appropriately selected, often younger and higher-demand patients, since these techniques depend on an intact, stable subchondral bed.
• Grade 4 (bone exposed): candidacy shifts toward osteochondral procedures — osteochondral autograft transfer for smaller contained defects, or osteochondral allograft for larger or bone-loss-associated lesions — because the reconstructive problem now includes replacing bone volume, not just resurfacing cartilage.
This grade-driven logic is exactly why accurate, reproducible MRI grading matters clinically: it is frequently the information a surgeon has in hand before ever entering the joint, and it directly shapes which procedure — and which graft, cell, or scaffold technique — is brought into the operating room as the primary plan.
| Product | Indication | Trial Design | Key Result |
|---|---|---|---|
| Grade 0–1 · Normal / Nearly normal | |||
| Grade 2 · Partial thickness | |||
| Grade 3 (A–D) · Deep, bone covered | |||
| Grade 4 · Full thickness / osteochondral |