Introduction

Knee osteoarthritis (OA) has a global prevalence of 3.8% [25] and is a major source of socioeconomic costs [18]. It has an increasing burden (years lived with disorder) of 64% over the last 2 decades [25] and is the most common cause of knee arthroplasty [2, 3].

In the past, an indirect approximation of articular cartilage measuring the minimum joint space width on radiographs represents the standard to detect progression of knee OA [1, 12]. Measuring minimum joint space width on radiographs is an indirect approach and is subject to error secondary to reproducibility of position of the joint, beam alignment, and distance between the joint and film [23]. The current gold standard to detect progression of knee OA is magnetic resonance imaging (MRI) [12, 40, 43]. MRI provides excellent soft tissue imaging and is accurate in measuring cartilage volume, thickness, and surface area [11, 14, 41, 42].

The Osteoarthritis Initiative (OAI) is a public free of charge database used to conduct longitudinal studies using prospective MRI data to analyze cartilage alterations in knee OA. A number of studies [13, 14, 26, 28, 42, 43] reported on changes in cartilage volume and thickness in ten tibial and six femoral subregions. Maschek et al. [26] showed a minimum loss of cartilage thickness of 136 µm in the most affected subregion (OV1) in knees with Kellgren and Lawrence grade (KLG) 2–4. Wluka et al. [45] defined OA progression as annual change in cartilage volume of 5.3%/year.

A number of factors are associated with progression of OA (age, previous knee injury, obesity) [18, 37]. Inter alia (BMI, physical activity, meniscal pathologies, cartilage or bone marrow lesions) [4, 24], limb alignment represents a risk factor for progression of OA of the knee [17, 20, 34,35,36]. Sharma et al. [36] showed a fourfold higher risk of OA progression with varus alignment (> 5°) and a odds ratio of 4.89 for progression of OA in valgus alignment (> 5°).

Although the quantitative loss of cartilage thickness and volume has been described in previous studies, nobody has studied yet, to our knowledge, the survival rate (Kaplan–Meier survival rate) over the time and can show a detailed timeline of cartilage loss in subregions of the knee.

Based on the OAI dataset, the study wants to describe a detailed timeline of cartilage survival. Progression of OA was analyzed to answer the following research questions: (1) Does mechanical alignment of the lower limb affect progression of OA (timeline of cartilage survival) in the different subregions of the knee? (2) Are there differences in the change of cartilage thickness and cartilage volume between the subregions of the knee?

Materials and methods

Cohort

The study investigates the Osteoarthritis Initiative (OAI) database, a multi-center, longitudinal, prospective observational study of knee OA [30]. A total of 4791 subjects aged between 45 and 79 years at risk of developing knee OA were enrolled in the OAI. Annual radiographs, MRI, and clinical assessment of knees and disease activity were performed for all participants over a period of 8 years.

The “Core Image Assessment sample” at OAI was designed to provide longitudinal structural outcomes. 600 knees, part of the “Core Image Assessment sample”, were defined as “Index knees”: symptomatic knees with Kellgren and Lawrence (K&L) grades 2 or 3 radiographic evidence of OA at baseline [30]. Eckstein et al. [13, 42] performed quantitative cartilage measurements on MRI (sagittal DESS sequence) of these knees at baseline, 12, 24 and 48 months (kMRI quant Cart sagDESS, Project 9A) [31]. Of the 600 index knees, the hip-knee-ankle standing radiographs were assessed to determine limb alignment (available at a single measurement at 12 months by Cooke et al. [30] (“flXR Knee Alignment Cooke”), ICC 0.99). 142 knees were excluded because no alignment measurements were available leaving 458 knees to be included in the current study. Subjects were grouped based on lower limb alignment into varus alignment (n = 158, hip-knee-ankle angle < 177°), physiologic alignment (n = 234, hip-knee-ankle angle 180° ± 3°) and valgus alignment (n = 66, hip-knee-ankle angle > 183°). 458 knees were available for analysis at baseline, 451 knees at 12 months, 458 at 24 months and 56 at 48 months.

Image analyses

The cartilage volume and cartilage thickness were analyzed on sagittal double-echo steady-state (DESS) sequences. Reporting details and nomenclature for MRI evaluation are reported on the OAI website [30, 31] and by Eckstein et al. [10]. The medial and lateral femorotibial cartilage was segmented manually, using proprietary software (Chondrometrics GmbH), with two readers blinded to disease severity, alignment, and the order of image acquisition [44]. One expert reader, with more than 5 years of MRI-based cartilage segmentation experience, performed quality control readings of all segmentations.

The following cartilage volume parameters were analyzed (Fig. 1): central medial femur (total, OAI variable “BMFVCL”, in mm3 and normalized, “BMFVCN”, in mm), central lateral femur (total, “BLFVCL”, in mm3 and normalized, “BLFVCN”, in mm), central medial tibia (total, “WMTVCL”, in mm3 and normalized, “WMTVCN”, in mm) and central lateral tibia (total, “WLTVCL”, in mm3 and normalized, “WLTVCN”, in mm). The following thickness parameters (in mm) as femoral subregions were analyzed: central area of the weight-bearing lateral femur (“BLFMTH”), central area of the weight-bearing medial femur (“CBMFMTH”), internal part of the weight-bearing medial femur (“IBMFMTH”), external part of the weight-bearing medial femur (“EBMFMTH”). Following thickness parameters (in mm) as tibial subregions were analyzed: internal part of the weight-bearing medial (“IMTMTH”) and lateral (“ILTMTH”) tibia, central area of the medial (“CMTMTH”) and lateral (“CLTMTH”) tibia, external area of the medial (“EMTMTH”) and lateral (“ELTMTH”) tibia, anterior part of the medial (“AMTMTH”) and lateral (“ALTMTH”) tibia as well as posterior area of the medial (“PMTMTH”) and lateral (“PLTMTH”) tibia.

Fig. 1
figure 1

The analyzed subregions at the tibia (left) and the femur (right). Central medial tibia (cMT) includes variables “WMTVCL”, “WMTVCN” and “CMTMTH”, external medial tibia (eMT) includes “EMTMTH”, internal medial tibia (iMT) includes “IMTMTH”, anterior medial tibia (aMT) includes “AMTMTH”, posterior medial tibia (pMT) includes “PMTMTH”. At the lateral tibia following subregions with variables are described: central lateral tibia (cLT) with variables “WLTVCL”, “WLTVCN” and “CLTMTH”; external lateral tibia (eLT) with “ELTMTH”; internal lateral tibia (iLT) with “ILTMTH”; anterior lateral tibia (aLT) with “ALTMTH” and posterior lateral tibia (pLT) with “PLTMTH”. At the femur subregion central medial femur (ccMF) includes variables “BMFVCL”, “BMFVCN” and “CBMFMTH”; external central medial femur (ecMF) includes “EBMFMTH”, internal central medial femur (icMF) includes “IBMFMTH”; on the lateral femur subregion central lateral femur (ccLF) with “BLFVCL”, “BLFVCN” and “BLFMTH” is analyzed

Full size image

Limb alignment (varus, valgus, physiologic) was identified on hip-knee-ankle radiographs (parameter: flXR_KneeAlign_Cooke01). Frontal alignment was defined as the angle between the mechanical axes of the femur and tibia [6, 7].

In the current study, progression of knee OA per year was defined by a minimum thickness loss of 136 µm [26] or/with volume loss of minimum 5% [45].

The study received IRB approval by the institutional review board at the authors’ institution (Hospital for Special Surgery, NY, USA, IRB #15013).

Statistical analyses

Demographics are shown in Table 1. Possible confounders (age, BMI and previous surgery) are analyzed using Student’s t test (BMI, age) and Pearson’s Chi square test (previous surgery) (Table 2).

Table 1 Demographics of 600 knees summarized at the “Core Image Assessment sample” (symptomatic knees with Kellgren and Lawrence grades 2 or 3 radiographic evidence of OA)
Full size table
Table 2 Possible confounding factors such as BMI, age and previous knee surgery do not show any statistical significant differences between knees reaching the endpoint of cartilage loss and knees with cartilage survival
Full size table

Kaplan–Meier survival curves were generated for cartilage survival. A loss of more than 136 µm or 5% of volume was counted as progression of OA (“event” in Kaplan–Meier survival analysis). Cases without an event were censored. For every subregion (volume: 2 femoral, 2 tibial; thickness: 4 femoral, 10 tibial), three survival curves (alignment dependent: varus, valgus and physiological) were prepared. Significant differences in survival time between each subregion and/or alignment type were determined by log-rank tests. Effect size was calculated according to Hedge’s g. A p value of less than 0.05 was considered to be statistically significant. All statistical analyses were performed using IBM SPSS Statistics software version 23 (Armonk, NY: IBM Corp.).

Results

Comparison between alignments showed statistical significant differences (Tables 3, 4), effect sizes showed a large effect in every comparison (Hedge’s g 2.13–9.03), except at the lateral femoral condyle (varus vs physiologic alignment Hedge’s g 0.15).

Table 3 Survival rates (in months) of cartilage volume depending on limb alignment (varus, valgus and physiologic)
Full size table
Table 4 Survival rates (Kaplan–Meier curve) of cartilage thickness in months depending on limb alignment
Full size table

Cartilage thickness survival rates and subregion differences in varus knees are demonstrated in Table 5. Table 6 summarizes the cartilage thickness survival rates and each difference between subregions in knees with valgus alignment. Physiologic aligned osteoarthritic knees showed differences in cartilage thickness survival rates as well as in their subregion analyses (Table 7).

Table 5 “OA-progression-timeline” in varus aligned, osteoarthritic knees with mean cartilage thickness survival rates (in months) are demonstrated
Full size table
Table 6 “OA-progression-timeline” in valgus aligned, osteoarthritic knees with mean cartilage thickness survival rates (in months) are demonstrated
Full size table
Table 7 “OA-progression-timeline” in physiological aligned, osteoarthritic knees with mean cartilage thickness survival rates (in months) are demonstrated
Full size table

Varus knees showed shortest cartilage volume survival rates at the medial femur (30.8 months, CI 95% 28.2–33.4), followed by the medial tibia (36.9 months, CI 95% 34.5–39.4) (Table 3). Valgus knees showed shortest cartilage volume survival at the lateral tibia (31.5 months, CI 95% 27.6–35.4) and at the lateral femur (36.2 months, CI 95% 32.4–40) (Table 7). Physiologic aligned osteoarthritic knees showed shortest volume survival rates at the medial femur (37.8 months, CI 95% 36–39.7) (Table 8).

Table 8 Mean volume cartilage survival rates (in months) depending on limb alignment. Statistically significant differences between medial and lateral femur and medial and lateral tibia are shown
Full size table

Discussion

The main finding of the current study, as one of the first, is the detection of a cartilage survival rate over time (Kaplan–Meier) and shows a strong relation to malalignment. Varus and valgus alignment accelerate the OA progression rate compared to osteoarthritic knees with physiologic alignment. Differences of cartilage survival in subregions contribute to more precisely understanding of an “OA-progression-timeline” and create a “time-map” of cartilage survival.

The study reveals that the cartilage survival is strongly influenced by limb alignment similar to prior reports in the literature [27, 28, 34, 36]. The current study suggests that in knees with physiologic alignment, a more homogeneous progression of OA is observed compared to the varus or valgus group. In contrast, varus knees show a broader time range of the cartilage loss in the anterior and posterior subregion of the medial compartment suggesting that little posterior translation of the femur occurs. In valgus knees, however, posterior translation characterizes the progression of OA. These findings might be explained by differences in normal kinematics of the knee with a relatively constant medial condyle and posterior translation of the lateral condyle contact point during a knee bending [8, 9, 38].

In relation to cartilage thickness in varus and valgus alignment, a statistical significant difference of survival rates can be shown in the anterior and posterior subregion of the lateral compartment (lateral tibia anterior: p = 0.014; lateral tibia posterior: p = 0.002). This suggests that in valgus knees an anterior to posterior translation characterizes the progression of OA, whereas in varus knees direction of progression is in medial direction (medial tibia central: p = 0.001 medial external tibia: p = 0.001).

A closer look into differences between subregions in each alignment type shows in varus knees in the central medial tibia and central medial femur the shortest survival rates. Not surprisingly, external subregions on the medial tibia and femur are affected early too and the shortest survival rate of the contralateral compartment was determined in the internal lateral tibia. This might be related to beginning medial subluxation [21, 22].

The last affected regions were anterior and posterior. These findings lead to following assumptions: (1) prior studies reported that insufficiency of the ACL is a main factor for posterior medial wear in varus OA knees [19, 29]. Therefore, degeneration of the ACL may not be present in the majority of subjects in the OAI cohort. (2) Anteromedial wear pattern in varus OA knees are reported in anatomical studies analyzing bone wafers collected from total knee arthroplasties [19, 32, 39]. However, these findings cannot be confirmed with OAI–MRI measurements. Rajgopal et al. [32] described an anteromedial pattern of varus OA knees, although 87% showed an intact ACL. In 99.5%, anteromedial wear was present. White et al. [39] described ACL sufficient knees with anteromedial wear pattern in 40% (zone B) and in 43% with central medial wear pattern (zone BC).

These differences to the current study might be an expression of different definitions of medial tibial subregions. Rajgopal et al. [32] described three medial tibial subregions, White et al. [39] defined four medial tibial subregions, from anterior to posterior, respectively. Exact cutoff values were not reported. On the other hand, Raju et al. [33] showed in a cadaver study a more central medial cartilage wear. Biswal et al. [5] showed in a longitudinal MRI study that cartilage lesions located in the central region of the medial compartment progress faster than cartilage lesions in the anterior and posterior portions of the medial compartment. Everhart et al. [16] described a more rapid progress of cartilage defects in the medial compartment than in the lateral compartment.

In valgus OA knees, lateral central and internal tibial cartilage thickness shows the shortest survival rates. Lateral femoral cartilage thickness shows a significant cartilage loss relatively late. The shortest survival time at the tibial subregions, followed by femoral subregions is an important differentiation to varus aligned knees. Eckstein et al. [15] also showed the highest amount of cartilage loss at the internal and central lateral tibia followed by lateral femoral subregions in valgus knees. In valgus knees, the contralateral medial compartment shows a mean thickness cartilage survival rate of 43.7 months (medial tibial subregions), whereas in varus knee the contralateral compartment shows a mean thickness cartilage survival rate of 41.4 months (lateral tibial subregions). These findings suggest that the valgus osteoarthritic remains longer a unicompartmental disease than knees with varus OA.

Physiologic aligned osteoarthritic knees show a very homogeneous pattern of cartilage thickness loss. Central femoral cartilage thickness shows shortest cartilage survival rate followed by the central lateral tibial. The longest survival rate was detected in the posterior medial tibial subregion (46.2 months). A comparable pattern to varus aligned knees (central medial femur, central medial tibia and external medial tibia) or valgus aligned knees (central lateral tibia, internal lateral tibia and external lateral tibia) could not been shown.

The current study has several limitations. First, the cutoff values (cartilage loss of 136 µm thickness or 5% volume over 1 year) are described in recent studies, but are selected randomly in the current study. Second, these cutoff values are signs of cartilage-loss’ only, they do not predict progression of clinical symptoms as increased pain, decreased range of motion or decreased quality of life. The cutoff values are no disease value itself.

Conclusion

Data of the current study suggest that in varus OA, the main cartilage loss occurs on the medial femur (central and external), medial tibia (central and external) and internal lateral tibia. In valgus knees, the lateral tibia (central subregion first, followed by internal and external subregion) with a posterior component are affected first. This study is the first to determine the rate of OA progression depending on lower limb alignment. Therefore, we can now inform our patients in detail about when a significant loss of cartilage is to be expected and in which subregion this will occur. The therapy regimen may need to be adapted after reaching the cutoff values. OA progression occured after an average of 37.6 months in physiologic aligned knees, 31.2 months in varus knees and 30 months in valgus knees. This confirms a considerable influence of limb alignment on cartilage survival.