What a bone marrow lesion actually is

An MRI report mentioning 'bone marrow oedema' or a 'bone marrow lesion' can feel alarming, yet the finding is far more common — and far less definitive — than the language implies.

Bone marrow lesions (BMLs) show up on MRI as areas of abnormal signal in the subchondral bone: the dense layer of bone that sits just beneath the cartilage surface of the knee joint. Rather than pointing to a single disease, this signal reflects a spectrum of changes that may be occurring simultaneously — fluid accumulation within the bone marrow, tiny trabecular microfractures invisible on plain X-ray, or localised reactive oedema triggered by stress on the joint. The MRI captures the consequence; it cannot always tell you the cause.

BMLs are among the most frequently encountered incidental findings on knee MRI, and they are especially prevalent in people with knee osteoarthritis. Crucially, the imaging label and the patient's symptoms are two separate things. Many BMLs produce no pain whatsoever, and recent evidence suggests that when pain does arise, it is often the combination of a BML with other subchondral changes — such as bone attrition or a subchondral cyst — that drives discomfort, rather than the BML in isolation.

One detail that matters for both prognosis and treatment is that BMLs are not fixed. They can expand, contract, or resolve over months, depending on loading patterns, activity levels, and other joint changes. This dynamic nature is what makes them useful as clinical markers — and it is also why, in many reversible cases, the right management can lead to meaningful improvement.

When and why BMLs actually cause pain

Whether a bone marrow lesion is actually causing a patient's pain turns out to be a more complicated question than the MRI report implies. A large 2025 study — the Bunkyo Health Study, involving 1,145 older adults — found that a BML on its own carries only a weak statistical association with knee pain (odds ratio 1.32, with confidence intervals crossing 1). In plain terms: a BML in isolation is not a reliable pain predictor.

What changes the picture is the state of the surrounding bone. When a BML co-exists with subchondral bone attrition — thinning and flattening of the bone surface — the odds of significant pain roughly double (OR 2.22). When it co-exists with a subchondral bone cyst, the risk is similarly elevated (OR 1.79). Importantly, these associations held regardless of whether the overlying cartilage was also damaged, pointing to a pain pathway rooted in subchondral structural compromise rather than cartilage loss alone.

The proposed mechanism runs through the bone itself: structural weakening disrupts the normal architecture of the subchondral plate, raising intraosseous pressure, impairing local microvascular flow, and sensitising the nerve endings embedded within subchondral bone. It is this convergence of mechanical and biological stress — not the MRI signal in isolation — that appears to generate sustained nociceptive input.

Location adds a further clinical dimension. Medial femorotibial BMLs are specifically associated with pain during walking and prolonged standing; lateral patellofemoral BMLs correspond more closely to pain on climbing stairs. For patients trying to map their own symptom pattern, this distinction can be genuinely useful. BML severity has also been shown to correlate with proximal tibial inclination, reinforcing that abnormal biomechanical loading plays a causal role in forming and sustaining these lesions — not just progressive tissue wear.

Longitudinal data from the Osteoarthritis Initiative, tracking over 1,400 varus knees across 24 months, showed that changes in BML score moved in step with changes in weight-bearing pain. When load is reduced or redistributed, lesion severity tends to follow — and so, often, does pain.

Conservative management: the logical starting point

For most patients with a newly identified bone marrow lesion, structured conservative management is both the appropriate first step and, in reversible cases, often sufficient.

First-line options centre on reducing mechanical load: relative rest, joint offloading via a crutch or unloading brace, and a short course of oral NSAIDs to manage pain and local inflammation. Physiotherapy should address the biomechanical drivers of the lesion — quadriceps activation, lower-limb alignment, and load distribution through the knee — rather than treating the joint in isolation. With this approach, reversible BMLs typically resolve within three to six months.

Beyond this foundation, a 2025 systematic review by Andriolo and colleagues identified additional benefit from extracorporeal shockwave therapy (ESWT), pulsed electromagnetic fields (PEMF), and hyperbaric oxygen — each capable of improving local bone perfusion and reducing intraosseous oedema. These are reasonable options to explore with a specialist when first-line measures plateau; none replaces the core principle of offloading and rehabilitation.

Not all additions to the conservative regimen carry the same weight, however. Vitamin D supplementation — sometimes considered because of its role in bone metabolism — does not appear to act as a structural modifier here. A two-year randomised controlled trial in symptomatic knee OA found no significant reduction in BML volume with 800 IU cholecalciferol daily compared with placebo; the confidence interval was wide enough to effectively exclude a clinically meaningful effect.

The important caveat is that conservative care has limits. BMLs co-existing with established subchondral attrition or bone cysts are less likely to resolve with offloading alone, and for these patients — where the structural picture driving pain is more complex — escalation to targeted injection therapy becomes a realistic next consideration.

Injection therapies that target the subchondral bone

Three injectable approaches have accumulated the most meaningful clinical data for BMLs, each working through a distinct mechanism.

Calcium phosphate subchondroplasty

The principle here is structural: a bone-substitute material is injected directly into the subchondral lesion to fill the defect, stabilise the compromised trabecular architecture, and redistribute mechanical load away from the damaged zone. It is the most extensively studied of the targeted injectables. Pooled data from eight studies showed mean VAS pain scores falling from 7.90 to 2.76, and IKDC functional scores rising from 30.5 to 53.0. A broader systematic review across ten studies (540 patients, follow-up ranging from six months to seven years) confirmed consistent patient-reported outcome improvements. A realistic caveat sits alongside these gains: total knee arthroplasty conversion rates of 14–30% were recorded at two years, reflecting the underlying severity of disease in many patients rather than a failure of the technique per se. The most commonly reported complication was extravasation of calcium phosphate material into the joint space.

BMAC subchondral injection

Bone marrow aspirate concentrate (BMAC), harvested from the posterior superior iliac crest, is delivered under fluoroscopic guidance directly into the BML rather than the joint cavity. A 2026 retrospective study in 62 knee OA patients with MRI-confirmed lesions reported significant improvements in KOOS, WOMAC, and VAS scores across all four follow-up points to 12 months. MOAKS BML scores were measurably reduced at three months. Patient characteristics mattered: younger individuals and those with BMI at or below 30 kg/m² achieved superior outcomes — a useful consideration when discussing suitability.

Intra-articular Arthrosamid (iPAAG)

A single intra-articular injection of polyacrylamide hydrogel (Arthrosamid) has been associated with reduction in patellofemoral BMLs in advanced knee OA. The mechanism here is indirect — the hydrogel acts as a sustained cushioning agent within the joint, reducing the mechanical stress transmitted to the subchondral bone rather than treating the lesion directly. This load-redistribution pathway distinguishes it from the two subchondral approaches above.

What the evidence can and cannot tell us yet

The tools to measure BMLs rigorously are now ready; the trials that use them are not.

Deep-learning convolutional neural networks trained for MRI-based BML segmentation have achieved Pearson R² of 0.94 against expert manual annotation, making automated volumetric measurement both feasible and reproducible at scale. That matters because it removes one of the traditional obstacles to running large prospective trials: the prohibitive time cost of manually segmenting lesions across hundreds of scans. What has not yet followed is a generation of well-powered, randomised controlled trials that exploit this readiness.

The specific gaps are worth naming clearly. No published trial has compared injection modalities head-to-head using BML volume change as a primary structural endpoint — existing studies rely on patient-reported outcome measures, which are important but cannot confirm whether the underlying lesion is resolving or merely less symptomatic. Radiological follow-up beyond approximately 12 months post-injection has not been published for any of the three approaches. And all current data originate from small, retrospective, or uncontrolled series; the signal around patient selection — that younger individuals with BMI at or below 30 kg/m² respond better to biological injection — remains hypothesis-generating rather than trial-confirmed.

None of this is a reason to dismiss the therapies. It is a reason to interpret the encouraging early signals honestly. The methodological infrastructure for better trials now exists; the research community has not yet fully used it.

Getting the right assessment for your knee

Persistent pain — especially symptoms that remain after 6–8 weeks of structured conservative management, or imaging showing BMLs alongside subchondral attrition or cyst formation — is the clearest signal that specialist input adds more than continued self-management.

A useful assessment covers more than an MRI review. Clinical history, weight-bearing examination, and lower-limb alignment assessment all contribute information the scan cannot supply alone. A medial femorotibial BML in a patient with significant varus alignment presents a different clinical problem to the same finding in a mechanically neutral knee; a BML co-existing with subchondral bone attrition carries meaningfully higher pain risk than an isolated lesion. These distinctions determine which treatment tier is appropriate and whether the lesion is on a trajectory that warrants early intervention rather than observation.

The treatment sequence follows a clear logic: conservative management first, targeted injection if conservative care does not resolve symptoms, and surgical discussion where structural compromise is substantial. Patient-specific factors — age, BMI, activity demands, and alignment — all influence which path is worth pursuing. Knees with BMLs alongside significant OA may be heading toward joint replacement; early specialist input can help identify and, where possible, slow that trajectory.

For patients weighing whether a formal assessment is the next step, a structured suitability questionnaire is available at amsk.co.uk.

Attending a specialist consultation with a copy of the MRI images and report, a note of precisely which activities provoke symptoms, and a clear account of what has already been tried makes that assessment considerably more productive than arriving with the radiology letter alone.

1. [1] Fast quantitative bone marrow lesion measurement on knee MRI for the assessment of osteoarthritis. (2022). https://doi.org/10.1016/j.ocarto.2022.100234 https://doi.org/10.1016/j.ocarto.2022.100234
2. [2] Bone marrow lesion severity was associated with proximal tibial inclination in early knee osteoarthritis. (2021). https://doi.org/10.1007/s00167-020-06378-7 https://doi.org/10.1007/s00167-020-06378-7
3. [3] Subchondral bone marrow aspirate concentrate injection improves clinical outcomes and reduces bone marrow lesions in knee osteoarthritis: 12-month retrospective analysis. (2026). https://doi.org/10.1186/s13018-026-06814-3 https://doi.org/10.1186/s13018-026-06814-3
4. [4] Effect of Vitamin D supplementation on synovial tissue volume and subchondral bone marrow lesion volume in symptomatic knee osteoarthritis. (2019). https://doi.org/10.1186/s12891-019-2424-4 https://doi.org/10.1186/s12891-019-2424-4
5. [5] Bone marrow lesion coexisted with subchondral bone attrition and/or subchondral bone cyst is associated with knee pain in knee osteoarthritis regardless of cartilage lesion: the Bunkyo Health Study. (2025). https://doi.org/10.1186/s13075-025-03644-2 https://doi.org/10.1186/s13075-025-03644-2
6. [6] Association of subchondral bone marrow lesion localization with weight-bearing pain in people with knee osteoarthritis: data from the Osteoarthritis Initiative. (2020). https://doi.org/10.1186/s13075-021-02422-0 https://doi.org/10.1186/s13075-021-02422-0
7. [7] Subchondroplasty in the treatment of bone marrow lesion in early knee osteoarthritis: a systematic review of clinical and radiological outcomes. (2022). https://doi.org/10.1016/j.knee.2022.10.004 https://doi.org/10.1016/j.knee.2022.10.004
8. [8] Calcium phosphate injection of symptomatic bone marrow lesions of the knee: what is the current clinical evidence?. (2020). https://doi.org/10.1186/s43019-019-0013-3 https://doi.org/10.1186/s43019-019-0013-3