Combined CT Measurements Sharpen Prediction of Kneecap Instability, Study Finds

Kneecap instability is one of the most frustrating problems in knee medicine. When the patella fails to glide smoothly along the femoral groove at the front of the knee, patients can experience recurrent dislocations, pain, and a gradual wearing away of the joint surfaces that may end in early arthritis. Diagnosing the condition accurately has always been difficult, because the underlying causes are many and varied: the shape of the trochlear groove, the alignment of the leg, rotational abnormalities of the femur or tibia, and the tension of the surrounding soft tissues can all contribute in different combinations from one patient to the next. A new study from Wuhan Fourth Hospital in China, published in BMC Medical Imaging, now offers a quantitative way forward, showing that a combination of computed tomography measurements can predict patellofemoral joint instability with markedly better accuracy than any single measurement alone.

The research team, led by radiologists Man Yang and Xiang Feng together with sports medicine specialist Tao Li, recruited 70 patients diagnosed with patellofemoral joint instability and compared them with 45 asymptomatic controls. Because the study was retrospective and used anonymized data, the ethics committee of Wuhan Fourth Hospital waived the requirement for individual informed consent, and the work was conducted in accordance with the Declaration of Helsinki. All participants underwent knee CT scanning, and the researchers extracted a set of standardized measurements designed to capture the alignment and rotational anatomy of the lower limb around the patellofemoral joint.

Four parameters formed the core of the analysis. The first, and historically the most widely used, is the tibial tubercle to trochlear groove distance, abbreviated TT-TG. This measures the horizontal offset between the bony bump on the tibia where the patellar tendon attaches and the groove in the femur where the kneecap should track. A large TT-TG distance means the extensor mechanism effectively pulls the kneecap laterally with each contraction of the quadriceps, a well-recognized risk factor for dislocation. The second measurement, TT-ME, replaces the trochlear groove landmark with the medial epicondyle of the femur, a point that may be easier to identify reliably when the trochlear groove itself is shallow or dysplastic.

The remaining two parameters extend the assessment beyond the transverse plane. The knee joint rotation angle, or KJRA, quantifies rotational malalignment of the knee as a whole, capturing the degree to which the joint is internally or externally twisted relative to the mechanical axis of the limb. The tibial tubercle torsion angle, TT-TA, goes further down the leg, measuring the torsional relationship between the tibial tubercle and the distal tibia. Rotational abnormalities of the tibia have increasingly been implicated in patellar maltracking, but they are often overlooked in routine clinical workup, which has traditionally focused on the TT-TG distance alone. By including these parameters, the researchers aimed to test whether a broader anatomical picture improves diagnostic discrimination.

Methodological rigor was a particular concern for the team. Measurements were taken using standardized methods, and interobserver reliability was formally assessed for the TT-ME distance, an important step because a measurement that different observers cannot reproduce consistently is of limited clinical value regardless of its theoretical appeal. Statistical analysis was performed with SPSS 19.0. The researchers first ran univariable regression analyses to identify which parameters differed between patients and controls, and then built a multivariable model to determine which factors remained independently associated with instability when the others were taken into account.

The results were clear-cut. Patients with patellofemoral joint instability showed significantly higher values than controls in all four CT parameters: TT-TA, KJRA, TT-ME, and TT-TG, with differences reaching statistical significance at the level of P less than 0.05. Interestingly, the patient group was also significantly younger than the control group, a finding consistent with the clinical observation that recurrent patellar dislocation tends to affect adolescents and young adults, particularly those engaged in sports. Youth, it turns out, was not merely a demographic footnote but a genuine component of the predictive model itself.

When the researchers entered all candidate variables into the multivariable regression, three emerged as independent predictors of patellofemoral joint instability. The tibial tubercle to midepicondylar distance carried an odds ratio of 1.535, meaning each unit increase in this measurement substantially raised the odds of instability. The tibial tubercle torsion angle contributed an odds ratio of 1.131, confirming that tibial torsion carries independent diagnostic information even after accounting for the classic transverse-plane measurements. Age, with an odds ratio of 0.914, acted protectively in the model: older individuals were less likely to present with instability, all else being equal. Notably, the conventional TT-TG distance and the knee joint rotation angle did not survive as independent predictors once the stronger variables were included, a result that challenges some long-standing diagnostic habits.

The performance of the resulting multi-parameter predictive model was the headline finding of the study. Evaluated with receiver operating characteristic curve analysis, the combined model achieved an area under the curve, or AUC, of 0.929, a figure generally interpreted as excellent diagnostic discrimination. This substantially outperformed the individual parameters on their own: TT-ME alone achieved an AUC of 0.893, while TT-TA alone managed only 0.722. In practical terms, combining the measurements allowed the model to correctly separate unstable from stable knees far more often than any single anatomical landmark could, demonstrating that the information carried by each parameter is partly complementary rather than redundant.

Accuracy alone, however, does not guarantee that a diagnostic model is useful at the bedside or in the clinic. To address this, the researchers employed decision curve analysis, a technique that quantifies the net benefit of acting on a model’s predictions across the full spectrum of probability thresholds. The analysis suggested potential clinical utility, showing consistent net benefit gains across a wide range of threshold probabilities. Calibration curves were also used to confirm that the model’s predicted probabilities matched observed outcomes, providing further reassurance that the model is not merely discriminating between groups but doing so in a quantitatively trustworthy way.

The implications for clinical practice are potentially significant. Surgeons planning procedures for patellar instability, from tibial tubercle osteotomy to rotational correction, rely on accurate anatomical assessment to choose the right operation for the right patient. A validated multi-parameter CT model could help identify which patients truly have pathogenic malalignment, support surgical decision-making, and reduce reliance on any single, imperfect measurement. The study does have limitations inherent to its design: it was retrospective, involved a relatively modest sample of 70 patients and 45 controls, and the findings will need external validation in independent cohorts before widespread adoption. The researchers received no specific funding for the work and declared no competing interests. Nevertheless, the central message stands out clearly: when it comes to predicting kneecap instability from CT imaging, the whole anatomical picture is worth more than the sum of its individual landmarks, and combining tibial tubercle to midepicondylar distance, tibial tubercle torsion angle, and patient age into a single predictive model raises diagnostic accuracy to a level that individual measurements cannot reach on their own.

Subject of Research: CT-based multi-parameter prediction of patellofemoral joint instability

Article Title: Application of CT imaging parameters in predicting patellofemoral joint instability: a multi-parameter combined analysis

Article References: Yang, M., Feng, X., Wang, Q., Zhang, Y., Liu, J., Hu, X., & Li, T. (2026). Application of CT imaging parameters in predicting patellofemoral joint instability: a multi-parameter combined analysis. BMC Medical Imaging. https://doi.org/10.1186/s12880-026-02814-1

Image Credits: AI Generated

DOI: 10.1186/s12880-026-02814-1

Keywords: patellofemoral joint instability, computed tomography, TT-TG distance, TT-ME distance, tibial tubercle torsion angle, knee joint rotation angle, predictive model, ROC analysis, decision curve analysis, orthopaedics, medical imaging, knee dislocation