Abstract
Background:
Patellar instability (PI) is a potentially debilitating knee condition that occurs most commonly in children and adolescents. Patellar height (PH) in the form of patella alta has been historically described, along with several others, as an anatomic risk factor for PI. However, numerous methods for evaluating PH exist without an established gold standard. Several ratio measurements have been proposed and studied including the Caton-Deschamps (CD) ratio. This can be measured on radiographs or magnetic resonance imaging (MRI) and is widely used due to its reliability to quantify PH across all age groups. A more recent measurement, the patellotrochlear index (PTI) quantifies PH by measuring the amount of patellar and trochlear chondral overlap on a sagittal MRI image and is expressed as a percentage. The PTI has been compared to the CD ratio as well as other PH measurements on MRI and has been proposed as a more accurate representation of clinically meaningful PH. We hypothesized that the PTI would have better inter- and intra-rater reliability than the CD ratio measured on standard radiographs, that most patients with PI would not have patella alta as represented by the PTI, and that there would be no correlation between the PTI and CD ratio.
Further, the effect of PH on other radiographic measurements involved in assessing PI has not been evaluated. Given the effect trochlear dysplasia (TD) has on other radiographic parameters, such as patellar tilt and the tibial tubercle to trochlear groove distance, we sought to evaluate the correlation between PTI and the lateral trochlear inclination (LTI) angle and lateral patellar inclination (LPI) angle in a cohort of children and adolescents with PI. We hypothesized that the PTI would have no correlation between LTI or LPI in this patient population.
Methods:
Patients aged 9 to 18 years treated for PI between January 2014 and August 2017 were identified. The CD ratio was measured on lateral radiographs and PTI was measured on sagittal MRIs (Figure 1). Based on a previous study, patella alta with regards to PTI was defined as < 12.5%. A CD ratio > 1.2 was defined as alta based on historic standards. The LTI and LPI were measured on axial MRI images using recently described techniques that have high or near-perfect reliability (LTI: Interrater ICC = 0.971, 95% CI: 0.88 – 0.97. LPI: Interrater ICC = 0.885, 95% CI: 0.77 – 0.95). All measurements were performed by two independent observers. For reliability analysis, a cohort of 30 patients were randomly selected. Measurements were performed for this cohort by three independent observers and re-performed after a minimum of two weeks. Inter- and intra-rater correlation coefficients were calculated from this subgroup, and regression analysis was performed on the entire cohort.
Results:
Sixty-five patients met inclusion criteria for this study. PTI measurements had less variability and near perfect agreement between observers (ICC=0.92, 95% CI: 0.83-0.96) and within observers (ICC=0.99, 95% CI: 0.97-0.99), while CD ratio measurements had moderate agreement between observers (ICC=0.62, 95% CI: 0.34-0.80) and high agreement within observers (ICC=0.82, 95% CI: 0.65-0.91). Only 3.1% of patients had a PTI < 12.5% and were classified as patella alta. 62.5% of patients had patella alta based on the CD ratio. Regression analysis demonstrated a significant but very weak correlation between PTI and CD ratio (r = -0.30, beta = -0.004, p = 0.0156). In this study, 71% of patients had an LTI <11°, classifying them as having TD (average LTI 4.2±11.9°). Additionally, 73% of patients had an LPI >13.5° classifying them as having excessive patellar tilt (average LPI 19.6±9.4°). No significant correlations were found between PTI and LTI (r = -0.0.13, p = 0.914), PTI and LPI (r = -0.092, p = 0.47), CD ratio and LTI (r = 0.19, p=0.14) or CD ratio and LPI (r = 0.162, p = 0.1959).
Conclusion:
The PTI was found to be a more reproducible and reliable measure of patellar height than the radiographic measured CD ratio. Additionally, PTI categorized a much smaller proportion of patients as having patella alta versus the CD ratio. No significant correlation was found between PTI or the CD ratio and measures of patellar tilt and trochlear dysplasia despite the majority of patients in this study having both. These results suggest that renewed investigation into the role of patellar height in PI is warranted and further studies should be conducted to determine what is clinically significant patella alta. These results also support prior work suggesting that the PTI is a more clinically relevant measure of PH. The weak correlation between PTI and the CD ratio is in line with prior work and if CD and other ratios are to continue to be used in clinical practice, revisiting normative values is warranted.
Patellotrochlear Index (PTI). The sagittal image with the longest section of the patella and the thickest patellar cartilage is selected. The length of the patellar chondral surface is measured – here it is 32.6mm. Then the length of the trochlea overlapping with the patellar chondral surface is measured in parallel by referencing a a perpendicular subtended from the patellar chondral surface measurement – here the length of overlapping trochlear chondral surface is 12.1mm. The PTI is calculated with dividing the trochlear chondral length by the patellar chondral length and converting to a percentage: (12.1 / 32.6) x 100 = 37%