Legg-Calvé-Perthes disease is a pediatric hip disorder characterized by unilateral or bilateral idiopathic osteonecrosis of the proximal femoral epiphysis1. The etiology of Legg-Calvé-Perthes disease is still unclear. Impairment of the blood supply due to extrinsic compression of vessels or intravascular occlusion has been suggested as a cause of Legg-Calvé-Perthes disease, following the observation of hypercoagulability in adults with idiopathic osteonecrosis2.
Coagulation abnormalities that can result in a hypercoagulable state, such as deficiency of antithrombin, protein C or S deficiencies, the presence of factor V Leiden or prothrombin G20210A, and elevated levels of lipoprotein(a) or fibrinogen, have been investigated in children with Legg-Calvé-Perthes disease. However, the results remain conflicting. Glueck et al. repeatedly reported coagulation abnormalities in 60% to 80% of children with Legg-Calvé-Perthes disease3-6. In most cases, these abnormalities were due to thrombophilia (protein C or S deficiency, activated protein C resistance, or the factor V Leiden mutation), but hypofibrinolysis was also observed. Other authors have failed to reproduce these results and found coagulation abnormalities in =12% of patients or no difference in coagulation factors between patients with Legg-Calvé-Perthes disease2,7-11 and controls12-17.
In the present study, we evaluated the contribution of a hypercoagulable state to the etiology of Legg-Calvé-Perthes disease by determining the association of the factor V Leiden mutation, the prothrombin G20210A mutation, protein C and S deficiency, high levels of factor VIII, and high levels of fibrinogen with the risk of Legg-Calvé-Perthes disease in a population-based case-control study.
Study Population
Patients and a first control group were recruited between 2000 and 2003 from the Ikazia and St. Clara hospitals in Rotterdam, the Netherlands. The study included 169 patients who had been diagnosed with Legg-Calvé-Perthes disease between 1965 and 2002 as well as thirty-eight controls who had been frequency matched on the basis of sex (control group 1). The controls were children treated for asthma in the same hospitals. All patients completed a questionnaire that included questions about the time of onset and diagnosis of the Legg-Calvé-Perthes disease. The diagnosis was confirmed by radiographic and physical examination by one of the participating orthopaedic surgeons. Exclusion criteria were other causes of necrosis of the hip, epiphyseal dysplasia, and femoral fracture.
A second control group (control group 2) consisted of all control subjects from a population-based case-control study on risk factors for venous thrombosis (the Leiden Thrombophilia Study [LETS]18). The 474 controls were adults and acquaintances or partners of patients with an objectively confirmed first episode of deep venous thrombosis.
The study design was approved by the medical ethics committee, and written informed consent was obtained from all participants or their parents.
Blood Collection
Blood samples were collected in 4.5-mL Vacutainers (BD, Franklin Lakes, New Jersey) containing 0.106 mol/L trisodium citrate. Blood plasma was separated by centrifugation for ten minutes at 3300 rpm at room temperature. Five milliliters of citrated plasma and leukocytes were stored at —70°C. DNA was isolated from the leukocytes by standard procedures. The blood-collection procedure was the same for the patients and both control groups.
Coagulation Analysis
The factor V Leiden mutation (G1691A) and the prothrombin mutation (G20210A) were detected by polymerase chain reaction as previously described19,20. Genotypes could not be determined in samples from three patients and three controls tested for the factor V Leiden mutation. Protein C and S antigen levels were measured with enzyme-linked immunosorbent assay with use of polyclonal antibodies21. Factor VIII coagulant activity was determined with a one-stage clotting assay22 on a coagulation analyzer (STA; Diagnostica Stago, Roche Diagnostics, Almere, the Netherlands) according to the instructions of the manufacturer. The fibrinogen concentration was determined according to the Clauss method with use of the Fibriquik Thrombin Reagent (Organon, Oss, the Netherlands) on a coagulation analyzer (STA). Dade Thrombin (Baxter, Miami, Florida) was used in the LETS23,24.
Statistical Analysis
Frequencies of abnormalities of each of the coagulation factors were compared between patients and controls by simple cross tabulation. Relative risk estimates for the abnormalities were obtained by the calculation of odds ratios. Ninety-five percent confidence intervals were constructed according to the Woolf method25. Adjustment for age and sex was performed, with use of logistic regression, for the analysis of the effects of factor VIII, fibrinogen, and proteins S and C.
We used adult reference values to determine abnormal levels of fibrinogen, protein S, and factor VIII that could result in a hypercoagulable state. To determine abnormal levels of protein C, we used a lower cutoff point than is usual for adults because protein C levels are lower in children. Levels were defined as abnormal when fibrinogen concentrations were >4.0 g/L, factor VIII levels were >150 IU/dL, protein C levels were =55 U/dL, and protein S levels were <67 U/dL26. Factor VIII levels were also categorized in four groups, and risk calculations were performed relative to the group with the lowest factor VIII level.
The influence of coagulation abnormalities was separately analyzed in patients with an early onset of the disease (at an age of five years or younger) and those with a late onset of the disease (at an age of older than five years).
Separate odds ratios for males and females with and without the coagulation abnormality were calculated relative to females without the abnormality. The risks for the groups with one or more or two or more coagulation abnormalities were calculated relative to the group with no abnormalities.
Source of Funding
A. Vosmaer received an educational grant from MSD (Merck Sharp and Dohme) to defray laboratory costs. The sponsor had no role in the design of the study, the collection or analysis of the data, or the writing of the report.
The general characteristics of the study population are summarized in Table I. There were more males than females in both the patient group (70% male) and the sex-matched pediatric control group (control group 1; 63% male). In the LETS control group (control group 2), 43% of the subjects were men.
The symptoms of Legg-Calvé-Perthes disease started, on the average, at the age of 5.7 years, and the mean age at the time of diagnosis was 6.1 years. The mean age at the time of blood collection was older than the age at onset because we tested most of the patients some time after the onset of the disease. However, the mean age at the time of blood collection was similar for the patients (12.5 years) and control group 1 (11.9 years) (mean difference, -0.58 year; 95% confidence interval, -2.4 to 1.3 years). The 474 subjects in control group 2 had a mean age of 46.6 years (range, 16.3 to 73.1 years).
Right and left hips were equally affected, and both hips were involved in 14.2% of the cases.
Genetic Coagulation Factors
The results of the analyses of the factor V Leiden and prothrombin G20210A mutations in the patients with Legg-Calvé-Perthes disease and the controls are shown in Table II. The factor V Leiden mutation was more common in the patients (sixteen of 166, 9.6%) than in all controls combined (sixteen of 509, 3.1%). All patients with the factor V Leiden mutation were heterozygotes. Only one control subject was homozygous for this mutation. The odds ratio for the development of Legg-Calvé-Perthes disease in the presence of the factor V Leiden mutation compared with the wild-type genotype was 3.3 (95% confidence interval, 1.6 to 6.7).
The prevalence of the prothrombin G20210A mutation was 5.3% (nine of 169) in the patients with Legg-Calvé-Perthes disease and 2.1% (eleven of 512) in the controls. An association was observed between the prothrombin G20210A mutation and Legg-Calvé-Perthes disease with an odds ratio of 2.6 (95% confidence interval, 1.0 to 6.3).
Other Coagulation Factors
The results of the analyses of the factor VIII, fibrinogen, protein C, and protein S levels are presented in Table II. Odds ratios were adjusted for age and sex. This was necessary because of the strong age difference between the cases and controls.
Elevated levels of factor VIII were associated with the risk of Legg-Calvé-Perthes disease (odds ratio, 7.5; 95% confidence interval, 2.2 to 25.2 when a level of >150 IU/dL was compared with a level of <100 IU/dL). Protein S deficiency increased the risk of Legg-Calvé-Perthes disease slightly (odds ratio, 2.8; 95% confidence interval, 0.7-10.8).
Neither increased fibrinogen concentrations (>4.0 g/L) nor low levels of protein C showed an association with the risk of Legg-Calvé-Perthes disease.
Influence of Sex
Legg-Calvé-Perthes disease was more common in males than in females, but we could not calculate the odds ratio for sex directly because the pediatric controls were matched to the cases with respect to sex. However, since we know the sex distribution in the population, an odds ratio for sex can be estimated to be 2.4 (95% confidence interval, 0.5 to 11.1).
Table III shows a comparison of risks associated with coagulation abnormalities for males and females separately. The combination of the factor V Leiden mutation and male sex increased the risk of Legg-Calvé-Perthes disease eighteen times (odds ratio, 18.0; 95% confidence interval, 5.6 to 58.3) as compared with the risk for females without the mutation. The prothrombin mutation increased the risk to a similar extent in females and males, although the numbers of subjects with the mutation were small. We found odds ratios of 4.0 (95% confidence interval, 0.6 to 24.4) for females with the prothrombin mutation and 5.2 (95% confidence interval, 1.8 to 15.1) for males with the mutation.
We could not calculate odds ratios for anticoagulant deficiencies for each sex since these deficiencies were too rare.
High levels of fibrinogen (>4.0 g/L) did not affect the risk of Legg-Calvé-Perthes disease in females but did affect the risk in males (Table III). Also, increased levels of factor VIII had virtually no effect on the risk of Legg-Calvé-Perthes disease in females. In contrast, increased levels of factor VIII were found to elevate the risk of Legg-Calvé-Perthes disease in males to an odds ratio of 31.4 (95% confidence interval, 5.3 to 184.9) when a level of >150 IU/dL was compared with a level of <100 IU/dL.
Effect of Coagulation Abnormalities on Time of Onset of Legg-Calvé-Perthes Disease
There was no clear difference in the effect of coagulation abnormalities between children with an early onset of Legg-Calvé-Perthes disease (at an age of five years or younger) and those with a later onset of the disease (at an age of older than five years). However, the effect of the following were stronger in children with an early onset than in those with a later onset: the presence of factor V Leiden (odds ratio, 5.3 [95% confidence interval, 2.2 to 12.7] compared with 2.7 [95% confidence interval, 1.0 to 7.1]), low levels of protein C (odds ratio, 8.6 [95% confidence interval, 1.7 to 43.7] compared with 0.9 [95% confidence interval, 0.07 to 10.7]), and high levels of fibrinogen (odds ratio, 1.5 [95% confidence interval, 0.4 to 5.8] compared with 0.7 [95% confidence interval, 0.2 to 2.7]). In contrast, the effect of the following were stronger in children with a later onset than in those with an early onset of the disease: high levels of factor VIII (odds ratio, 3.3 [95% confidence interval, 0.7 to 16.7] compared with 7.7 [95% confidence interval, 1.9 to 31.9]) and protein S deficiency (odds ratio, 2.7 [95% confidence interval, 0.3 to 25.0] compared with 4.5 [95% confidence interval, 1.0 to 20.0]). We could not directly calculate the effect of the prothrombin mutation in these two groups because this mutation was not observed in children with an early onset of the disease.
Presence of Any Coagulation Abnormality
All of the coagulation abnormalities were considered together, and odds ratios were calculated for the presence of one or more coagulation abnormalities or two or more coagulation abnormalities relative to the absence of any abnormality (Table IV). The risk of Legg-Calvé-Perthes disease increased with an increasing number of coagulation abnormalities, with an odds ratio of 2.2 (95% confidence interval, 1.1 to 4.2) for the presence of one or more abnormalities and an odds ratio of 3.5 (95% confidence interval, 0.8 to 15.7) for the presence of two or more abnormalities as compared with no abnormalities.
When these odds ratios were calculated separately for males and females, we found that the number of coagulation abnormalities had almost no effect on the risk in females but had a clear effect in males. Compared with females without any coagulation abnormalities, males had an eightfold increase in the risk of Legg-Calvé-Perthes disease in the presence of one or more abnormalities (odds ratio, 8.1; 95% confidence interval, 3.0 to 22.1) and a fifty-eight-fold increase in the presence of two or more coagulation abnormalities (odds ratio, 58.2; 95% confidence interval, 4.5 to 747.8).
The etiology of Legg-Calvé-Perthes disease is still unclear, but it likely involves successive vascular occlusions, in which hypercoagulable disorders may play a role. However, because of the rarity of the disorder, previous studies on this subject often involved small numbers of cases and the results were inconclusive. Discrepancies in results may also be explained by differences in selection of participants, in laboratory measurements, or in statistical analyses. We evaluated the association of the factor V Leiden mutation, the prothrombin G20210A mutation, protein C and S deficiency, and high levels of factor VIII and fibrinogen with Legg-Calvé-Perthes disease in a large group of patients. We found that both the factor V Leiden and the prothrombin mutation, protein S deficiency, and high levels of factor VIII were associated with an increased risk of Legg-Calvé-Perthes disease. The risk also increased gradually with an increasing number of coagulation abnormalities. This pattern suggests involvement of a thrombotic component in the etiology of Legg-Calvé-Perthes disease.
The higher frequency of the factor V Leiden mutation in cases than in controls was consistent with the findings in other reports2,5-7,10,27. We found the factor V Leiden mutation in 9.6% (sixteen) of 166 cases and in 3.1% (sixteen) of 509 controls, rates that are similar to the rates of 4.9% to 11% in cases and 0.7% to 5% in controls described in the other reports. However, there have been studies in which no association between the factor V Leiden mutation and the risk of Legg-Calvé-Perthes disease could be established8,14.
The presence of the prothrombin G20210A mutation also increased the risk of Legg-Calvé-Perthes disease, with an odds ratio of 2.6 (95% confidence interval, 1.0 to 6.3), a finding that was in line with the odds ratio of 2.1 (95% confidence interval, 0.4 to 8.5) in one other report8. Others who have investigated this mutation found no association with Legg-Calvé-Perthes disease2,6.
Several authors reported an association between protein S deficiency and the risk of Legg-Calvé-Perthes disease, which is in agreement with our results3,4,7,14, but just as many opposite results have been published6,9,10,13,15,16. The results of studies of protein C deficiency are inconclusive and are hard to compare because different definitions of deficiency were used3-5,7,12,17. In our study, protein C deficiency was not observed more often in the patients than in the controls. The low frequency of these deficiencies has resulted in a low power to find an effect in all studies, including ours.
We found no association between high levels of fibrinogen and Legg-Calvé-Perthes disease, in agreement with all other studies in which this relationship was examined4,9,10,15,28. We found an association between increasing levels of factor VIII and an increased risk for Legg-Calvé-Perthes disease, a relationship that has not been described before, to our knowledge.
Our results suggest that coagulation abnormalities have a contributing role in the etiology of Legg-Calvé-Perthes disease. A thrombus can obstruct venous outflow, leading to increased intraosseous venous pressure, reduced arterial flow, hypoxia, and osteonecrosis of the femoral head29. The veins of the femoral head might be more prone to intravascular occlusion because they are very thin-walled, have a very sluggish blood flow, and spiral around arteries30,31.
Legg-Calvé-Perthes disease affects boys five times more often than it does girls32, while coagulation abnormalities are equally distributed across the sexes. The difference in the effects of coagulation abnormalities between males and females suggests that coagulation abnormalities have a greater impact on the circulation of the femoral heads of males. Chung described the anatomy of the vasculature around the femoral head in boys as being different from that in girls33, which could mean that, in the case of thrombophilia, venous obstruction occurs more easily in boys than in girls. Another explanation could be that boys experience more trauma at young ages. Clot formation after damage to vessels around the femoral head could then result in larger clots when thrombophilia is present, which would lead to more serious obstruction of the blood flow and more damage to the femoral head.
Our study had some limitations, which are to a large extent the result of difficulties with the performance of venipunctures in healthy minors. We could not use adult reference values for all coagulation factors to determine abnormalities. Factor VIII, fibrinogen, and protein S levels reach adult levels early in childhood, but protein C levels do so much later in childhood—i.e., around the age of sixteen years26. The cutoff point usually used to determine deficiency in adults is 67 U/dL26. Because most of our patients were children, we used a lower cutoff point of 55 U/dL and adjusted for age. It should also be noted that we performed only one measurement and did not conduct genetic or family studies for protein C or protein S.
We included a second group of control subjects because of the small number of pediatric controls. This led to the inclusion of one small hospital-based control group and a large control group of adults in our study. These controls are entirely appropriate for analyses that are not related to asthma (the disease affecting the pediatric control group) or age. Thus, they are appropriate for genetic analyses and, in all likelihood, for the comparison of levels of factor VIII, fibrinogen, and protein S. Furthermore, we adjusted for age in the analyses, so age-related differences should have been taken into account. Protein C levels, however, are clearly age-dependent. Therefore, we used a stringent cutoff level and performed an additional analysis restricted to the pediatric control group for this parameter, which yielded results similar to those of the analysis that included both control groups.
We tested most of the patients some time after the onset of the disease or even after recovery from the disease. Therefore, the measurements may not be representative of those at the time that the disease occurred. If transient coagulation abnormalities were the cause of the disease, we may not have been able to detect them at these later ages. However, coagulation abnormalities are for a large part genetically determined, so it is likely that they are present at any age.
Our results suggest that coagulation abnormalities contribute to the development of Legg-Calvé-Perthes disease and therefore that the etiology has a thrombotic component. The question is whether treatment with anticoagulants could modify the process of Legg-Calvé-Perthes disease, which can be serious and can result in severe osteoarthritis later in life. Treatment with anticoagulants can slow or stop progression of osteonecrosis in adults with coagulation abnormalities34; therefore, it can be speculated that this treatment will also shorten or reverse the acute phase of Legg-Calvé-Perthes disease and reduce the common complication of osteoarthritis in adulthood. However, Legg-Calvé-Perthes disease is often diagnosed at a late stage, when anticoagulation treatment is probably no longer effective. Additionally, anticoagulants obviously introduce a bleeding risk. Therefore, both the risks and the benefits of such a treatment should be considered carefully.
In conclusion, in this study we found that thrombotic abnormalities play a role in the etiology of Legg-Calvé-Perthes disease, particularly in males.