Meta-analysis of hypercoagulability genetic polymorphisms in perthes disease

Authors


  • The authors have no financial relationships relevant to this article to disclose.

ABSTRACT

Perthes disease is an osteonecrosis of the femoral epiphysis with unclear etiology. This study aimed to systematically review the association between genetic determinants of hypercoagulability (Factor V Leiden, prothrombin II, and methylenetetrahydrofolate reductase; MTHFR) and Perthes disease. PubMed and Scopus searched from inception to January 2012, data extraction and quality assessment were performed. The odds ratio (OR) for the allele effect was pooled, and heterogeneity and publication bias were assessed. Twelve case–control studies met inclusion criteria and had sufficient data for extraction. There were 824 cases and 2,033 controls with a mean age range of 6.1–14.7 years. The prevalence of the minor allele in controls was 0.015 (95% confidence interval (CI): 0.008, 0.023), 0.012 (95% CI: 0.008, 0.017), and 0.105 (95% CI: 0.044, 0.167) for factor V Leiden, prothrombin II, and MTHFR, respectively. The factor V Leiden allele increased the risk of Perthes with a pooled OR of 3.10 (95% CI: 1.68, 5.72), while prothrombin II and MTHFR had non-significantly pooled OR 1.48 (95% CI: 0.71, 3.08), and 0.97 (95% CI: 0.72, 1.30), respectively. The factor V Leiden mutation is significantly related to Perthes disease, and its screening in at-risk children might be useful in the future. © 2013 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 32:1–7, 2014.

Perthes disease is an osteonecrosis of the femoral head that occurs in children. The blood supply to the femoral epiphysis is partially and temporarily disrupted causing a deformity of the femoral head. Although it was described over 100 years ago, its etiology is still controversial; possible causes include trauma, transient synovitis of the hip, socioeconomic problems, smoking, and coagulopathy.[1-3] Due to the pattern of ischemia, coagulation defects have been postulated to impair the blood supply to the femoral epiphysis as reported in adults.[4] Mutation of genes controlling these factors results in a prothrombotic tendency which may lead to vascular occlusion of the femoral epiphysis.

Recently genetic polymorphisms which have been investigated in relation to Perthes include factor V Leiden, prothrombin II, and methylenetetrahydrofolate reductase (MTHFR).[2] The factor V Leiden mutation is located on the first chromosome 1q23, exon 10 (MIM# 612309). A missense nucleotide substitution occurring at position 1691 or 1746 (G → A) results in resistance to cleavage by activated protein C and an increased risk of venous thromboembolism.[5] The prothrombin II mutation is located in 11p11–q12 (MIM#176930) at position 20210 (G → A) and activates the coagulation cascade which increases the risk of thrombosis. The MTHFR gene is located at 1p36.3 (MIM#607093). A nucleotide change at position 677 (C → T) which leads to higher homocysteine levels has been associated with coronary heart disease, and thromboembolic events.[6] These factors have been demonstrated as risk factors for Perthes disease in some studies[7-9] but not in others.[2, 10, 11]

In view of these conflicting results and the rare incidence of Perthes, we thought that a meta-analysis of previous studies would increase power and precision. We therefore carried out a systematic review and meta-analysis to determine association between factor V Leiden, prothrombin II, and MTHFR polymorphisms and Perthes disease.

METHODS

Search Strategy

Studies which addressed association between coagulation gene polymorphisms and Perthes disease were identified from two main databases (Medline and Scopus) since their inception to January 2012. Other specific journals, that is, orthopedic journals (Journal of Bone and Joint Surgery, Journal of Pediatric Orthopaedics, Clinical Orthopaedic and Related Research), and reference lists were also searched. The search strategy is given in detail in appendix 1.

Inclusion Criteria

Observational studies (case–controls, cross-sectional studies, and case series with population control data) which assessed the association between Perthes and factor V Leiden or prothrombin II or MTHFR were included. Perthes disease was diagnosed by a combination of history, physical examination, and radiographs.

Exclusion Criteria

Other causes of osteonecrosis of the femoral epiphysis such as Gaucher's disease, steroid induced osteonecrosis, skeletal dysplasia, hematologic diseases, and traumatic injury were excluded. We also excluded studies with inadequate description of control samples.

Data Extraction

Two authors (PW and CT) extracted all data independently. We used a standardized data extraction form including demographic data, family history of thrombophilia, severity of Perthes (Lateral Pillar Classification; A, B, and C), and genotype frequency. Any disagreement was resolved by consensus. Allele frequency was identified based on genotyping reports.

Quality Assessment

Two authors (PW and TW) separately assessed the quality of studies by using a previous tool (Risk of Bias Assessment for Genetic Association Studies).[12]

Statistical Analysis

The prevalence of the minor allele for each study was pooled across studies. A “per-allele” odds ratio (OR), equivalent to an additive genetic model under Hardy–Weinberg equilibrium (HWE), along with 95% confidence interval (CI) was estimated for each study.[13] Heterogeneity of the allele effect was assessed using a Q-test and the degree of heterogeneity was quantified using I2. If heterogeneity was present (I2> 25% or Q-test was significant), the allele effects were pooled using a random-effects model, otherwise the fixed-effects model was used. Publication bias that might occur from including only significant results, English, or cited studies was assessed using Egger's test.[14, 15] All analyses were performed using Stata statistical software, release 12, College Station, TX: Stata Corporation; 2012). The level of significance was set at p-value ≤ 0.05, except for the heterogeneity test, for which a level of 0.10 was used.

RESULTS

As described in Figure 1, 240 studies were identified as potentially eligible. One hundred sixty-five of them were duplicates leaving 75 studies to review based on titles. Another 35 studies were not related to Perthes disease. From 40 studies included in the abstract review, 26 of them were not related to genetic mutations. Fourteen studies (12 case–controls, and two case series) focused on Perthes disease and inherited coagulopathy. However, two of them had insufficient data on controls and were excluded from review, leaving 12 case–control studies in the analysis. Seven studies were from Europe[2, 8, 11, 16-19] and the others were from North[7, 9, 20] and South America.[10, 21]

Figure 1.

Study flow diagram of data retrieval.

Demographic data are presented in Table 1. There were 824 children in the Perthes group and 2,033 children in the control group. The mean age ranged from 6.1 to 14.7 years. Between 51% and 86% of participants were male and 63%–81% of participants in the Perthes group had severe involvement (lateral pillar stage B or C). Nine studies[2, 7-11, 16, 19, 20] had newborns to adolescents as controls whereas three studies[17, 18, 21] had adult controls whose average age ranged from 29 to 35 years.

Table 1. Demographic Data
AuthorYearDesignEthnicityMean AgeMale (%)N CaseN ControlFamily History of Thromboliphilia (%)Lateral Pillar B/C (%)Controls (No Perthes)
  1. CC, case-control study.
Glueck[18]1997CCUSA11 641011.6 Children
Arruda[10]1999CCBrazil6.781.861296  Newborns
Hayek[11]1999CCIsrael9.670.56250 62.9Children
Sirvent[14]2000CCFrance7862222  Children
Eldridge[7]2001CCUSA1477.25556  Adolescents
Balasa[9]2004CCUSA6.662.172197  Children
Szepesi[15]2004CCHungary14.772.34730 80.9Adults
Lopez-Franco[16]2005CCSpain  90203  Adults
Yilmaz[17]2005CCTurkey7.176.1467935.476.1Children
Kenet[2]2008CCIsrael 76.5119276 76.1Children
Sanders[19]2009CCBrazil6.675202145 Adults
Vosmaer[8]2010CCNetherlands6.150.7166509  Children
Total     8242,033   

Risk of bias for genetic studies was assessed by two authors (PW, TW) as shown in Table 2. Each domain represent some aspect of quality of the studies with “yes” indicating low risk of bias, and “no” indicating high risk of bias. The risk of bias was lowest for selective outcome reporting and population stratification, in which all studies rated well. The criteria for diagnosis of Perthes were clearly described in 11 out of 12 studies and the risk of ascertainment bias of each case was therefore low. No study assessed HWE, and many had high risk of confounding (9/12). Some studies did not mention any details about quality control for genotyping (8/12), or ascertainment of controls (5/12). We, therefore, classified the risk of bias as unclear.

Table 2. Risk of Bias Assessment
AuthorYearAscertainment of PerthesAscertainment of ControlsQuality Control for GenotypingPopulation StratificationConfounding BiasSelective Outcome ReportHWE
  1. HWE, Hardy–Wienberg equilibrium.
Glueck[18]1997YesUnclearUnclearYesNoYesNo
Arruda[10]1999YesUnclearUnclearYesNoYesNo
Hayek[11]1999YesYesYesYesNoYesNo
Sirvent[14]2000YesYesYesYesYesYesNo
Eldridge[7]2001NoYesUnclearYesNoYesNo
Balasa[9]2004YesYesUnclearYesYesYesNo
Szepesi[15]2004YesUnclearUnclearYesNoYesNo
Lopez-Franco[16]2005YesUnclearYesYesNoYesNo
Yilmaz[17]2005YesYesUnclearYesNoYesNo
Kenet[2]2008YesYesYesYesNoYesNo
Sanders[19]2009YesUnclearUnclearYesNoYesNo
Vosmaer[8]2010YesYesUnclearYesYesYesNo

Factor V Leiden Mutation

From the eligible 12 studies,[2, 7-11, 16-21] factor V Leiden allele frequencies between case and control groups are described in Table 3. The pooled prevalence of A alleles were 5.8% (95% CI: 3.8%, 7.7%) and 1.5% (95% CI: 0.8%, 2.3%) in cases and controls, respectively. The allele effect was highly heterogeneous across studies with I2 of 67.5% (Fig. 2). The pooled OR of A versus G alleles with the random effect model was 3.10 (95% CI: 1.68, 5.72), suggesting the odds of having Perthes disease was 3.10 times significantly higher with the A than G allele. Egger's test suggested that there was no evidence of publication bias (coefficient = 1.03, SE = 1.39, p = 0.472).

Table 3. Estimations of Minor (A) and Major (G) Allele Frequencies and FVL Effects on Perthes
AuthorCaseControl
YearA AllelePrevalenceNo. AllelesA AllelePrevalenceNo. AllelesOR95% CI
  1. FVL, Factor V Leiden; OR, odds ratio; CI, confidence interval.
Glueck[18]199790.07012820.0102027.561.61, 35.59
Arruda[10]199960.04912240.0075927.62.11, 27.37
Hayek[11]199940.03212450.0501000.630.17, 2.42
Sirvent[14]200010.0234400443.070.12, 77.41
Eldridge[7]2001100.09111060.0541121.770.62, 5.04
Balasa[9]2004160.111144140.0363943.391.61, 7.14
Szepesi[15]200490.09494006013.140.75, 230
Lopez-Franco[16]200530.01718050.0124061.360.32, 5.75
Yilmaz[17]2005120.1309220.01215811.702.56, 53.55
Kenet[2]200870.029238260.0475520.610.26, 1.43
Sanders[19]200970.1754080.01942811.143.8, 32.61
Vosmaer[8]2010160.048332160.0161,0183.171.57, 6.41
Pooled 1000.0581,648880.0154,0663.101.68, 5.72
95% CI  0.038, 0.077  0.008, 0.023   
Figure 2.

Forest plot of factor V Leiden gene effects on Perthes.

Prothrombin II Mutation

There were five studies[2, 8, 10, 11, 18] which provided adequate allelic data for prothrombin II (G20210A) mutation analysis (Table 4). One study[10] was not included in pooling due to zero cells for minor alleles in both case and control groups. The pooled prevalence of allele A was 2.3% (95% CI: 1.6%, 3%) among cases, and 1.2% (95% CI: 0.8%, 1.7%) among controls. The allele effect was mildly heterogeneous across studies with an I2 of 26.0% (Fig. 3). The pooled OR was 1.48 (95% CI: 0.71, 3.08) which suggested that the A allele carried a 50% higher risk of Perthes disease when compared with the G allele. However, this pooled OR was not statistically significant. There was no evidence of publication bias suggested from Egger's test (coefficient = −2.45, SE = 1.78, p = 0.304).

Table 4. Estimations of Minor (A) and Major (G) Allele Frequencies and Prothrombin II Mutation on Perthes
AuthorCaseControl
YearA AllelePrevalenceNo. AllelesA AllelePrevalenceNo. AllelesOR95% CI
  1. OR, odds ratio; CI, confidence interval.
Arruda[10]199900122003923.20, 8,310.5
Hayek[11]199910.00812420.021000.40.05, 3.34
Lopez-Franco[16]200540.02218440.014062.230.55, 9.03
Kenet[2]200840.017238110.025520.840.26, 2.67
Vosmaer[8]201090.027338110.0111,0242.521.03, 6.13
Pooled 180.0231,006280.0122,4741.480.71, 3.08
95% CI  0.016, 0.030  0.008, 0.017   
Figure 3.

Forest plot of prothrombin II mutation effects on Perthes.

MTHFR Mutation

Only four studies (2, 10, 11, 18) had sufficient allelic data for the MTHFR polymorphism. The prevalence of the minor T allele in case and control groups was very similar, that is, 10.4% (95% CI: 4.6%, 16.2%) and 10.5% (95% CI: 4.4%, 16.7%), respectively (Table 5). The OR for T versus C alleles was homogenous (I2 = 0) with the pooled OR of 0.97 (95% CI: 0.72, 1.30) as shown in Figure 4.

Table 5. Estimations of Minor (T) and Major (C) Allele Frequencies and MTHFR Mutation on Perthes
AuthorCaseControl
YearT AllelesPrevalenceNo. AllelesT AllelesPrevalenceNo. AlleleOR95% CI
  1. MTHFR, methylenetetrahydrofolate reductase; OR, odds ratio; CI, confidence interval.
Arruda[10]199940.033122240.0415920.80.27, 2.36
Hayek[11]1999160.129124120.121001.090.49, 2.42
Lopez-Franco[16]2005220.122180480.1184061.040.61, 1.78
Kenet[2]2008360.140258820.1495520.930.61, 1.42
Pooled 780.1046841660.1051,6500.970.72, 1.31
95% CI  0.046, 0.162  0.044, 0.167   
Figure 4.

Forest plot of MTHFR effects on Perthes.

DISCUSSION

We systematically reviewed studies from various countries focusing on inherited hypercoagulable factors. The effects of three polymorphisms (i.e., factor V Leiden, prothrombin II, and MTHFR) were assessed in groups of 2,857, 1,167, and 825 individuals, respectively. We found that factor V Leiden was significantly associated with Perthes disease, with the minor A allele carrying an approximately three times higher odds of disease than the G allele, although this showed high heterogeneity. Our findings also suggested a possible 1.5-fold increased risk of disease with the A allele of the prothrombin II polymorphism relative to the G allele, although this was not statistically significant. There was no association between MTHFR polymorphism and Perthes disease.

Perthes disease is avascular necrosis of the femoral epiphysis produced by multiple possible etiologies. However, one of the causes of occlusive femoral epiphyseal vessels is likely to be hypercoagulability, which may be precipitated by smoking,[22, 23] inherited thrombophilia, or fibrinolytic coagulopathy.[1, 24] Many inherited hypercoagulable states have been suspected to be related with Perthes disease such as protein C and protein S deficiency, antiphospholipid antibodies, anticardiolipin, factor V Leiden, prothrombin II, and MTHFR. Unlike other hypercoagulable factors, factor V Leiden, prothrombin II and MTHFR are not confounded by age, vaccination or infection,[2] being genetic factors essentially assigned randomly at conception (also known as Mendelian randomization). Factor V Leiden is the most common inherited prothrombotic state. Its mutation resists activated protein C degradation[5] which removes one of the brakes on the coagulation cascade.[25] Among previous reports, two studies[8, 9] revealed significant relationship between factor V Leiden and Perthes disease with ORs between 3.3 and 3.4. However, the others[7, 18] showed non-significant ORs. After pooling data in our systematic review, the findings supported its association with Perthes disease, with a population attributable risk (PAR) of 3.1%.

The prothrombin II polymorphism was the second possible prothrombotic tendency explored. The magnitude of the association with Perthes disease was 1.48 but this was non-significant, likely due to the low frequency of the allele and the small sample size. For MTHFR, the results of the meta-analysis demonstrated an effect size close to 1. Despite the wide CI, this point estimate likely indicates no association with Perthes disease.

The method of polymorphism detection might affect ORs of mutation related to the Perthes disease. From all included studies, they used genomic DNA polymerase chain reaction (PCR) digested by restriction enzyme, that is, endonuclease Mnl-1 for factor V Leiden mutation,[26] endonuclease Hind III for prothrombin II mutation,[27] and endonuclease Hinf-1 for MTHFR.[28] Only two studies[7, 19] used commercial detection kits for factor V Leiden mutation but their ORs were comparable with the others.[2, 8-11, 16-18, 20, 21]

Differences in ethnicity may cause confounding in genetic association studies. Two studies were conducted in Jewish populations which showed a trend towards a protective effect from inherited hypercoagulability whereas other Caucasian studies showed a risk effect. We, therefore, performed a sensitivity analysis by excluding the two studies and found that the pooled OR was as high as 4.28 (95% CI: 2.70, 6.80) for factor V Leiden and 2.43 (95% CI: 1.15, 5.15) for prothrombin II. However it is unlikely that these polymorphisms behave fundamentally differently in terms of biology between ethnic groups and the differences may be due to chance or different environmental interacting factors.

Our study has some strength. We extensively searched and identified all relevant genetic association studies assessing gene effects on Perthes disease. Data were pooled applying standard meta-analytic methods for genetic association studies.[29] The pooled prevalence of minor alleles and pooled ORs for factor V Leiden, prothrombin II, and MTHFR have been provided. There are however some limitations. The pooled estimate for factor V Leiden was moderately to strongly heterogeneous. We were unable to use more sophisticated pooling methods, for example, pooling genotype groups, and checking genetic models since most studies did not report genotypic data. This also meant that we could not assess HWE and 8/12 (66.7%) did not give details about quality control, so some bias from genotyping error might be present. Although all included studies were case–control design, 5/12 (41.7%) studies did not clearly define controls and thus ascertainment bias might be present. The appropriateness of using adult controls is also debatable, although removing the adult-control studies[17, 18, 21] still led to a significant risk for factor V Leiden (pooled OR = 2.68; 95% CI: 1.38, 5.23). Severity of Perthes disease was unable to retrieve in 8/12 (66.7%) studies. Therefore, ascertainment bias among cases might be occurred. From studies reported Lateral Pillar Classification,[2, 11, 17, 19] the results revealed that patients with factor V Leiden mutation tends to have more severe course of disease.[17]

Physicians might response to Perthes disease with factor V Leiden mutation as (1) anticoagulation therapy that has limitation of results in late stage and risk of bleeding complications in young children,[20] (2) early surgical intervention in early stage to prevent poor results since factor V Leiden mutation tends to have more severe disease,[17] and (3) patient information for probable thromboembolic risks in adulthood.[19]

In conclusion, our meta-analysis suggests that factor V Leiden might increase the odds of Perthes disease about three fold. If correct, this allele may explain about 3% of the cases of Perthes disease in the general pediatric population. We could not detect any associations between prothrombin II and MTHFR polymorphisms and Perthes disease. However, ethnicity may affect the relationship between factor V Leiden as well as prothrombin II and Perthes disease. More high-quality studies and studies among Asians are still needed.

APPENDIX 1

The search terms were

  1. Perthes, Perthes disease, Legg–Calve–Perthes, LCPS, or LCPD
  2. Factor V Leiden or G1691A
  3. Prothrombin II or G20210A
  4. Methylenetetrahydrofolate reductase or MTHFR
  5. Gene* or genetic*
  6. Polymorphism*
  7. Mutation*
  8. The combination between 1 and 2 or 3 or 4 or 5 or 6.

Ancillary