Measures of Adverse Patellofemoral Morphology Predict Risk of Local Cartilage Damage Progression: A Yale/MOST Collaborative Study

Abstract

Background:

The relationship between anatomic risk factors for patellofemoral morphology and patellofemoral (PF) cartilage damage in the general population remains unclear. This study aimed to determine whether 3-D based patellofemoral metrics of morphology are associated with progressive lateral PF cartilage damage.

Study Design:

Cross-sectional Study. Level of Evidence 2.

Methods:

We analyzed non-weight-bearing computed tomography scans of knees from a subset of participants enrolled in the community-based Multicenter Osteoarthritis Study (MOST). Baseline and two-year follow-up knee magnetic resonance images were scored for progressive PF cartilage damage using the magnetic resonance imaging osteoarthritis knee score (MOAKS). Tibial tubercle–trochlear groove distance (TT-TG), patellar tilt angle (PTA), external tibiofemoral rotation (eTFR), patellar height, entry point to trochlear groove angle, and entry point to transition point angle (EP-TP) were measured for each knee. To assess the association of each morphology measure with progressive cartilage damage, generalized estimating equation logistic regression models were fit using continuous and natural cubic spline models.

Results

We analyzed lateral patellofemoral cartilage damage in 389 participant knees, mean age 53.79 ± 5.51 years, body mass index 28.48 ± 5.13 kg/m²). TT-TG (β = 0.23, OR = 1.26, p = 0.036), eTFR, (β = 0.24, OR = 1.27, P = 0.048), and EP-TP spline model (Z= 2.09, P = 0.036) all demonstrated significant positive associations with worsening lateral PF cartilage damage.

Conclusion

The results demonstrate significant associations between 3D anatomical metrics and progressive lateral patellofemoral cartilage damage. Elevated tibial tubercle-trochlear groove distance, external tibiofemoral rotation, and entry point-transition point angles may be keys to understanding the mechanical etiology of lateral patellofemoral arthritis.

Introduction

Osteoarthritis (OA) is a common condition, with a global prevalence of 595 million in 2020.⁴⁰ Factors besides trauma contributing to OA development include, among others, variation in joint morphology and disruption in cartilage and bone metabolism.¹⁴ The patellofemoral (PF) compartment of the knee is often affected by OA and is a major source of knee pain and loss of quality adjusted life years (QALY)⁷.

Treatment of PF pain and of PF OA can be challenging, as root causes are often ambiguous. For example, it is important to differentiate between activity-based overload and morphological causes such as malalignment. While the first one can be mostly resolved by short-term activity reduction, the second one might need corrective therapy such as bracing⁵ or surgery. Abnormal alignment can lead to repeated microtrauma, inflammation, and eventually cartilage loss which can progress to OA. Unequal distribution of forces across the joint surface, due to patellar malalignment can lead to degenerative changes to the underlying articular surfaces.^19,24,31 Kalichman et al. found that abnormal ranges of patellar alignment indices were correlated with patellofemoral OA.¹⁷

Many measures describing structural factors of the patellofemoral joint stem from the treatment of patellofemoral instability. Among those factors, trochlear dysplasia and different alignment factors, such as tibial tuberosity to trochlea groove (TT-TG) distance stand out, as their correlation to OA has been reported^{6,12,13,18,24,26,32}. Trochlear dysplasia, characterized by a shallow or misshapen trochlear groove, can disrupt proper patellar tracking, predisposing individuals to patellar maltracking.^4,21 The maltracking might manifest as increased joint loads and contribute to OA progression, which often can be seen in patients with patellar instability²⁵. Yu et al. and Beitler et al. recently described the entry-point-transition point angle (EP-TP) and entry-point to trochlear groove (EP-TG) angle, as a mean to quantify trochlear groove dysplasia.⁴⁶ EP-TG focuses on the laterality of the patellar entry, while EP-TP focuses on the obliquity of the path. Both are currently being used to predict patellar instability.^3,35

A popular metric for PF surgical decision-making is the TT-TG distance, which quantifies the lateral pull of the patellar tendon.^2,34,43 Elevated TT-TG is linked to lateral maltracking of the patella.¹¹ Other similar measurements affecting the PF joint are patellar tilt angle (PTA), patellar height, sagittal TT-TG (sTT-TG), and external tibiofemoral rotation (eTFR). These measures in addition to TT-TG, EP-TG, and EP-TP were used in this study and explained in more detail in Supplement 1 along with background references. ^{3,6,8,9,13,18,20,23,28,30,32,35–39,41,44–46}

Cartilage damage is the signature pathologic feature of OA. This study was designed to evaluate the extent to which measures of lateral PF dysplasia are associated with increased risk for progressive PF cartilage damage and, if so, to define a threshold for risk elevation. Quantification of this relationship could provide potential targets for conservative treatment or corrective surgery to enhance long-term knee joint health.

The purpose of this study was to determine the relationship between certain morphologic measurements and misalignment of the PF joint and progressive lateral PF cartilage damage. We hypothesize that measures quantifying more severe metrics are linked to a higher risk of progressive lateral patellofemoral cartilage damage.

Methods

Multicenter Osteoarthritis Study

The Multicenter Osteoarthritis Study (MOST) is a research initiative funded by the National Institute on Aging of the National Institutes of Health (NIH) that focuses on individuals aged 45 years and older who have or are at high risk of developing knee OA.³³ Our study’s data collection was initiated at the year 12 examination of MOST at which 1500 new participants were recruited. Included were persons from the community who had mild knee OA or no OA. Those with radiographs showing Kellgren and Lawrence grade 3 or 4 knee OA in any compartment of either knee were excluded as were those with severe knee pain. At the study’s baseline examination, non-weight bearing CT and MRI scans of the knees were obtained for all included subjects. As part of the two-year follow-up examination repeat knee MRI’s were obtained. All data used in this study was collected between 2016–2018. This study was approved by institutional review boards at Boston University, University of California, San Francisco and the two clinical sites, The University of Iowa and the University of Alabama at Birmingham.

Imaging Parameters

CT imaging was conducted at two separate locations using dual-energy CT technology. Specifically, University of Iowa employed a Siemens SOMATOM Force scanner with certain imaging parameters (80/150 kVp, 250 mAs, 0.8 mm pitch, tin filtration at 150 kVp, rotation speed 15 ms), while University of Alabama at Birmingham utilized a GE Discovery CT750HD scanner with different parameters (80/140 kVp, 260 mAs, 0.9 mm pitch, 0.8 s exposure, rotation speed 50 ms). The raw projection data underwent reconstruction according to the parameters outlined by Jarraya et al.¹⁶ The images analyzed in this study were based on the 80 kVp reconstructions due to their superior bone-to-soft-tissue contrast.

Knee MRIs were acquired using an extremity based 1.5 Tesla magnet. Each selected knee was read at both timepoints by one of 2 experienced musculoskeletal radiologists (AG and FR) according to the MRI Osteoarthritis Knee Score (MOAKS) for cartilage damage in each knee’s subregions.¹⁵ Additionally, Western Ontario and McMaster Universities Arthritis Index (WOMAC), hip knee alignment and PF Kellgren & Lawrence Grade was acquired ^22,33 to describe the cohort, but not used for analysis. Interreader reliability for cartilage damage scores was high.²²

Participant Selection

To study the relationship between PF dysplasia and cartilage damage, we limited the sample with CT scans to those aged 45 to 60 years at the baseline exam (see Figure 2; Year 12 of the overall MOST study), a total of 446 knees. Of those in the targeted age range, we selected all persons whose two knee MRI showed progressive PF cartilage damage over the 2-year follow-up in either medial or lateral PF compartment. We then selected an equal number of persons from the MOST cohort who had no progressive cartilage damage in the PF compartment over these 2 years as controls. For analyses, we took cases that had progressive lateral PF cartilage damage vs. those with no damage. For each knee to be studied, we had a CT scan from the baseline exam to assess measures of PF morphology.

Figure 2:

Flowchart of cohort inclusion and exclusion criteria and respective cohort sizes.

Measures of PF Morphology

All selected CT scans were segmented in Synopsys Simpleware (Sunnyvale, CA) using the CT knee auto segmentation algorithm (i.e. AS Ortho module). A total of 21 landmarks were placed on the bony anatomy in a semi-automated process previously described by Park et al.³⁰ Once landmarks were verified with manual review, landmarks were used to calculate 3D metrics. Using 3D landmarks positioned on each 3D model, several key parameters were measured: the TT-TG distance, sagittal TT-TG distance (sTTTG), patellar tilt angle (PTA), relative external tibiofemoral rotation (reTFR), and patellar height. EPTP and EPTG were measured on anterior views of the proximal trochlea. A detailed description of this process is given in Supplement 1. Members of the team who were segmenting 3D models and placing landmarks were blinded to the clinical history and MRI outcomes of all MOST participants.

The measures of PF morphology can be separated into two categories: (1) alignment measures (i.e, TT-TG, reTFR sTTTG, PTA, patellar height) and femoral trochlea dysplasia measures (EPTP and EPTG).

TT-TG quantifies the lateral displacement of the tibial tubercle relative to the trochlear groove on the coronal plane. sTTTG evaluates the relative anteroposterior positioning of the tibial tubercle concerning the trochlear groove in the sagittal plane. PTA measures the angular orientation of the patella in relation to the posterior femoral condyles. reTFR assesses the rotational alignment between the tibia and femur. Patellar height describes the vertical position of the patella relative to the trochlea. EPTG angle describes the angle between the entry point (EP), or the midpoint of the medial and lateral ridges of the proximal trochlea, and a line down the apex of the distal trochlear groove (TG) that is perpendicular to medial and lateral condyles. Similarly, EPTP measures the angle between the EP and the transition point (TP), the point at which the trochlear groove changes from an oblique trajectory to vertical as it enters distally. Detailed description, depictions, and background information for each metric is given in Supplemental 1 Section “Description of Metrics”.

Statistical Methods

The outcome measure for our analyses was progressive lateral PF cartilage damage developing or worsening over 2 years. We defined this as an increase of at least a half grade in MOAKS score in a lateral PF subregion.²² There were 2 such subregions, one encompassing the lateral patella and the other the lateral trochlea. Analyses were subregion specific with generalized estimating equations (GEE) used to adjust for the correlation of findings in the 2 subregions, allowing us to present the risk of lateral PF cartilage damage by knee. The relation of each measure of PF dysplasia could be linear or nonlinear (for instance if there were a threshold above which the risk of damage increased) or the risk of damage could increase in a stepwise fashion. We tested continuous linear, quintile, and natural cubic spline versions of the standardized dysplasia measure, the latter to address possible non-linearity. (see Supplement 1 “Selection of Statistical Model”).²⁹ All models were adjusted for age, sex and BMI. For each measure of dysplasia, we attempted to determine the model (e.g. linear vs. nonlinear) that provided the best fit to the data based on the smallest QICu value. Once the best model was selected, we proceeded to use the result from this model to assess the relationship of the measure of dysplasia with cartilage damage. For example, if the cubic spline model provided the best fit for a specific measure of dysplasia, we report only the result of the cubic spline model. A Pearson correlation matrix was created to understand correlation between different measures of dysplasia (for more details see Supplement 1 Table 3)

Results

Of 1942 eligible knees in the MOST database, 446 were selected based on subject age and 2-year PF cartilage damage on MRI (half with and half without). Measures of PF morphology were acquired for 445 knees and 1 knee was excluded due to part of the knee being outside the scanning window. A further 56 knees were excluded due to missing MOAKS scores, leaving 389 knees with MOAKS data for both subregions of interest (i.e., lateral trochlea and lateral patella). All 389 knees had complete covariate data and were included in the final analysis. Demographics are outlined in Table 1. BMI ranged from 18.9 to 44.0 kg/m².

Table 1. Demographic characteristics of the study population and descriptive statistics for 3D patellofemoral measurements.

Continuous variables are presented as mean ± standard deviation (SD) with range, while categorical variables are presented as counts and percentages. Metrics include tibial tubercle–trochlear groove distance (TTTG), relative external tibiofemoral rotation (reTFR), patellar tilt angle (PTA), entry point to trochlear groove angle (EPTG), sagittal TTTG (sTTTG), patellar height, and entry point to transition point angle (EPTP). These measurements quantify anatomical variations relevant to PF dysplasia and cartilage damage. BMI – body mass index.

Variable	Total (N = 389)
Age, years	53.79 ± 5.51**
BMI, kg/m2	28.5 ± 5.1
Female	206 (53%)
Patellofemoral Kellgren & Lawrence Grade
0	248 (64%)
1	115 (30%)
1.9#	9 (2%)
2	13 (3%)
3	2 (1%)
4	1 (0%)
WOMAC Knee Pain Status
No Pain	87 (43%)
Pain	114 (57%)
Hip Knee Alignment*
Valgus	119 (31%)
Neutral (defined as −1,0, or +1 HKA)	2 (0%)
Varus	265 (69%)
Baseline MOAKS Cartilage Score– lateral PF Joint (range for subregions 0–6)
Both 0	217 (56%)
Trochlea 0, patella Non-Zero	108 (28%)
Trochlea Non-Zero, Patella 0	25 (6%)
Trochlea Non-Zero, Patella Non-Zero	39 (10%)
TTTG, mm	12.80 ± 3.70
reTFR, mm	1.75 ± 4.89
PTA, °	10.63 ± 11.38
EPTG, °	19.06 ± 9.45
STTTG, mm	1.97 ± 4.36
Patellar Height, dimension less	0.80 ± 0.15
EPTP, °	27.17 ± 23.70

^*Varus defined when HKA was negative; Valgus when HKA was positive. Neutral was defined if limb had an HKA score of 0 (neither varus nor valgus).

^**Study design was for subject ages 45–60 but inadvertently, 15/389 persons over age 60 were also studied.

^#KL 1.9 is a knee with osteophytes but no joint space narrowing on x-ray.

58 trochlea and 16 patella experienced progression of lateral PF cartilage damage. Of these, 4 participants experienced progression in both the femur and patella. The overall values of measures of dysplasia are shown in Table 1. The correlation matrix showed that EPTP and EPTG had a high correlation of r = 0.78; all other correlations were either negligible or low (see Supplement 1 Table 3).

The linear model was selected as providing the best fit for the following measures of dysplasia: TT-TG, reTFR, Ep-TG, and PTA. Natural cubic splines models showed the best fit for the following measures of dysplasia: patellar height, sTT-TG, and EP-TG. The quintile model was not selected for any of the measures. Details for model selection can be found in Supplement 1 Table 4.

TT-TG, reTFR, and EP-TP were significantly related to progressive PF cartilage damage in their respective models. EP-TG and PTA showed a trend towards significance. The estimated effect (i.e., the increase in cartilage damage progression risk per standardized unit increase), standard error, estimate confidence interval, Z-statistic, and p-value of the linear models are shown in (Table 2). ReTFR demonstrated the highest standardized increase in risk of 27.1% per 4.89 degrees followed by TT-TG with 25.9% per 3.70 mm.

Table 2. The association of standardized measures of patellar morphology with patellofemoral cartilage damage progression over 2 years.

Estimates showed an increase in risk per standard deviation increase. (e.g., Risk of PF OA progression increased by 23% for every 3.70 mm TT-TG.)

Parameter	Risk of progression per s.d.	95% Confidence Limits	Z-statistic	P-value
TT-TG	0.230	[0.0146, 0.4451]	2.09	0.036
reTFR	0.2399	[0.0017, 0.4781]	1.97	0.048
EP-TG	0.2712	[−0.7365, 0.2457]	1.89	0.059
PTA	0.2472	[−0.0056, 0.4999]	1.92	0.056

The splines of the EP-TP, sTT-TG, and patellar height are displayed in Figure 2 with only the third spline term of EP-TP indicating significance (Z-statistic= 2.09; p=0.036). Patellar height spline term 3 tended towards significance (p=0.078). All spline terms can be found in Supplement 1 Table 3. The predicted probability of cartilage damage progression for EP-TP angles over 50 degrees was 11% and increased to 24% for an EP-TP over 70 degrees.

Discussion

This study evaluated the relationship between patellofemoral morphology metrics and the progression of lateral PF cartilage damage using data from the MOST cohort. We found that metrics were related to cartilage damage progression in a population without frank patellar instability. Significant associations were observed for tibial tubercle–trochlear groove (TT-TG) distance, external tibiofemoral rotation (eTFR), and entry point–transition point (EPTP) angle.

Our finding that higher TT-TG values increase the risk of progressive cartilage damage aligns with previous studies identifying TT-TG as contributing to PF osteoarthritis (OA). Previous studies were cross sectional ^26,42,47 and, as such, left open the possibility that OA in the PF joint caused the extreme values of dysplasia and not vice versa. Our longitudinal study showed convincingly that specific anatomical patterns are likely to accelerate progressive cartilage loss, worsening OA.

Similarly, the significant association for eTFR indicated that rotational malalignment was associated with progressive lateral PF cartilage damage.. Lateral tracking is influenced by a multitude of factors such as the lateralization of the patellar tendon force vector and a medialization of the trochlea groove. In this study we have shown that elevated TT-TG, which quantifies the lateralization of this vector and is routinely measured from CT or MRI in PF instability patients, is a risk factor for progressive lateral PF cartilage damage. Notably, the EPTP angle spline model also showed a significant positive relationship, supporting the hypothesis that patellar tracking abnormalities lead to cartilage damage. Overall, our results show that lateral patellar tracking and malalignment are correlated with progressive PF cartilage damage and that 3D metrics of PF morphology can be used to quantify risk.

Although the EPTG and patellar tilt angle (PTA) linear models did not show statistically significant associations with cartilage damage, both demonstrated trends toward significance, suggesting their potential relevance in larger or more targeted cohorts. Increased patellar tilt is among factors correlated with tight lateral structures pulling the patella laterally, thus contributing to lateral tracking and potentially increased loads on the lateral facet⁴⁸. Other metrics, including patellar height, and sagittal TT-TG distance (sTTTG), did not demonstrate significant associations with progression of lateral PF cartilage damage.

Many of the PF morphology measures presented in this study are often measured as inputs to clinical decision making for the treatment of PF instability, especially in cases where surgery might be indicated by clinical history. Affected patients are predominantly young, female, and show either no or only early signs of OA, and their PF metrics tend to be higher and more severe¹⁰. By contrast, this study has been conducted on an older, sex-balanced group, with similar OA status and milder metrics. Still, knowing the impact of these measures on the older cohort could help physicians estimate the risk of future cartilage damage in patellar instability patients based on specific metric aberrations. This could help in choosing the correct treatment to not only stabilize a PF joint joint but also improve the likelihood of pain-free knee function in the long-term, with possibility of reduced arthritis risk long term. For example, a anteromedializing tibial tubercle transfer reduces TT-TG and PTA, normalizing patellar tracking and reducing lateral PF overload¹. While surgery might be the most effective means to reduce effects of severe dysplasia, the findings of this study could also lead to more targeted conservative management of patients with OA risk. For example, Mousavi et. al. have recently shown that in addition to improving hip external rotation strength, specialized treatment of increased external tibiofemoral rotation is more effective than routine physiotherapy in treating malalignment and the associated pain and movement impairments²⁷.

Our measures of dysplasia were acquired from 3D models created from CT-scans. This process allowed us to standardize the process and increase its reliability (see supplement 1). It mitigated known issues with measuring in 2D³⁸. Additionally, it enabled us to incorporate new continuous metrics of trochlea dysplasia, such as EPTP⁴⁵ and EPTG³. Previously, other studies have examined the link between some of these measures and PF cartilage loss. Still, to our knowledge, this study is the first to employ a standardized 3D based approach, enabling the investigation of a multiple measures in a reliable and reproducible fashion.

Limitations

Strengths of the study include its robust sample size and use of advanced modeling techniques to characterize complex relationships. We did not carry out analyses of pain data. Most participants had little or no knee pain. Furthermore, the relevance of findings in an older population recruited from communities to a younger one with patellofemoral instability remains unclear. The study cohort was predominantly derived from two sites, with youngest study subjects aged 45 years old and included those with little if any radiographic OA. We believe that our results can be generalized to other populations of knees, but more studies should be conducted to better understand our measures’ reliability in younger patients with dysplastic knees. Another limitation of our study is that all measurements were performed on supine CT scans, which may not represent the kinematics of a weightbearing joint. Initial attempts to include weightbearing scans were limited by poor scan resolution, making accurate measurements challenging.

Conclusion

The results demonstrate significant associations between abnormal 3D anatomical metrics and lateral patellofemoral cartilage damage progression. Elevated tibial tubercle-trochlear groove (TT-TG) distance, tibiofemoral rotation, and entry point-transition point (EP-TP) angles may be keys to understanding the mechanical etiology of lateral patellofemoral cartilage damage and progression to patellofemoral osteoarthritis.

Figure 1:

Entry Point to Transition Point (EP-TP) and Entry Point to Trochlear Groove (EP-TG) are measurements based on our consistent observations that patellar tracking normally follows the shape of the underlying trochlear groove. These metrics describe how oblique the initial movement of the patella (from a fully extended knee) can be in a patellofemoral patient as the patella moves from lateral to medial as well as distally before being captured in the deepening trochlear groove. Please see Supplement 1 for more information,

Figure 3. Cubic spline models illustrating the relationships between anatomical variables and lateral patellofemoral cartilage damage progression.

The splines depict non-linear associations, capturing potential threshold effects and variations in risk across the range of measurements. Shaded regions represent confidence intervals. Blue dots correspond to participants who did not experience lateral progression while orange asterisks correspond to participants who experienced damage progression.

Clinical Relevance:

The findings of this study can help inform physicians of the prognosis of patients presenting with patellofemoral disease.

Key Terms:

Patellofemoral disease, patellofemoral morphology, cartilage prognosis, osteoarthritis

What is known about the subject:

Previous studies have identified links between some of the measures discussed in this paper and patellofemoral osteoarthritis.

What this study adds to existing knowledge:

To our knowledge, this study is the first to use longitudinal data from a high-quality community-based knee study to examine how 3-D patellofemoral structural measurements affect cartilage damage. This may allow us to generate quantitative estimates of how specific morphologies and malalignment impact prognosis.