Review of Lipid and Lipoprotein(a) Abnormalities in Childhood Arterial Ischemic Stroke

Abstract

National organizations recommend cholesterol screening in children to prevent vascular disease in adulthood. There are currently no recommendations for cholesterol and lipoprotein (a) testing in children who experience an arterial ischemic stroke. While dyslipidemia and elevated lipoprotein (a) are associated with ischemic stroke in adults, the role of atherosclerotic risk factors in childhood arterial ischemic stroke is not known. A review of the literature was performed from 1966- 4/2012 to evaluate the association between childhood arterial ischemic stroke and dyslipidemia or elevated lipoprotein (a). Of 239 citations, there were 16 original observational studies in children (with or without neonates,) with imaging-confirmed arterial ischemic stroke and data on cholesterol or lipoprotein (a) values. Three pairs of studies reported overlapping subjects, and 2 were eliminated. Among 14 studies, there was data on cholesterol in 7 and lipoprotein (a) in 10. After stroke, testing was performed at > 3 months in 9 studies, at ≤3 months in 4 studies and not specified in one study. There were 5 case-control studies: 4 compared elevated lipoprotein (a) and one compared abnormal cholesterol in children with arterial ischemic stroke to controls. A consistent positive association between elevated lipoprotein (a) and stroke was found [Mantel-Haenszel OR 4.24 (2.94–6.11)]. There was no association in one study on total cholesterol, and a positive association in one study on triglycerides. The literature suggests that elevated lipoprotein (a) may be more likely in children with arterial ischemic stroke than in control children. The absence of confirmatory study on dyslipidemia should be addressed with future research.

Introduction

Stroke is a leading cause of childhood disability. Steno-occlusive cerebral arteriopathy is identified in approximately half of children with an arterial ischemic stroke (AIS), the pathophysiology of which is not well understood, but may be associated with infection.(1–4) The majority of acquired cerebral arteriopathy in the adult is attributed to atherosclerotic disease. Atherosclerosis-related risk factors were documented in only 2% of the International Pediatric Stroke Study registry (2003–2007) for childhood AIS but that data was not systematically collected as it was for other risk factors.(5)

Childhood dyslipidemia is a cause of adult atherosclerotic disease. The prospective childhood cohort in Muscatine, Iowa, found that adults with a history of high total cholesterol in childhood were 40–50% more likely to have upper quartile carotid intimal-medial thickness, a marker of atherosclerotic disease.(6) Some atherosclerosis was found in almost all children at autopsy, where death was primary due to trauma, and atherosclerosis was positively associated with non-high-density lipoprotein cholesterol (HDL-c) and negatively associated with HDL cholesterol.(7, 8)

Atherosclerosis is the accumulation of esterified cholesterol within smooth muscle cell.(9) Characteristics of an atherosclerotic plaque are proliferation of smooth muscle cells and intracellular lipid inclusion.(10) In animal study atherosclerosis can be slowly induced with a high fat diet. A high fat diet in the presence of inflammation results in plaque with fatty and proliferative components, typical histologic features of atherosclerosis.(11) The inflammatory state speeds the process of atherosclerosis and leads to diffuse large and small vessel involvement.(12, 13)

On a cellular level, low-density lipoprotein cholesterol (LDL-c) promotes normal cellular division until a threshold dose after which increasing concentrations are cytotoxic. The same effect is not seen with increasing concentrations of HDL-c where a cellular proliferative state continues to a stable threshold.(14) Intracellular LDL-c is tightly regulated by LDL receptor expression, and transmembranous diffusion even at high extracellular concentrations is low. Maximal function of the LDL receptor is seen at interstitial LDL-c levels of 25mg/dL. A higher interstitial level will cause down regulation of receptor expression.(9) Typical human LDL-c interstitial levels are approximately five fold higher than is required by the cell, and the highest of mammalian species, but a normal LDL receptor will continue to optimize intracellular concentrations.(15) When instead LDL-c is in a cationized state, as can occur with inflammation, LDL-c will diffuse freely across the cell membrane.(9)

High Lp(a) is associated with AIS in children and adults.(16, 17) Lp(a) is attached to LDL-c through a disulfide bond, it has athero- and thrombogenic roles and structural similarity with plasminogen.(18–20) Heritability is an important determinant of Lp(a) and it may be more likely elevated in children from families with stroke compared with children from families where there is no history of stroke. (21, 22) Lp(a) has been associated with stroke recurrence in children with AIS.(23) Lp(a) may even be associated dyslipidemia as was found in a sample of Spanish children.(22)

Normative values for cholesterol in childhood are based on population distributions that have evolved historically. Internationally the rate of increase in overweight and obese children was unprecedented at the turn of this century and prevalence remains at historic high levels.(24–26) Childhood obesity correlates with dyslipidemia.(27) Among adolescents and young adults admitted with stroke in the US Nationwide Inpatient Sample, there was an approximate 4-times increase in the prevalence of lipid disorders between 1995 and 2008.(28) Other US data suggests a trend for increasing incidence of childhood and young adult stroke.(28, 29)

Longstanding atherosclerotic disease as is seen in adults is unlikely to be present in childhood, and without pathologic data, we propose that the pathology of some steno-occlusive arteriopathy in childhood AIS may be atherosclerotic. The aim of this review is to summarize the literature that suggests or supports a role for dyslipidemia and Lp(a) as risk factors or associated factors in childhood AIS.

Methods

Search strategy

An electronic search was performed using Medline via PubMed, OVID, Web of Science and The Cochrane Library, for publications in the English language from 1966 through April 30th 2012, matching a combination of “childhood” or “pediatric“ with “lipid,” or “cholesterol”, “dyslipidemia”, or “lipoprotein a” with the term ”stroke.” Subject headings and abstracts were reviewed by a single neurologist (SS), for eligibility of data. If potentially appropriate, the study was reviewed in full including the reference list. Of the studies discovered on reference lists, only one additional study contained relevant data.

Selection criteria

Original observational studies in children (n>1), a distinguishable subgroup of children or children with neonates where there was imaging-confirmed AIS, were included if the study contained subject-level or sample distribution values for a cholesterol parameter: total cholesterol (TC), triglycerides (TG), HDL-c, LDL-c or (Lp(a). Studies exclusively on neonates were excluded. Early and delayed testing protocols were included.

Five studies that reported performance of lipid testing in children with ischemic stroke were not included. In four studies, the authors commented that there were no lipid abnormalities, but individual or aggregate values were not available for the full sample. (30–34) In the fifth study, modern radiologic evaluation for stroke was not available and the diagnosis of ischemic stroke was not confirmed.(35)

Data extraction

The following data was extracted from relevant studies: country of origin, enrollment dates, year of publication, inclusion and exclusion criteria, study design, number of subjects, age at stroke and lipid or Lp(a) testing, interval to testing, individual or distribution estimates for TC, LDL-c, HDL-c, TG, Lp(a), critical abnormal values, and where available, association estimates with stroke.

Analysis

Review Manager (vs. 5.2, Cochrane IMS) was used to calculate the Mantel-Haenszel odds ratio (MH OR), the Breslow-Day test statistic and to generate the forest plot (Figure 1). The summary MH OR is an estimate of the association of Lp(a) with childhood AIS. The Breslow-Day test of heterogeneity tests the null hypothesis that there is homogeneity of ORs across studies. Estimates for cholesterol data are discussed in relation to National Cholesterol Education Program (NCEP) and American Heart Association (AHA) recommendations for children, where TC should be <170mg/dL, LDL-c <110 mg/dL, and HDL-c >35 mg/dL.(36) There is no standard for a healthy triglyceride level in children, but levels >200mg/dL are associated with obesity.(37) A critical value for Lp(a) of >30mg/dL was accepted based on risk threshold study on childhood venous thromboembolism.(38) International reference for prevalent childhood dyslipidemia (≥1 lipid abnormality) varies widely and has been estimated at ~20% among 12–19 year olds in the US.(27, 39–41) By cholesterol parameter in this NHANES sample of 12–19 year old children (1999–2006): LDL-c ≥130 mg/dL was seen in 7.6%; HDL-c ≤35 mg/dL was seen in 7.6%; and triglycerides ≥150 mg/dL was seen in 10.2%.(41)

Figure 1.

Association betweenabnormal Lp(a) and AIS

Data reporting

Tables with median and range of lipid values and their association with stroke are used to present the data from 14 studies. These estimates are as reported in the primary study except in one study where individual-level data was available and an estimate for subjects with ischemic stroke was recalculated with exclusion of subjects with hemorrhagic stroke.(42).

Results

An electronic topic search yielded 239 citations. Of these, 15 studies plus one discovered through a reference list contained relevant data. Three pairs of studies were reviewed for overlapping subjects. In the first set, the primary author was contacted and overlap was estimated at less than one third. Both studies are reported with a notation about the overlap in Table 4.(43, 44) A second set consisted of a larger study and subgroup analysis.(45, 46) The larger study was reported.(46) The final set of studies were published in the same year, from the same University hospital, and enrollment dates were not available. The subject number differed by three and there was overlapping authorship.(47, 48) Without a response from the primary authors, the smaller study with fewer estimates was not reported.(47)

Table 4.

Summary Lipoprotein(a)

Author	Country	Enrollment dates	Inclusion (N=neonate, C=Child)	Excluded	Number of subjects w/AIS	Number of subjects tested	Age (yrs) at stroke, median, range	Delayed testing (months)	Lp(a) > 30 mg/dL (%)	Lp(a) (mg/dL), median, range	Study design/comment
Nowak-Göttl U., et al.(72)	Germany	1992–1996	N+C	-	36	36	birth-18	acute and >6	22 (Lp(a)>50mg/dL on repeated testing)	15 (0–145)	case series (prevalence of risk factors)
Nowak-Göttl U., et al.(71)	Germany	10/1995–10/1998	C	Cardiac, infectious, collagen vascular and metabolic disorders, cerebral vasculopathy	148	148	4.5 (0.5–16)	1.5–3	26.4*		case control (association of Lp(a) with stroke)
Sträter R, et al.(70)	Germany	10/1995–10/1998	N+C	Non-cardiac disease	38	38	neonate-18	NA	18.4		case control (association of Lp(a) with stroke)
Nowak-Göttl U., et al.(44)	Germany	7/1996–8/2006	N+C	-	282	282	4 (0.1–18)(age at blood collection)	6–12	25.5*	14 (0–168)	cross sectional (correlation in families)
Ganesan V., et al.(1)	UK	1978–2000	N+C	congenital hemiplegia, and silent infarction	212	41	5 (21 days-19.7)	acute	22		retrospective, comparison of risk factors in stroke subgroups
Beheiri A., et al.(50)	Germany	01/2002 and 12/2004	N+C	cerebral vascular disease	103	103	4 (1–15) (age at blood collection)	12–18	29.1		case control (association of Lp(a) with stroke)
Teber S, et al.(51)	Turkey	1/1992–2/2006	N+C	-	52	52	3.2 +/− 4.4 (mean, SD) ; range(0–15)	3–6	26.9	11.9 (1.9–140)	case control (association of Lp(a) with stroke)
Lynch JK, et al.(73)	US	1990–2000	N+C	Cardiac, chromosomal, metabolic or hematologic disease and history of cancer, trauma, or CNS infection	59	59	0.67 (1 day-12)	> 3	20.3		case series (prevalence of risk factors and comparison with published rates)
Simma B, et al.(42)	Austria	1984–2005	C	porencephaly, head injury	15	15	6 (0.9–14)	4–216	40		retrospective, descriptive
Barreirinho S, et al.(74)	Portugal	1987–1999	C	-	21	21	4.5 (0.17–13)	acute	28.6		prevalence of Lp(a)(case control for other inherited risk factors)

^*overlapping subjects

Fourteen studies (n= 1,071) are reported in the tables. One study compared TC and TG in children with a history of AIS to controls (Table 1). This study (n_AIS=75), found no association between total cholesterol and AIS [OR 1.18 (0.81–1.72)] and a positive association between triglycerides and AIS [OR (2.27–16.20)].(48) Four studies compared Lp(a) values in children with a history of AIS to controls (Figure 1). Each study found a significant positive association between Lp(a) and AIS. A summary MH OR was [4.24 (2.94–6.11)] and study ORs were homogeneous [Breslow-Day Chi-Square 4.59, p= 0.2040].

Table 1.

Odds Ratio for Cholesterol and Triglyceride

Author	Subjects (nabnormal/ntotal)	Controls (nabnormal/ntotal)	Delayed testing	TC: OR, 95%CI, p value	TG: OR, 95%CI, p value
Kopyta I. et al.(48)	NA/75	NA/71	NA	1.18 (0.81–1.72), 0.39	6.06 (2.27–16.20), <0.001

NA=not available

Among 14 studies, 7 reported data on cholesterol and 10 reported data on Lp(a). Sample frequency data is reported in Tables 2, 3 and 4. The range of study-specific average values in mg/dL for cholesterol parameters: TC (medians 155–166, and a mean of 172), LDL-c (medians 85–101, and mean of 95], HDL-c (medians 47–60, and a mean of 55) and TG (medians 64–88, and a mean of 112), are comparable with NCEP and AHA recommendations.(Table 2)

Table 2.

Summary cholesterol parameters

Author	Country	Enrollment period	Included (N=neonate, C=Child)	Excluded	Number of subjects w/AIS	Number of subjects tested	Age (yrs) at stroke, median	Age range at time of stroke (yrs)	Age range at time of testing (yrs)	Delayed testing (months)	Fasted	TC (mg/dL) median (range)	LDL-C (mg/dL) median (range)	HDL-C (mg/dL) median (range)	TG (mg/dL) median (range)	Study design/comment
Normal reference values (NCEP and AHA)												<170	<110	>35	<200*
Kopyta I., et al.(48)	Poland	NA (published 2010)	C	-	75	75	8.2	0.75–17	0.8–18	NA	yes	171.7 (32.9) mean SD)	94.7 (34.8) mean (SD)	54.5 (27.8) mean (SD)	111.6 (32.8) mean (SD)	case control (association of lipids with stroke)
Normann S., et al.(43)	Germany	1/2004–1/2009	N+C	-	72	72	5	term-18	0.5–18	6–12	yes	155 (110–336)	-	-	-	case control (association of lipids with testosterone
Nowak-Göttl U., et al.(44)	Germany	7/1996–8/2006	N+C	-	282	282	4 (age at testing)	NA	0.1–18	6–12	yes	161 (91–249)	85 (24–177)	55 (25–104)	-	cross sectional (correlation in families)
Abram HS., et al.(49)	US	1973–1991	C	head trauma, cardiac, hematologic & metabolic disorders	42	TC 41; LDL 37; HDL 38; TG 41	6	0.75–18	2.0–33	12–234	yes (excl. 5)	166 (116–182)	101 (60–121)	47 (34–88)	64 (26–139)	case series (risk factors for poor outcome)
Glueck CJ., et al.(46)	US	1968–1982	C	Identified cause of stroke	11	11	7	1.25–17	1.33–21	>6	yes	158 (120–204)	97 (40–123)	60 (46–63)	88 (60–121)	cross sectional (correlation in families)

NA=not available

^*no recommended value (<200 from obesity literature)

Table 3.

Proportion abnormal by cholesterol parameter

Author	Country	Excluded	TC, ratio (%)	LDL-C ratio (%)	HDL-C ratio (%)	TG ratio (%)	Critical values	Delayed testing (months)	Fasting	Study design/comment
Reference values NHANES (12–19 years 1999–2006)(41)				7.6% (≥130 mg/dL)	7.6% (≥35 mg/dL)	10.2% (≥150 mg/dL)
Abram HS., et al.(49)	US	head trauma, cardiac, hematologic& metabolic disorders	10/41 (24)	7/37 (19)	7/38 (18)	11/41 (27)	age and sex %tile: >90th TG, TC, and LDL-C; < 10th HDL-C*	12–234	yes	case series, risk factors for poor outcome
Ganesan V., et al.(1)	UK	congenital hemiplegia & silent infarction	10/117(9)	-	-	36/117 (31)	TC >5.4mmol/L (>210mg/dL); TG >1.4micromol/L**	acute	no	retrospective, comparison of risk factors in stroke subgroups
Simma B, et al.(42)	Austria	porencephaly & head injury	borderline 5/15 (33), absolute 2/15 (15)	borderline 3/15 (20), absolute 2/15 (15)	3/15 (20)		“Age-related values” Ref.[Borderline TC: 170–199 mg/dL and LDL-c 110–129 mg/dL; High TC ≥200 mg/dL and LDL-c ≥130 mg/dL]	4.8–216	no	retrospective, descriptive

^*This is equivalent to the reference values used for 12–19 year olds in the NHANES analysis.

^**1.4 μmol/L is as reported in the original study however 1.4 mmol/L (=124.6mg/dL) is a common threshold value.

The range of prevalence (%) estimates for abnormal cholesterol parameters were: TC (9–24), LDL-c (15–20), HDL-c (18–20), and TG (27–30). If we compare these prevalence estimates to population estimates among 12–19 year olds in the US (NHANES) where the threshold for an abnormal value is comparable we find that the population prevalence of high LDL-c (7.6%) appears to be less than in two samples of children with AIS (15% and 19%)(42, 49); that the population prevalence of low HDL-c (7.6%) may be lower than in two samples of children with AIS (18% and 20%)(42, 49); and that the population prevalence of high TG (10.2%) may also be lower than one sample of children with AIS (27%)(49). In a second study with prevalence of abnormal TG, the critical value was lower at 1.4 mmol/L or 124.6mg/dL.(1)(Table 3) The median Lp(a) values (12–15mg/dL) were below threshold for abnormality (>30mg/dL) but the range of prevalent elevated Lp(a) was appreciable at 18–40%.(Table 4)

There were no studies with lipid testing prior to stroke. In 9 studies, testing for lipid levels was done at > 3 months after the stroke, in 4 studies testing was ≤3months and in one study information on interval to testing was not available. Of 7 studies with cholesterol values, 5(71%) were fasting samples.

Conclusions

This review of the literature confirms an association between Lp(a) and childhood AIS. A recent meta-analysis on thrombophilia in childhood stroke reported a similar pooled OR (6.53, 4.46–9.55).(17) The difference in the pooled ORs between this study and the meta-analysis is due to two additional studies included here(50, 51) and the non-inclusion of two studies used in the meta-analysis because one reported only neonates and the second did not provide the control data in the primary manuscript.(44, 52)

A gap in the literature on dyslipidemia and childhood AIS is noteworthy and is the main limitation in this review. Only a single study compares cases with controls. Other studies report average lipid values and prevalence estimates for abnormal parameter but comparison of these values to external samples has limited validity. There are additional issues with aggregation of estimates from males and females and from a heterogeneous sample of childhood ages. The distribution of cholesterol lipids varies with age and by gender for age. There is a pre-adolescent peak in boys and girls with a subsequent decrease in boys from 12–16 years, not seen in girls this age and return to preadolescent values by late adolescence in boys with continued higher values in girls of the same age.(53, 54) This concern is not shared for Lp(a) values where there is considerably less variability by age and strong heritability.(21)

Aggregating values as proportion abnormal has the advantage of standardization by age and gender. This was done in 2 of 3 studies reporting the prevalence of an abnormal cholesterol parameter. The interpretation of these proportions still requires a reference value and it appeared that abnormal LDL-c, HDL-c and TG values were higher in samples of children with AIS than modern prevalence estimates from adolescents in the US, but this reference population is undoubtedly different from the AIS samples: US (1973–1991), UK (1978–2000) and Austria (1984–2005). Appropriate reference samples would now be difficult to locate.

Another potentially important limitation to the validity of the findings relates to the interval to testing from the time of stroke. Both cholesterol and Lp(a) values have been found to be influenced by an ischemic event. After an acute myocardial infarction, the immediate, 3 and 6 month values are similar to baseline, but in the interval between this time, and by the first hospital day, all lipid concentrations decrease by 15–25% of baseline values.(55–57) Lp(a) similarly drops by 10–25%, but unlike cholesterol, it can rebound on days 7-1 1 before returning to baseline values by 3 months.(58) This phenomenon has not been studied as extensively after cerebral infarction in adults and has not been studied at all after childhood ischemia. In one adult stroke cohort of Northern Manhattan, serum Lp(a) and other lipid levels obtained within 24 h of acute ischemic stroke did not significantly change at 4 weeks.(59)

The effect of some of these methodological concerns would be to bias the result towards a null effect. Among studies reporting average and prevalence abnormal cholesterol values, the interval from stroke to testing was >3 months in 5/7 (71%) studies. If lipids decrease after ischemia as is seen after a myocardial infarction these 2 studies report lower than actual values.

The direction of effect for other biases will depend on the specific study methods. A long interval to testing (>12months), noted in 2 studies may include values that are unrelated to lipid metabolism at the time of the stroke. The direction of effect would depend on the relative hormonal and metabolic change at the time of testing. A single estimate obtained by aggregating children with a range of ages will similarly be biased by the age make-up of the sample and the variable contributions of age-related hormonal and metabolic effects. It may be important to analyze childhood data by age categories. The nonlinear trend in cholesterol parameter distribution by age may be associated with a similar nonlinear estimate of risk by age. The expected pre-adolescent cholesterol peak in cholesterol overlaps with the peak in childhood AIS [5.7 years, interquartile range, 1.7–11.6 years], reported from the International Pediatric Stroke Study registry (2003–2007).(5)

Like age, certain subgroups of children with AIS having other risk factors may have a higher risk of stroke with dyslipidemia than other subgroups. In many of the studies presented here the authors excluded children with AIS who additionally had one of a list of other risk factors. The original desire for this may have been to study dyslipidemia as an independent risk factor, but as the literature on childhood AIS expands we are learning that multiple risk factors are the norm, such as a recent study showing an association between cardio-embolic stroke and recent infection.(60) If dyslipidemia is a risk factor only as it interacts with other risk factors, such as an infectious or inflammatory state, the selection of a sample of children with AIS and few or no other identified risk factors, will have no association with stroke.

The single case control study on cholesterol deserves particular mention because as this would otherwise be considered the best source of evidence there are a number of concerns in the study design where the effect of bias would be towards a null association. Control subjects were older and pre-pubertal (mean age 10.3 versus 8.2 in cases) and more likely female (69% versus 59% among cases). Both of these differential demographics of the control group would result in higher cholesterol in the control group. Unfortunately information on interval to testing was not available which further limits the value that can be placed on their findings.(48)

In December, 2011 the National Heart, Lung, and Blood Institute (NHLBI) with American Academy of Pediatrics (AAP) endorsement, recommended universal lipid screening for children at 9 to 11 years of age and again at 17 to 21 years of age.(61) The most recent, 2008 AHA guidelines on the management of stroke in infants and children does not discuss testing lipids. (62) In keeping with previous literature, our review suggests an association between high Lp(a) and childhood AIS. Niacin is effective at lowering Lp(a) levels,(63) but there is currently no data demonstrating that niacin reduces the risk of stroke.(64, 65) On the other hand, modification of dyslipidemia positively impacts secondary stroke prevention in adults.(66) Randomized controlled study on statin use in children with hyperlipidemia has demonstrated efficacy and safety.(67–69)

An ongoing prospective study, the Vascular Effects of Infection in Pediatric Stroke (VIPS) study, will test the association between infection, childhood AIS and arteriopathy.(4) Four of the studies presented here discuss a vascular abnormality (stenosis or occlusion) in their cohort but the data was not available to correlate dyslipidemia with arteriopathy.(1, 46, 49, 70) Further exploration of the role of dyslipidemia in childhood stroke could not only identify a modifiable risk factor but could shed light into the pathophysiology of arteriopathy. The diverse properties of statins may be especially beneficial if there is an athero-inflammatory process involved in childhood AIS with arteriopathy. The biologic plausibility and failure of the past literature to demonstrate no association are cause for future research on dyslipidemia in childhood AIS.

Future study should be designed with the limitations of these historic studies in mind. An appropriate reference group is necessary to estimate the effect size and there should be group or individual-level matching by age category and gender to limit confounding by these variables. In order to learn about how dyslipidemia might interact with other risk factors, a study design should include all childhood AIS with accounting for major risk factor categories in the demographics and analysis. A case-control design using an available large research cohort of childhood AIS with demographic and stroke risk factor information is ideally suited to provide evidence for an association. Still anticipated limitations to such a study are an absence of uniform lipid testing in an existing cohort and variability in the interval to testing. It will be necessary to compare the sub-cohort that is tested to the untested sub-cohort and account for biases in testing in the final analysis. The interval to testing should also be categorized (i.e. <24 hours, ≥24 hours-1 week, >1 week-3 months, >3months- 6months) and data analyzed by testing interval. Future prospective testing in a childhood cohort identified rapidly after stroke with testing at (i.e. <24 hours; 24 hours, 7 days; 14 days; 3 months and 6 months) and prospective follow-up for stroke recurrence would be positioned to estimate the association between dyslipidemia and AIS, to assess risk of a recurrence stroke due to dyslipidemia and to address the influence of a stroke on lipid values in children.