The Painful Total Knee Replacement

The presentation. Pain in a knee that has undergone total knee arthroplasty, persisting or recurring beyond the expected recovery window of approximately three months. It is a pattern, not a diagnosis, and the task is to convert "painful knee replacement" into a mechanism. The first priority is to exclude periprosthetic joint infection, actively, with serology and aspiration rather than by impression.

The pain-free interval. Ask about it first, before examining anything, and pin down the date the pain started relative to the operation. Pain that never went away against pain after a good result is the single most powerful branch point in the entire differential: the answer halves it before you touch the patient.

• Never pain-free points to infection, instability, malrotation, an unrecognised extrinsic cause, or the wrong operation for the original pain
• New pain after a genuine period of good function points to aseptic loosening, wear and osteolysis, late haematogenous infection, periprosthetic fracture, or patellar clunk

The character of the pain. Each mechanism has its own complaint.

• Start-up pain settling after a few steps: loosening
• Giving way and swelling on stairs or slopes: instability
• Pain only on stairs and rising from a chair, localised anteriorly: patellofemoral or malrotation
• Rest and night pain with warmth: infection

Where it hurts. Point tenderness at the tibial tubercle or the medial joint line, or a Tinel sign over the infrapatellar branch of the saphenous nerve, each redirects the differential.

Adequate films. Confirm the pattern on a weight-bearing AP, a true lateral with the femoral condyles superimposed, a skyline at 30 degrees and a full-length hip-to-ankle alignment film. A supine AP alone will miss instability and coronal malalignment. Serial films compared with the immediate postoperative radiograph are worth more than any single film.

Describing the films in a viva.

"These are weight-bearing AP and lateral radiographs of a cemented total knee replacement. I would first want the immediate post-operative films for comparison and a long-leg alignment view. I am assessing component position and alignment, the cement–bone interfaces on both sides for progressive or circumferential lucency, evidence of subsidence or migration, the patella and joint line height, and the periprosthetic bone for osteolysis or periosteal reaction. My clinical priority in any painful knee replacement is to exclude periprosthetic joint infection with CRP, ESR and aspiration before attributing the pain to a mechanical cause."

What mimics the pattern. Some painful knees are not a painful TKR, and some findings are not really there.

• Normal early postoperative pain within the first three months, steadily improving, with no systemic features, normal inflammatory markers and an unremarkable examination. This is expected recovery, not a painful TKR, and it warrants reassurance and rehabilitation.
• Residual stiffness and soft-tissue soreness from an under-rehabilitated but well-fixed, well-aligned knee.
• Radiographic pseudo-lucency from beam obliquity: a lucent line that disappears on a properly tangential repeat view was never there.
• Metal artefact on cross-sectional imaging simulating osteolysis when artefact reduction sequences have not been used.

Plain radiographs. First line in every case, alongside the immediate postoperative films for comparison. They show component position, coronal and sagittal alignment, the cement–bone interfaces, subsidence, joint line height, patellar tracking, osteolysis, polyethylene thickness asymmetry and fracture. They change management by identifying loosening, fracture or gross malalignment that mandates revision planning.

View larger

Knee aspiration. Performed off antibiotics for at least two weeks, under strict aseptic technique, for synovial white cell count, neutrophil percentage and culture held for extended incubation. Alpha-defensin and synovial CRP add value where the count is equivocal. A positive result changes management completely, converting the problem from mechanical revision to a debridement or staged revision pathway.

Alpha-defensin is two different tests. The laboratory assay pools at 95 per cent sensitivity and 97 per cent specificity. The lateral flow cassette used at the bedside pools at 85 and 90 per cent, with sensitivity significantly lower, and performs no better than a leukocyte esterase strip. A negative cassette is weak reassurance, not an exclusion.

Rotational CT (Berger protocol). Indicated for anterior knee pain, patellar maltracking or unexplained pain with well-fixed components. It measures femoral component rotation against the transepicondylar axis and tibial tray rotation against the tibial tubercle, and also demonstrates component overhang and maps osteolysis. It changes management by justifying component revision for malrotation rather than an isolated polyethylene exchange.

CT with metal artefact reduction. Maps osteolysis or confirms equivocal loosening.

Stress radiographs or examination under anaesthesia. Used where instability is suspected but not confidently demonstrated in clinic. They quantify coronal opening and flexion gap laxity, and support the decision between a thicker polyethylene insert and a more constrained implant.

Nuclear medicine. A triple-phase bone scan or a labelled leucocyte scan with marrow subtraction is reserved for the genuinely unexplained painful replacement when serology, aspiration and CT are unhelpful. It is uninterpretable within twelve months of implantation because of expected periprosthetic remodelling. Its main value is a strongly negative scan, which makes both loosening and infection unlikely and supports non-operative management. Hybrid SPECT/CT localises the compartment when plain films are unimpressive.

View larger

View larger

Diagnostic local anaesthetic injection. Powerful and underused. Abolition of pain after an intra-articular hip injection confirms referred hip pain and settles the hip question; abolition after a targeted block at a Tinel point confirms a neuroma. Either redirects surgery away from the knee entirely.

When no further imaging is needed. Expected early recovery needs none. Equally, when the diagnosis is established clinically, by an extensor lag with a palpable gap or a classic hip referral pattern, proceed on that basis rather than accumulating negative scans.

Where to start. Periprosthetic joint infection is the must-not-miss; instability and malalignment are the commonest mechanical cause of persistent pain after TKR. Wear, osteolysis, periprosthetic fracture and patellar clunk are late mechanical failures after years of good function, and each has a specific operation once identified. The table sets each cause against the feature that discriminates it and the test that confirms it.

View larger

Once the pain-free interval is established, interrogate the knee in order.

1. Exclude infection actively, and score it. Send CRP and ESR in every painful knee replacement, no exceptions; their result decides whether the pathway becomes septic or aseptic. A normal pair substantially lowers but does not exclude the probability of infection, because low-virulence organisms such as Cutibacterium acnes and coagulase-negative staphylococci frequently give normal serology. Aspirate if either marker is raised, the story is atypical, there was no pain-free interval, or there was prolonged wound drainage.

The 2018 ICM score. Apply it as a weighted score rather than a checklist. Two positive cultures or a sinus tract are diagnostic alone; otherwise the minor criteria are summed:

• Synovial white cell count over 3,000/µL, alpha-defensin and leukocyte esterase: 3 points each
• Serum CRP over 1 mg/dL and D-dimer over 860 ng/mL: 2 points each
• ESR over 30: 1 point

Six or more is infected. A score of 2 to 5 is inconclusive: it does not mean aseptic, it means the answer is deferred to intraoperative histology, purulence and culture, and the operation must be planned accordingly.

View larger

2. Intrinsic or extrinsic to the knee? Examine the hip through a full range with logroll and internal rotation, perform a straight-leg raise and a neurological examination of the limb, and palpate the distal pulses. Pain in a dermatomal distribution, or reproduced by hip rotation with an unremarkable knee examination, redirects investigation entirely. Approximately one in five painful replacements is explained here, which is why the hip and spine are examined, and the pulses felt, before cross-sectional imaging of the knee is ordered.

• Hip osteoarthritis refers anteromedial knee and thigh pain, with a pain-free knee examination and restricted, painful hip rotation.
• The lumbar spine refers pain from the L3–L4 nerve roots; lumbar stenosis gives buttock-to-calf pain worsened by walking, with a normal knee examination.
• Vascular claudication has a reproducible walking distance, absent pulses, and relief with standing still rather than sitting.

3. Read the films against the immediate postoperative series. Retrieve the immediate postoperative films and compare component position, cement mantle, joint line height and coronal alignment side by side. Look for lucency, subsidence, change in component position, osteolysis, asymmetric polyethylene thickness on weight-bearing views, patella baja and periprosthetic fracture. A single film in isolation answers very little; a series answers most of the fixation question.

When a lucent line matters. Thin, non-progressive lucent lines under the tibial tray are seen in many asymptomatic, well-functioning knees, and a stable non-progressive 1 mm line at the cement–bone interface is normal. Loosening requires a lucency that is progressive, greater than 2 mm or circumferential, or component migration or subsidence on serial films.

4. Define the mechanical phenotype clinically. Assess stability in full extension, at 30 degrees of mid-flexion and at 90 degrees of flexion, then extensor mechanism integrity, patellar tracking and the presence of a clunk. These findings are almost entirely clinical; the static radiograph will not make them for you.

• Coronal laxity in extension: collateral or extension gap imbalance
• Laxity at 90 degrees with recurrent effusion and stair difficulty: flexion instability
• Symptomatic laxity at 30 degrees with a raised joint line and a downsized femur: mid-flexion instability

5. Scan to answer a question. Decide the question before ordering cross-sectional imaging and state it on the request; a rotational CT is ordered to answer the malrotation question, not as a fishing trip. Nuclear imaging is a late resort.

6. No diagnosis, no operation. If every investigation is negative, say so explicitly to the patient and give the number rather than an opinion. Across 24,540 first revision knee arthroplasties in the UK National Joint Registry, revision for unexplained pain left a mean Oxford Knee Score of 25, with only 55 per cent of patients improving enough to count as responders. That is among the poorest results of any indication, with only stiffness worse (24 points, 48 per cent), and it has to be set against a re-revision risk that compounds and a serious 90-day complication rate of several per cent. Revision for aseptic loosening, by contrast, has among the better outcomes once the diagnosis is secure.

Unexplained pain is a diagnosis, not a failure. Revision without a diagnosis reliably fails, and an exchange of polyethylene "to see" is not the answer. Reassess the pain phenotype, consider central sensitisation and CRPS, and remember that anxiety and depression are the commonest preoperative predictor of dissatisfaction. Manage it with structured non-operative care, pain services and physiotherapy, and re-image after an interval: serial radiographs at six-month intervals may yet reveal progressive loosening that was not apparent, and a diagnosis found late is still better than an operation done blind.