Article Text
Abstract
Background Tibiofemoral alignment has a role in knee osteoarthritis (OA), but which factors contribute to alignment is unknown.
Objective To investigate familial aggregation of tibiofemoral alignment in participants of the GARP (Genetics ARthrosis and Progression) study.
Methods The tibiofemoral anatomical angle on semiflexed knee radiographs was measured in sibling pairs (mean age 60 years, 81% women) with primary OA with multiple joint involvement. Radiographic OA was assessed according to the Kellgren–Lawrence (KL) method. Heritability estimates of the tibiofemoral angle were calculated by comparing twice the between-sibling variance with the total variance; adjustments were made for age, gender, body mass index, history of meniscectomy, lower limb fracture and in analyses including all knees, for KL score.
Results 360 subjects representing 180 families were studied. The mean (SD) tibiofemoral angle of right and left knees in the probands was 182.7 (2.9)° and 182.8 (2.6)°, respectively; similar angles were measured in the siblings. Radiographic knee OA (KL score ≥2) was present in 27% of the knees. Stratified analyses in sib pairs with non-osteoarthritic right or left knees showed adjusted heritability estimates of the tibiofemoral angle of the right and left knees of 0.42 (95% CI 0.02 to 0.82) and 0.56 (95% CI 0.19 to 0.93). In addition, adjusted heritability estimates of the tibiofemoral angle in all right and left knees were calculated, being 0.48 (95% CI 0.18 to 0.78) and 0.50 (95% CI 0.21 to 0.79), respectively.
Conclusion The alignment of the tibiofemoral joint is influenced by familial factors, implying that tibiofemoral malalignment may add to the genetic predisposition for knee OA development. These results need to be confirmed in other study populations.
Statistics from Altmetric.com
Introduction
Knee osteoarthritis (OA) is a prevalent musculoskeletal disorder and one of the leading causes of disability world wide. The prevalence of knee OA rises with age, hence the social and economic burden of the disease is increasing rapidly.1 Currently, no cure is available and the identification of potentially modifiable risk factors is therefore crucial.
Knee OA is a heterogeneous disease caused by an interplay of local biomechanical factors, such as mechanical stress, meniscectomy, overweight and muscle weakness, and a systemic predisposition, determined by age, gender, hormonal, metabolic and genetic factors.2 The genetic predisposition of knee OA has extensively been studied.3 4 In a twin study in women, heritability for knee OA was found to be 39%,5 and susceptibility genes for knee OA have been demonstrated.3 4 How aetiological factors underlie this genetic predisposition is the focus of ongoing research. Genetic factors in OA may not be directly related to the homoeostasis of cartilage or bone, but may be related to other risk factors, including local mechanical factors.
Knee malalignment is a local biomechanical factor leading to unfavourable loading of the knee.6 Several studies have shown an association between knee malalignment and structural progression of knee OA.7,–,9 Recently, also, an association between knee malalignment and development of knee OA was shown.10 The finding that malalignment may precede OA had been shown earlier in animal studies.11 The association of malalignment with knee OA development suggests that malalignment as a result of OA modifies the OA process, and also that malalignment can precede and lead to knee OA in itself.12
Which factors contribute to knee malalignment in subjects without knee OA remains to be investigated. It might be that knee malalignment is under genetic control. Malalignment of the knee could then partially explain knee OA heritability. Sib pair studies have shown that knee structures and other mechanical factors, including bone size, cartilage volume and quadriceps strength, are under strong genetic control, independently of each other, and may contribute to the pathogenesis of knee OA.13,–,15
The aim of this study was to investigate the genetic contribution to knee (mal)alignment. In a large sib pair study of 360 subjects with OA at multiple sites we measured the tibiofemoral angle (as anatomical axis) on knee radiographs to estimate the heritability of knee alignment. The ultimate goal was to provide further insight into the knee alignment–knee OA relationship.
Patients and methods
Study design and patients
All subjects were participants of the GARP (Genetics ARthrosis and Progression) study, a prospective cohort study, comprising Caucasian sib pairs with OA at multiple joint sites. Details of recruitment and patient selection of the GARP study have been reported elsewhere.16 Briefly, patients (probands) between 40 and 70 years of age with OA at multiple sites were recruited through rheumatologists, orthopaedic surgeons and general practitioners. Subsequently, siblings were introduced via their eligible probands. OA at multiple sites was defined as symptomatic OA in at least two of the following joint sites: hands, knees, hips or spine. Symptomatic knee OA was defined according to American College of Rheumatology criteria with pain or stiffness on most days of the previous month and presence of radiographic osteophytes.17 Patients with secondary forms of OA, familial syndromes with Mendelian inheritance pattern, major congenital or developmental disorders and metabolic-genetic disorders related to joint disease were excluded. For this study, cross-sectional data from the baseline visit were used. Two patients with bilateral knee prostheses were excluded, along with their siblings. The GARP study was approved by the medical ethics committee of the Leiden University Medical Centre and written informed consent was obtained from all participants.
Clinical assessment
Clinical characteristics—age, gender, weight (kg), height (m), body mass index (BMI; kg/m2), history of meniscectomy or lower limb fracture—were recorded by standardised questionnaires.
Knee radiographs
Standardised non-fluoroscopic weightbearing/semiflexed posterior–anterior radiographs of the right and left knee were obtained in a single centre. A SynaFlex x-ray positioning frame was used to facilitate the uniform anatomical alignment of the knee. An experienced radiographer obtained all the radiographs by a standardised protocol. Radiographic severity of knee OA was graded by a single experienced musculoskeletal radiologist (HMK) according to the Kellgren–Lawrence (KL) scoring method18: a five-scale scoring system with ascending severity. A KL score of ≥2 denotes OA in a particular joint. The intrareader reproducibility, based on the examination of 40 radiographs, of the KL scores depicted as the intraclass correlation coefficient (95% CI) was 0.92 (95% CI 0.86 to 0.96).16
Measurement of tibiofemoral alignment
All knee radiographs were digitised via a film digitiser at a resolution corresponding to a pixel size of 0.1 mm. After image calibration, digital callipers of medical imaging software (Ortho-CMS, Medis, Leiden, The Netherlands) were employed with the help of a computer mouse to measure tibiofemoral anatomical alignment. One trainer examiner (AT) measured all tibiofemoral angles. Anatomical knee alignment was defined as the medial angle subtended by a line drawn through the middle shaft of the femur with respect to one drawn through the middle shaft of the tibia as described by Moreland et al.19 The origin of these lines was 10 cm from the knee joint margins when included in the field of view on the radiograph, otherwise the furthest distance from the joint margins was used. The tibiofemoral angle was measured in degrees with two decimals on a continuous scale. The reader was blinded to the patient's characteristics during assessment. The intraobserver and interobserver reproducibility for assessing the tibiofemoral angle, depicted as intraclass correlation coefficients, were 0.98 (95% CI 0.97 to 0.99) and 0.96 (95% CI 0.95 to 0.98), respectively. The reproducibility was based on the repeat assessment of a random sample of 40 patients (80 knees) by the original reader (AT) and a second trained reader (JB).
Statistics
Familial aggregation (heritability) of the tibiofemoral angle was estimated by comparing twice the between-sibling variance with the total variance.The heritability estimates indicate the fraction of the total variance that is explained by shared genetic and environmental factors. Estimates in non-osteoarthritic knees were adjusted for age, gender, BMI, history of meniscectomy or fracture. Estimates in all knees were additionally adjusted for KL score. All analyses were made using SPSS for windows 14.0 (SPSS, Chicago, Illinois, USA). Heritability estimates are provided with 95% CI.
Results
Population description
A total of 378 patients representing 189 families were eligible for this study. Radiographs of 11 patients (four within two families and seven from independent families) were unavailable for the measurement of the tibiofemoral angle; hence these families were excluded leaving 180 families to be examined (response rate 95.2%). Clinical and radiographic characteristics of the excluded patients were similar to those of the whole study population (data not shown).
In this study, 81% of subjects included were women and symptomatic knee OA was present in one-third of the study population (table 1). Radiographic knee OA was present in 40% of the patients. Approximately three-quarters of the patients had OA involvement of the hands (73%) and spine (80%), 25% had OA involvement of the hips.
Eight patients from separate families (six probands and two siblings) had unilateral knee replacements (seven right, one left) leaving 712 knees to be measured. Radiographic knee OA was evident in 28.5% and 25.1% of probands and siblings respectively. A small minority of subjects had a history of meniscectomy or lower limb fracture (table 2).
Heritability estimates of the tibiofemoral angle
A stratified analysis was performed in knees with a KL score <2, that is, non-osteoarthritic knees. Ninety-three families had two siblings each with non-osteoarthritic right knees and 109 families had two siblings each with non-osteoarthritic left knees. Heritability estimates adjusted for age, gender, BMI, and history of meniscectomy or fracture were 0.42 and 0.56 for the right and left knee, respectively (table 3). Analyses in all knees were performed as well (table 3). After adjustments for age, gender, BMI, history of meniscectomy or fracture, as well as for KL score the heritability estimates were 0.48 and 0.50 for the right and left knee, respectively.
Discussion
As far as we know, this is the first study which systematically investigates familial aggregation of tibiofemoral alignment in a large number of sib pairs. We estimated a heritability of up to 56%, suggesting a substantial genetic contribution to the tibiofemoral alignment.
Our results are in contrast with an earlier study by Zhai et al, who reported a zero heritability estimate for the tibiofemoral angle.20 The difference in results might be explained by the difference in study population, especially the difference in the number of families included—45 families in the previous study, compared with 180 families in this investigation. Another explanation might be a different ethnic background of the subjects in the study by Zhai et al. They studied subjects from Tasmania, whereas Caucasians with Dutch ancestry were included in this study. Earlier studies in Chinese subjects have shown that knee alignment is different in Chinese in comparison with Caucasians, suggesting that ethnicity has a role which may be genetically determined.21 22 Moreover, in the study by Zhai et al subjects were younger and more often male. The difference in studied subjects is also illustrated by the mean tibiofemoral angle of the right knees studied by Zhai et al, which was 180.4º. Moreover, Zhai et al studied right knees only. Finally, the method for assessment of the tibiofemoral angle used by Zhai was different: manually versus digitised computerised. Hence, not only were right knees of fewer families studied by Zhai et al, but the subjects studied and the method used were different from those in this study, which may explain the difference in results.
Results from our study suggest a strong genetic contribution to tibiofemoral alignment. What processes lead to malalignment is not clear. Most information about malalignment is available for people with knee OA for whom factors such as bone attrition, meniscal degeneration and subluxation, cartilage loss, osteophytes and ligament damage have been suggested to contribute to knee malalignment.23 24 Which factors contribute to knee malalignment in subjects without knee OA remains to be investigated. One could hypothesise that ligament laxity or form and size of the bone may contribute to (mal)alignment of the knee. These processes could all be under genetic control and could be involved in OA development. For example, it has been suggested that the frizzled-related protein gene (FRZB), which has been associated with knee OA development, is involved in skeletal morphogenesis.3 25
Whether and how tibiofemoral alignment is an intermediate leading to joint vulnerability and eventually knee OA has to be further elucidated. The association between tibiofemoral alignment and knee OA development has scarcely been studied. Brouwer et al have reported a clear association in a large population study—the Rotterdam study. However Hunter et al failed to show such an association in the Framingham study.10 26 Our study, demonstrating a genetic contribution to knee malalignment independent of knee OA, is in line with the results from Brouwer et al and indicates that knee malalignment may be a risk factor for knee OA development. Further research in longitudinal cohorts is warranted to obtain more insight into the contribution of tibiofemoral alignment in knee OA pathogenesis.
Several confounders might potentially bias the heritability estimate of tibiofemoral alignment, such as sex, BMI and radiographic OA. To minimise the effect of confounding we stratified for radiographic OA and we adjusted for potential confounders in our analyses. However, in this study only a minority (26.8%) of the knees had radiographic OA and no familial aggregation for knee OA in itself was demonstrated.16 Analyses in all knees with adjustment for KL score gave the same results. Consequently, we concluded that the heritability estimates reflect the tibiofemoral alignment independent of radiographic knee OA.
This study has several limitations. The first limitation concerns the use of semiflexed knee radiographs for the assessment of the tibiofemoral angle. Although fully extended full limb radiography is the “gold standard” for the assessment of the tibiofemoral angle, the need for specialised equipment, high costs and radiation exposure to the pelvis make it impractical for large epidemiological studies. Knee radiographs avoid all these disadvantages and the tibiofemoral anatomical axis can be easily produced from them. Assessment of the tibiofemoral angle via the tibiofemoral anatomical axis has recently been validated24 27; furthermore, a strong correlation has been shown with the mechanical axis,27 28 and therefore, the tibiofemoral anatomical axis can be used in OA epidemiological studies. In two of these validation studies knee radiographs in flexion (semiflexed or fixed-flexion) were used.24 28
The second limitation concerns the lack of definite cut-off points for the neutral alignment. Hence we have used the tibiofemoral anatomical angle as a continuous variable for the comparisons to overcome this limitation.
Third, this study has the limitations of any other sib pair study—namely, that a shared environment may result in falsely higher estimates of heritability. We ascertained that exposure to strenuous physical labour or participation in elite sport was not shared between the sibs. Information on dietary habits in childhood was, however, not available. Furthermore we adjusted for as many relevant confounders as possible, such as BMI, meniscectomy and lower limb fractures.
Fourth, although adjustments included most of the potential confounding factors, some factors such as bone mineral density, were not included. However, since the influence of this confounding factor on tibiofemoral alignment is not known, we do not expect that this would substantially change the results. Finally, this study was conducted in middle-aged patients with familial OA at multiple joint sites, where the most affected joint sites were the hands and spine. The majority of patients were women. Whether this phenotype has influenced the findings on the heritability for the tibiofemoral angle is unclear.
Therefore, it is our opinion that similar studies are warranted in other study populations to confirm these results.
Acknowledgments
We are grateful to Medis for the use of Ortho-CMS medical imaging software. This study was performed when AT was a Pfizer articulum fellow.
References
Footnotes
-
Funding The GARP study was financially supported by Pfizer, Groton, USA.
-
Competing interests None.
-
Patient consent Obtained.
-
Ethics approval This study was conducted with the approval of the medical ethics committee of the Leiden University Medical Centre.
-
Provenance Not commissioned; externally peer reviewed.