| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Lipids, Lipoproteins, and Cardiovascular Risk Factors |
1 St. Paul’s Hospital Lipid Clinic and the University of British Columbia Department of Pathology and Laboratory Medicine, Vancouver, BC, Canada.
2 The Centre for Health Evaluations and Outcomes Sciences, St. Paul’s Hospital, Vancouver BC, Canada.
aAddress correspondence to this author at: Department of Pathology and Laboratory Medicine, University of British Columbia, Room B180, 1081 Burrard St., Vancouver, BC, V6Z 1Y6, Canada. Fax 604-806-8590; e-mail dtholmes{at}interchange.ubc.ca.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: This was a retrospective analysis of clinical and laboratory data from a large multiethnic cohort of HFH patients at a single, large lipid clinic in Vancouver, Canada. Three hundred and eighty-eight patients were diagnosed with possible, probable, or definite HFH by strict clinical diagnostic criteria. Multivariate Cox regression analysis was used to study the relationship between several established CVD risk factors, Lp(a), and the age of first hard CVD event.
Results: An Lp(a) concentration of 800 units/L (560 mg/L) or higher was a significant independent risk factor for CVD outcomes [hazard ratio (HR) = 2.59; 95% confidence interval (CI), 1.53–4.39; P <0.001]. Other significant risk factors were male sex [HR = 3.19 (1.79–5.69); P <0.001] and ratio of total to HDL-cholesterol [1.18 (1.07–1.30); P = 0.001]. A previous history of smoking or hypertension each produced HRs consistent with increased CVD risk [HR = 1.55 (0.92–2.61) and 1.57 (0.90–2.74), respectively], but neither reached statistical significance (both P = 0.10). LDL-cholesterol was not an independent predictor of CVD risk [HR = 0.85 (0.0.71–1.01); P = 0.07], nor was survival affected by the subcategory of HFH diagnosis (i.e., possible vs probable vs definite HFH).
Conclusion: Lp(a) is an independent predictor of CVD risk in a multiethnic HFH population.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Lipoprotein(a) [Lp(a)] is a plasma lipoprotein composed of an LDL particle to which the glycoprotein apolipoprotein(a) is covalently linked by means of a disulfide bridge to apolipoprotein B100. Apolipoprotein(a) bears striking homology to plasminogen. Although Lp(a) is an established risk factor for CVD in the general population (2), there is some controversy as to its utility as a predictive factor for CVD in the HFH population. Whereas some studies have found that increased Lp(a) is associated with increased risk of CVD(3)(4)(5)(6)(7), others have not(8)(9)(10)(11)(12)(13)(14) or have found it useful in the prediction of very early coronary artery disease (CAD) only(15).
Interestingly, Lp(a) is somewhat resistant to therapeutic lowering with statins, although prolonged high-dose statin therapy (16) has been shown to decrease concentrations by 
25%. Nicotinic acid also lowers Lp(a)(17), whereas combined estrogen and progesterone therapy has been shown to decrease concentrations in postmenopausal women(18)(19). Nevertheless, the relative resistance to treatment allows for the interpretation of Lp(a) concentrations despite the presence of lipid-lowering therapy.
We have observed that HFH patients with high Lp(a) appear to develop CVD more frequently and earlier than those with lower concentrations. We therefore hypothesized that Lp(a) is an independent predictor of CVD risk in the HFH population. Additionally, we sought to assess the relative importance of Lp(a) as a predictor of CVD compared with selected established CVD risk factors.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
) and inclusion criteria (Table 2
) for HFH. Diagnostic criteria were based on those of the Simon Broome Register. Exclusion criteria (Table 2
) were chosen to prevent dilution of the cohort by patients with dyslipidemias other than HFH. For example, many patients with familial combined hyperlipidemia (FCH) would meet the diagnostic criterion of LDL-C 
5 mmol/L (Table 1
) and could confound results if not excluded. Thus, to minimize contamination of the cohort with patients having FCH, those meeting criteria for possible or probable HFH but having triglyceride (TG) concentrations 3.0 mmol/L were excluded. There were 44 exclusions based on TG concentrations and other clinical grounds. Finally, of the remaining 452 patients, 64 had no Lp(a) data recorded in their chart, leaving 388 patients in the cohort. This study was approved by the Ethics Review Board of St. Paul’s Hospital.
|
|
data collection
Demographic, clinical, and laboratory data were collected from the patient charts. Information on patient age, anthropometrics, cardiac risk factors, laboratory results, and hard cardiovascular outcomes were acquired on the 388-patient cohort. Because data acquisition was limited to the patient records, no information after the most recent visit was obtained. To examine the potential for selection bias, demographic, laboratory, and clinical data were also extracted on the 64 patients for whom no Lp(a) data were available.
demographics, diagnostic category, and anthropometrics
Patient date of birth, sex, presence of arcus cornealis, and presence of tendon xanthoma were noted. Additionally, clinical diagnostic categorization was recorded as possible, probable, or definite HFH.
clinical cardiac risk factors
Body mass index (BMI) on the most recent visit, alcohol use, and information on smoking and hypertension status were also obtained. Smoking history was determined from the patient record and was categorized dichotomously as ever smoked or never smoked. Presence of hypertension was defined as mention of the diagnosis in the chart or the prescription of antihypertensives for the purpose of lowering blood pressure (as opposed to, for example, the prescription of angiotensin-converting enzyme inhibitors to patients after myocardial infarction).
biochemical cardiac risk factors
The earliest untreated lipid profile was recorded: total cholesterol (TC), HDL-cholesterol (HDL-C), TGs, ratio of TC to HDL-C (TC/HDL-C), and calculated LDL-C (using the Friedewald equation). In the event that there was no untreated lipid profile available (13.7% of patients), the worst lipid profile was taken as a surrogate, assuming that this represented the lipid profile when medication compliance was poorest.
LP(A) measurements
All Lp(a) measurements were performed with the Mercodia 2-site IRMA. Because Lp(a) was usually measured after referral to our clinic, patients were rarely off all treatment when it was measured. However, because Lp(a) is fairly resistant to lipid-lowering therapies, this is unlikely to substantially affect the quality of the data. Accordingly, the mean of all Lp(a) determinations for each patient was collected under the assumption that this best represented a patient’s exposure to this atherogenic lipoprotein.
cvd endpoints
CVD endpoints were defined as history of myocardial infarction, coronary or peripheral arterial angioplasty with or without stent placement, coronary artery bypass graft, peripheral arterial bypass, repair of abdominal aortic aneurysm, carotid endarterectomy, and stroke. Indications of CVD that were qualitative or were vulnerable to interpretive subjectivity, such as presence of carotid plaques on ultrasound, history of chest pain, positive exercise stress tests, or history of transient ischemic attack, were not considered endpoints. The nature and year of the first CVD endpoint were acquired from the chart based on patient clinical history or procedural reports.
statistical analysis
We used a Cox proportional hazard regression model in SPSS (Ver. 11.5) to assess the association of risk factors to hard cardiovascular outcomes. For the purpose of analysis, Lp(a) results were divided into low (<800 units/L) and high (800 units/L). The cutpoint of 800 units/L represented the 75th percentile of Lp(a) for the HFH cohort at our clinic in a previous unpublished pilot project. The multivariate regression model was confined to the variables considered to be of greatest interest and importance: sex, smoking status, hypertensive status, LDL-C, TC/HDL-C, and Lp(a) concentration. TC/HDL-C was used in lieu of TC and HDL-C separately because this ratio is used in the Canadian guidelines for the management of dyslipidemia (20). Information on alcohol consumption, BMI, and the presence of arcus cornealis was used for descriptive analysis of the cohort and for comparison with the 64 patients for whom no Lp(a) data were available to ensure that no selection biases with respect to other potential cardiac risk factors were present. Time zero for purposes of the multivariate analysis was the year of birth, and follow-up terminated on the year of the first hard cardiovascular endpoint. Patients were censored at the time of death, after their last visit to the lipid clinic, or at the age of 80 years, whichever occurred first. The clinical and laboratory characteristics of the 388 HFH patients with Lp(a) data were compared with the characteristics of the 64 patients for whom there were no Lp(a) data.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
. Significant differences between these 2 groups were confined to age, distribution among the 3 diagnostic categories, and the extent of alcohol use.
|
The Lp(a) distribution of the cohort showed positive skewing. The median (interquartile range) was 399.0 (168.3–915.0) units/L, and there was no significant difference in the median Lp(a) concentration between men and women (P = 0.82, Mann–Whitney test). The Lp(a) distribution statistics are summarized in Table 4
. There were 278 patients whose Lp(a) concentration was categorized as low (<800 units/L) and 110 as high (800 units/L). Of these, 32 patients (11.5%) in the low group had a first CVD event in the follow-up period compared with 29 (26.4%) in the high group. The mean age of first CVD event was 50.0 years (range, 23–84 years) in the cohort as a whole: 51.4 years in the low Lp(a) group (range, 23–84 years) and 48.4 years (range, 32–70 years) in the high Lp(a) group (P = 0.36). By sex, the mean (range) age of first CVD event was 49.0 (29–78) years among men and 51.8 (23–84) years among women (P = 0.42).
|
In the multivariate regression model, only sex, Lp(a), and TC/HDL-C ratio were significant predictors of CVD events with hazard ratios (HRs) of 3.19 [95% confidence interval (CI), 1.79–5.69; P <0.001], 2.59 (1.53–4.39; P <0.001), and 1.18 (1.07–1.30; P = 0.001), respectively (Table 5
). Neither history of smoking nor hypertension was a significant predictor of CVD outcomes (both P 0.10), but each produced an HR consistent with increased risk of CVD [HR = 1.55 (0.92–2.61) and 1.57 (0.90–2.74), respectively]. Kaplan–Meier survival curves were constructed for the 2 statistically significant dichotomous variables, Lp(a) concentration and sex (Fig. 1
). Separation of survival curves based on these 2 variables was significant in each case (P <0.001, log-rank test). There were no significant differences in event-free survival based on diagnostic category of HFH.
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Larger studies (10)(11)(14)(15) with cohorts ranging from 262(15) to 526(10) found no correlation between Lp(a) concentration and CVD in the general HFH population. However, Jansen et al.(7) recently published a very large retrospective analysis of classic risk factors in HFH with data from 2400 Dutch HFH patients. In that study, Lp(a) data were available on 1698 patients, and the relative risk of CVD for Lp(a) >300 mg/L (430 units/L) was 1.50 (95% CI, 1.20–1.79).
The results of the present study confirm the value of Lp(a) as a predictive factor for CVD in the HFH population. Although the HR associated with high Lp(a) was larger in our study than in that of Jansen et al. (7), 2.59 vs 1.50, the difference is almost certainly attributable to the higher cut point of 800 units/L (560 mg/L) used in our study compared with 430 units/L (300 mg/L) used in the Dutch study. Although we are not aware of studies other than those described that evaluated Lp(a) as a risk factor for CVD in HFH cohorts per se, a large study of 1043 Dutch children from HFH kindreds determined that children affected by HFH whose Lp(a) was >300 mg/L (430 units/L) had a 1.45-fold higher incidence (95% CI, 0.99–2.13) of having a parent with HFH suffering from premature CVD(22). Thus, in addition to the present study, the 2 largest studies(7)(22) evaluating the value of Lp(a) as a predictor of CVD risk in HFH also affirmed its utility.
With respect to sex, our study confirmed the findings of other similar studies demonstrating that, apart from age, male sex is the most important predictor of CVD risk. In our cohort, the HR for male sex was higher than the HR observed by Jansen et al. (7), 3.19 vs 2.82, but of comparable magnitude. Authors of other studies have reported similar findings with respect to sex-associated risk(4)(5)(13)(14)(15).
The TC/HDL-C ratio was not analyzed as a risk predictor in most other studies because the 2 factors were evaluated as separate variables, with HDL-C generally performing well as a predictor of CVD risk. However, in one case–control study of 66 patients (33 with CAD and 33 without CAD), the TC/HDL-C ratio was found to be useful as a CAD risk factor (odds ratio for TC/HDL-C <6.5 = 0.28; 95% CI, 0.1–0.8; P = 0.016) (23). The utility of TC/HDL-C ratio in this context has not yet been fully explored.
Although smoking and hypertension failed to reach the conventional standard for statistical significance, both point estimates are consistent with increased risk of CVD. Smoking has been found to be predictive of CVD by others (7)(14)(15)(24), but this has not been demonstrated by all studies. Some reported no effect(11) or sex-specific effects(25). Hypertension has also previously been shown to be predictive of CVD in the HFH population(7)(10), but this finding is, again, not universal(11)(15).
LDL-C failed to predict CVD endpoints, perhaps contrary to clinical intuition. However, the same has been observed in other studies (7)(10)(25). We agree with Jansen et al.(7) that the explanation likely lies in the fact that all patients in an HFH cohort have high LDL-C by definition and are therefore at high risk for CVD. To statistically resolve the effect of LDL-C on CVD risk within this group is therefore difficult without a highly powered study. Furthermore, it has been our clinical observation that HFH patients with very high LDL-C are treated more aggressively, which may further confound matters.
Alcohol intake, the presence or absence of arcus cornealis, and BMI were not analyzed as risk factors for CVD in the multivariate model, as mentioned. Descriptive results (Table 3
) are presented for interest and reference.
Our study is strengthened by the fact that the HFH population in our clinic is ethnically heterogeneous: 55% were non-Dutch Europeans, 20% were Asian, 10% were French Canadian, 5% were Dutch, 3% were Scandinavian, 2% were African, and 5% were other. Our results are therefore applicable to the multiethnic HFH populations seen in North American clinics. This contrasts with some previous studies, which focused on genetically (11)(13) or ethnically(7) homogeneous HFH populations. In addition, we chose relatively objective CVD endpoints, electing to reject a history of angina, positive exercise and nuclear medicine studies, or transient ischemic attacks as endpoints. Although this makes statistically significant differences more difficult to demonstrate, it lends credence to the results.
There are several limitations to this study from the perspective of patient selection. One limitation is that the cohort is derived from patients who had been referred to a lipid clinic. This biases the sample toward those with more severe dyslipidemia and/or established CVD, which means that the cohort may not represent the general HFH population. Similarly, our failure to include patients who had been discharged previously from the clinic (typically because their lipid concentrations were well controlled) also biased the cohort toward more severely affected patients. Conversely, our failure to include data from deceased former patients biased the sample toward less severely affected individuals. Finally, the 64 patients without Lp(a) data could obviously not be included in the main cohort. The question arises as to whether this might affect results. Statistically significant differences in this group and the main cohort were confined to age, diagnostic categorization, and alcohol consumption (Table 3
); although the origin of these differences is unclear, it is unlikely that they constitute a bias that would affect results for the following reasons. First, the age difference in the groups merely indicates that if the 64 patients had been included in the study, the mean follow-up period would have been shorter. There is no reason to believe that this would affect the manner in which Lp(a), sex, TC/HDL-C, hypertension, and smoking were related to CVD risk. Second, although alcohol intake was not included in the multivariate model, by univariate modeling it was not a significant predictor of CVD risk (data not shown). Furthermore, to our knowledge, alcohol intake has never been shown to be an independent predictor of CVD risk in HFH. Third, we found no differences in survival based on HFH diagnostic categorization; therefore, this difference is not likely to affect results. Finally, the proportion of patients with no Lp(a) data was comparatively small (64 patients), and therefore, the omission of these patients is less likely to be important. In a study of this design, these weaknesses are unavoidable.
Because inclusion was based on clinical criteria instead of genetic analysis, there are likely some individuals in our cohort who do not have an LDL receptor mutation. For example, among those classified as definite HFH, there could potentially be a proportion, likely 1% (26)(27), with familial defective apolipoprotein B100, a condition almost clinically indistinguishable from HFH(28). Additionally, among those classified as probable or possible HFH, there may be some patients with familial combined hyperlipidemia (FCH), as "combined" is not always an accurate descriptor of this condition, in which increased TGs are not always observed. Results must be interpreted with these factors in mind. However, because most North American lipid referral centers apply clinical and not genetic criteria for diagnosis of HFH, the results are nevertheless of practical utility.
In conclusion, we have performed a retrospective cohort study to evaluate Lp(a) as an independent risk factor for hard CVD outcomes in an ethnically heterogeneous HFH population. In our cohort, male sex, Lp(a), and TC/HDL-C ratio were significant independent risk factors for CVD in decreasing order of importance. Smoking and hypertension were associated with increased HRs but failed to reach statistical significance. LDL-C was not useful in stratifying risk of CVD. We conclude that Lp(a) is a clinically important independent predictor of CVD in ethnically heterogeneous HFH populations. Thus, we recommend that Lp(a) be assessed at the time of presentation and considered along with other CVD risk factors in the management of patients with HFH.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  J. Staples, P. Taylor, A. Magil, J. Frohlich, S. M. Johnston, M. Koschinsky, C. Chan-Yan, and A. Levin Progressive kidney disease in three sisters with elevated lipoprotein(a) Nephrol. Dial. Transplant., May 1, 2008; 23(5): 1756 - 1759. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
