Routine wrist MRI misses many scapholunate ligament tears, arthroscopy study finds

Standard practice wrist magnetic resonance imaging is highly reliable when it reports the scapholunate ligament as intact, but it misses a large share of actual tears, according to a retrospective study published in Archives of Orthopaedic and Trauma Surgery. In a cohort of 864 patients whose preoperative MRI findings were compared with subsequent wrist arthroscopy, the scan showed a specificity of 91.1 percent for detecting scapholunate ligament lesions but a sensitivity of only 42.0 percent.

The scapholunate interosseous ligament, the structure at the center of the study, is the most commonly injured carpal ligament and a leading cause of chronic wrist pain and instability. Typical causes include hyperextension injuries of the wrist and repetitive trauma, such as prolonged use of crutches. Untreated tears can progress through a predictable cascade of carpal malalignment, culminating in scapholunate advanced collapse wrist with radiocarpal and midcarpal osteoarthritis. Early and accurate diagnosis is therefore considered essential to prevent irreversible degenerative changes, which is why the diagnostic performance of everyday imaging matters for patients with wrist injuries.

Wrist arthroscopy remains the diagnostic gold standard, providing direct visualization and grading of ligament integrity under dynamic loading conditions. However, because arthroscopy is invasive, magnetic resonance imaging has been widely adopted as the primary non-invasive imaging modality for assessing intrinsic wrist ligaments. Reported diagnostic accuracy has varied widely, with standard practice clinical MRI often demonstrating limited sensitivity, in the range of 19 to 79 percent, for scapholunate ligament tears despite reasonable specificity of 69 to 98 percent. Partial tears are consistently underdetected compared with complete ruptures. The authors of the new study noted that large-scale data from standard clinical imaging, performed under variable conditions without standardized research protocols, had remained scarce.

The research team, led by Sebastian Spodzieja and colleagues at the Vulpius clinic in Bad Rappenau, Germany, reviewed all consecutive patients who underwent preoperative wrist MRI followed by diagnostic or therapeutic wrist arthroscopy at their institution between January 2013 and December 2021. Patients were included if they had preoperative MRI with an original report available. Exclusion criteria included MRI performed after arthroscopy, examinations at 3 Tesla, an interval of more than one year between MRI and arthroscopy, prior surgery on the scapholunate ligament, operations not performed by a board-certified hand surgeon, and incomplete medical records. Reports that did not explicitly mention the scapholunate ligament were also excluded.

Of 889 patients included in the analysis, complete paired data were available for 864, with a mean age of 42.4 years, a range of 10 to 83 years, and 51.3 percent male. The mean interval between MRI and arthroscopy was 51.5 days, ranging from 1 to 295 days. The right wrist was operated in 56.9 percent of cases. MRI technical details showed that contrast medium was used in 21.3 percent of examinations and a dedicated wrist coil in 37.7 percent. Arthroscopy was performed in a dry technique using standard portals, with both radiocarpal and midcarpal compartments systematically inspected and ligament integrity assessed dynamically with a probe.

Grading followed the Geissler classification, in which grade I describes attenuation or haemorrhage, grade II a partial tear with a gap of less than one millimeter under probe pressure, grade III an incomplete tear through which the probe passes, and grade IV a complete tear. Partial tears, grades I through III, were grouped together for the primary analysis. MRI reports were categorized as intact, partial tear, or complete tear, and arthroscopic findings were recorded by the operating surgeon. All procedures were performed or supervised by three experienced board-certified hand surgeons, and no attempts were made to retrieve missing information from examiners, in order to keep the assessment of routine report quality unbiased.

The MRI reports classified the ligament as intact in 725 cases, or 83.9 percent, partially torn in 79 cases, or 9.1 percent, and completely torn in 60 cases, or 6.9 percent. Arthroscopy told a somewhat different story: intact ligaments in 676 cases, or 78.2 percent, partial tears in 111 cases, or 12.9 percent, and complete tears in 77 cases, or 8.9 percent. Agreement between the two modalities was limited, with a simple kappa of 0.339 and a weighted kappa of 0.399, values the authors classify as only fair agreement. A McNemar test also indicated a statistically significant discrepancy between the imaging and surgical findings.

When findings were dichotomized as any lesion versus intact, MRI showed a sensitivity of 42.0 percent, a specificity of 91.1 percent, a positive predictive value of 56.8 percent, and a negative predictive value of 85.0 percent. When partial lesions were excluded and only complete tears were considered, performance improved: sensitivity rose to 51.5 percent, specificity to 97.3 percent, positive predictive value to 67.3 percent, and negative predictive value to 94.9 percent. Even so, the authors concluded that the negative predictive value remained insufficient to reliably rule out clinically relevant ligament injuries, underscoring the limited ability of routine MRI to exclude scapholunate pathology, particularly partial lesions.

Subgroup analyses revealed patterns in how imaging technique influenced performance. Sensitivity was higher without a documented dedicated wrist coil than with one, at 48.3 percent versus 32.4 percent, and higher with contrast medium than without, at 50.0 percent versus 39.4 percent. The highest sensitivity, 66.7 percent, occurred in examinations without a wrist coil but with contrast, although this subgroup comprised only 106 patients and the numbers were small. Sensitivity also increased over the study period, from 33.3 percent in 2013 to 54.2 percent in 2021, while specificity remained stable between 86.4 and 96.8 percent. The authors interpreted these figures as indicating only a modest gain in sensitivity compared with benchmarks reported in earlier standard-practice studies, including a 2013 analysis reporting 32 percent sensitivity and a 2020 analysis reporting 38.5 percent sensitivity with 91.0 percent specificity.

The study’s authors emphasized that the marked discrepancy between MRI and arthroscopy in detecting partial lesions is the most important result, because partial scapholunate injuries are common and may precede more advanced instability. The ligament’s thin dorsal, membranous, and palmar components can be difficult to assess on routine static MRI, while arthroscopy allows direct visualization and probing. The clinical implication, they wrote, is that a negative MRI should be interpreted cautiously when suspicion of ligament injury remains high; persistent pain, mechanical symptoms, or clinical signs of instability should prompt further evaluation, and arthroscopy should be considered when definitive structural assessment is needed. They positioned MRI as complementary to, rather than a replacement for, arthroscopy, noting that MRI remains valuable for identifying alternative causes of wrist pain and detecting clearly advanced tears.

Several limitations accompany these findings. The retrospective design introduces selection bias, since only patients who proceeded to arthroscopy were included. The MRI examinations were performed in standard practice with heterogeneous protocols and variable scanner settings, which the authors chose deliberately to reflect real-world clinical care in a predominantly rural German healthcare setting where access to 3-Tesla imaging is limited. The interval between MRI and surgery was not uniform, so progression of pathology in some patients cannot be excluded, and arthroscopic grading itself is operator dependent, with previous studies showing only fair to moderate interobserver agreement for several arthroscopic wrist findings. Missing data in a subset of examinations were also acknowledged, although the cohort size of 864 paired observations was large enough, in the authors’ view, to support the conclusion that in non-specialized clinical settings, persistent clinical suspicion despite a negative MRI should prompt consideration of arthroscopic evaluation.

Subject of Research: Diagnostic accuracy of standard practice wrist MRI for detecting scapholunate ligament injuries compared with arthroscopy

Article Title: Diagnostic accuracy of standard practice wrist MRI for scapholunate ligament injuries: a retrospective arthroscopic correlation study

Article References: Spodzieja, S., Boever, J., Rau, J., Bruckner, T., Müller, H., Wulf, J., Unglaub, F., & Cavalcanti Kussmaul, A. (2026). Diagnostic accuracy of standard practice wrist MRI for scapholunate ligament injuries: a retrospective arthroscopic correlation study. Archives of Orthopaedic and Trauma Surgery, 146(1), Article 373. https://doi.org/10.1007/s00402-026-06541-8

Image Credits: AI Generated

DOI: 10.1007/s00402-026-06541-8

Keywords: wrist MRI, scapholunate ligament, arthroscopy, diagnostic accuracy, sensitivity, specificity, carpal instability, Geissler classification, hand surgery, retrospective study, wrist pain, ligament tears