Introduction
Lipoprotein(a) (Lp(a)) is a plasma lipoprotein that combines a low-density lipoprotein cholesterol (LDL-C)-like particle with an apolipoprotein B100 (ApoB100) that is covalently bound to an apolipoprotein(a) (ApoA1) [1]. The prothrombotic, atherogenic, and proinflammatory properties of Lp(a) may contribute to early development and progression of atherosclerotic plaques in childhood, which progress over time. This transition is critical because the clinical impact of atherosclerosis manifests only in adulthood, thus establishing Lp(a) as a causal and independent risk factor for early-onset atherosclerotic cardiovascular disease (CVD)-related events in adulthood, namely ischemic heart disease, stroke, and peripheral vascular disease [2-7].
Serum Lp(a) levels are mostly genetically determined (approximately 90%) and have been positively correlated with a family history of CVD [8-11]. There is evidence that these levels stabilize at approximately the age of 5 and tend to persist into adulthood [7,12-14]. Although evidence suggests the early stabilization of Lp(a) levels, some studies indicate they may continue to change throughout childhood, influenced by genetic and environmental factors, such as epigenetics. A Dutch study revealed significant intraindividual variability in children referred to a pediatric lipid clinic and followed up to the age of 20, with Lp(a) levels showing a 70% fluctuation between measurements and a tendency to increase with age until the end of this follow-up period. However, this study measured Lp(a) in mg/L (Lp(a) mass) rather than the recommended nmol/L (number of particles), which limits the ability to accurately quantify particle number and, as acknowledged by the authors, may have yielded different results than the latter quantification method [15], highlighting the need for unit standardization. This technique-associated variability may undermine the reliability of a single lifetime quantification, as endorsed by European guidelines [16,17], and the estimation of CVD risk, potentially compromising future diagnosis and treatment perspectives.
Hence, to identify and manage childhood dyslipidemia and mitigate early-onset CVD, the American expert guidelines for Cardiovascular Health and Risk Reduction in Children and Adolescents recommend lipid screening in children aged 9 to 11 years, followed by a second screening at 17 to 21 years old; however this screening does not include Lp(a) measurement unless in children with previous hemorrhagic or ischemic stroke, or a family history of premature CVD [18]. The European Society of Cardiology also recommends Lp(a) screening in youth meeting the same criteria and in cases with family history of elevated Lp(a) and no other risk factors [17]. This proactive approach acknowledges the importance of early risk factor identification, particularly in the context of CVD—which remains the leading cause of mortality worldwide—and recognizes that atherosclerosis is a main underlying factor in the development of the disease [19,20].
In Portugal, death associated with atherosclerosis was 14.3% in 2016, representing a significant clinical and economic burden [21-24]. The progression and severity of atherosclerosis relate to the presence and persistence of several modifiable and non-modifiable CVD risk factors, including high levels of LDL-C, Lp(a)—which is 6 times more atherogenic than LDL-C—and other forms of dyslipidemia [2,4,7,25]. These factors tend to cluster and reinforce each other, increasing disease morbidity and mortality. It is widely considered that Lp(a) levels are minimally influenced by lifestyle factors or classical lipid-lowering therapies, and in the absence of targeted therapies, managing the other CVD risk factors is of utmost importance. Therefore, recognizing individual risk factors during childhood, including Lp(a), can lead to the implementation of effective strategies for CVD prevention later in life [4,18].
Despite its critical relevance and the significant incidence of Portuguese adults with high Lp(a) levels, data concerning the prevalence of elevated Lp(a) levels in the Portuguese pediatric population remains scarce. A comprehensive characterization of Lp(a) levels within this population and their correlation with CVD risk profile is essential to identify potential risk factors that may help clinicians develop effective preventive strategies or delay the onset of CVD in adulthood. Thus, this study aimed to investigate the clinical significance of Lp(a) levels in a Portuguese pediatric sample that underwent serum Lp(a) testing as part of lipid disorder screening – prompted by obesity, hypercholesterolemia, and/or family history of premature CVD–spanning from childhood to adolescence. Moreover, it sought to characterize the CVD-associated risk factors, evaluate the correlation of Lp(a) levels with the patient’s lipid profile, and assess which CVD risk factors were associated with elevated Lp(a) levels in this population.
Methods
1. Study design and study participants
This was a cross-sectional, retrospective, single-center study involving pediatric patients (<18 years) followed at the Pediatric Department of the tertiary university Centro Materno-Infantil do Norte Dr. Albino Aroso (Centro Hospitalar Universitário Santo António, Porto) in Portugal. The study enrolled patients referred to the Clinical Nutrition Unit of the same department who underwent serum Lp(a) testing due to lipid disorder screening. These patients were evaluated due to conditions known to increase CVD risk, such as obesity, eating disorders, dyslipidemia, kidney or cardiac transplants, nephrotic syndrome, family history of premature CVD or dyslipidemia. No exclusion criteria were applied in this study.
This study complies with the Declaration of Helsinki and was approved by the local Ethics Committee – Comissão de Ética do Centro Hospitalar Universitário Santo António/Instituto Ciências Biomédicas Abel Salazar (N/ REF.ª 2022.197 (157-DEFI/158-CE).
2. Data collection and outcome measures
Data were retrospectively collected from the hospital medical records between August 2018 and June 2022. It involved collecting demographic (sex and age) and clinical data, including body mass index (BMI), medical history (as defined in Supplementary Table 1), and laboratory values such as Lp(a), total cholesterol, high-density lipoprotein cholesterol (HDL-C), non-HDL-C, calculated LDL-C, triglycerides, apolipoprotein B (ApoB), Apo A1, high-sensitivity C-reactive protein (hs-CRP), homeostatic model assessment for insulin resistance (HOMA-IR), fasting glucose and insulin.
Serum Lp(a) was measured by an immunoturbidimetric assay, through the Roche Cobas Integra 400 plus chemistry analyzer (Roche Diagnosis, Swiss) as per manufacturer instructions. This assay applies a monoclonal antibody against an unrepeated epitope of Lp(a), determining the concentration of Lp(a) particles. The reference reagent, used for calibration, was the IFCC SRM 2B. All measurements were carried out by a single laboratory, at Laboratório de Química Analítica Unidade Local de Saúde de Santo António, to reduce bias.
Lp(a) group classes were defined following the recommendations of the 2022 Atherosclerosis Society consensus statement, illustrating a pragmatic approach for cardiovascular risk assessment: <75 nmol/L, 75–125 nmol/L, >125 nmol/L. The age groups were defined as 0–4 years, 5–8 years, 9–11 years, and 12–18 years. This approach was based on previous evidence that full LPA gene expression occurs in the first years of life [15,26], that Lp(a) reaches adult levels by the age of five [14], and on the recommendation for lipid screening in the 9–11 years old population [18].
3. Statistical analysis
The patient’s sociodemographic and clinical characteristics were characterized using descriptive analyses. This approach was performed within specific subgroups based on the age group (0–4 years, 5–8 years, 9–11 years, and 12–18 years) and the Lp(a) risk levels (<75 nmol/L, 75–125 nmol/L, and >125 nmol/L). Measures of central tendency, namely median, and of dispersion, such as quartiles (P25, P75), describe numerical variables. Absolute and relative frequencies characterize categorical variables.
The distribution of patients presenting each Lp(a) risk level was described between age groups and according to the presence or absence of some medical conditions (i.e., early-onset family history of CVD, dyslipidemia, family history of dyslipidemia, malnutrition, obesity, and low-grade inflammation). Trends through Lp(a) risk levels regarding other binary categorical variables were evaluated using the Cochran–Armitage test, while the association between 2 ordinal variables was assessed by performing the Kendall Tau test. To assess the correlation between Lp(a) levels and other quantitative variables, the Spearman coefficient was obtained.
A multivariate logistic regression was conducted to evaluate the association between some risk factors and the presence of Lp(a) values ≥75 nmol/L. The model was adjusted for the age group, sex, BMI (z score), family history of dyslipidemia, early-onset family history of CVD, levels of fasting glucose, ApoA1/ApoB, and levels of LDL-C. Several models were created by a stepwise method for variable selection. The best model was obtained by comparing the Akaike information criterion, which indicates the quality of the model in comparison to others. Afterwards, the best model was compared to the null model (i.e., only with the intercept) to further evaluate its quality. Odds ratios (OR) were calculated to quantify the eventual protective or risk effect of the independent variables regarding the presence of Lp(a) values ≥75 nmol/L.
As the missing data were completely at random, a complete case analysis was applied to all analyses. P values obtained from multiple comparisons were adjusted by Benjamini-Hochberg correction. The significance level was set at 0.05. Any missing values in each variable were excluded from the analysis.
All statistical analyses and graphs were performed using RStudio software version 4.2.2 (R Foundation for Statistical Computing, Austria).
Results
1. Patients' demographic and clinical characteristics
This study included 792 pediatric patients who attended a clinical nutrition appointment at a pediatric department. The sex distribution was well-balanced, with 56.9% female participants, and a median (interquartile range) age of 12.0 (9.0–14.3) years. The distribution of patients by age groups was as follows: 33 patients (4.2%) were 0–4 years old, 120 (15.2%) were 5–8 years old, 203 (25.6%) were 9–11 years old, and 436 (55.0%) were 12–18 years old. Five patients (0.6%) were less than 2 years old.
The demographic and clinical characteristics of patients by age group are summarized in Table 1. The distribution of early-onset family history of CVD (ranging from 19.6% to 26.7%) and family history of dyslipidemia (ranging from 39.3% to 53.4%) had discrete variations amongst all age groups, with the former reaching higher values in children aged 0–4 and the latter in children aged 5–8 years. Four patients had familial hypercholesterolemia, 2 of them in the 12–18 years old category. The frequency of patients with low-grade inflammation (ranging from 36.8% and 63.6%) was highest in older patients. On the opposite, obesity (ranging from 28.1% and 70.6%) and dyslipidemia (ranging from 12.2% and 15.2%) were more prevalent in children aged 0–4 and 5–8 years. Arterial hypertension was not prevalent in this sample regardless of the age group (<3.7%).
Table 1 also details the laboratory levels observed in each age group. In particular, median Lp(a) levels were 39.2 (12.4–116.3) nmol/L in children aged 0–4 years, 23.2 (11.8–60.6) nmol/L in the 5–8 year olds, 29.3 (13.3–125.1) nmol/L in the 9–11 year olds, and 36.6 (13.1–107.6) nmol/L in the 12–18 year olds, which were not statistically different between them (H=6.007, P=0.111) (Supplementary Fig. 1).
Concerning treatment for a lipidic disorder, only one child was prescribed statin. She was a 16-year-old female, with a BMI z score of 0.69, Lp(a) level of 1.1 nmol/L, record of dyslipidemia with family history of dyslipidemia, early-onset family history of CVD, and familial hypercholesterolemia.
2. Association of inflammation with HOMA-IR and BMI
The association between inflammation, as indicated by hs-CRP levels, and both HOMA-IR and BMI (z score) was evaluated. Individuals with low-grade inflammation had significantly higher HOMA-IR values compared to those without low-grade inflammation (3.860 [2.825–5.975] vs. 3.070 [2.027–5.050], P<0.001). Similarly, BMI (z score) was significantly higher in individuals with low-grade inflammation than in those without low-grade inflammation (2.120 [1.827–2.400] vs. 1.650 [0.460–2.070], P<0.001). This suggests that low-grade inflammation is associated with higher levels of insulin resistance (HOMA-IR) and BMI.
3. Lp(a) levels in pediatric patients
Fig. 1 shows the distribution of serum Lp(a) levels in the overall sample, with a median level of 31.80 [13.00; 101.40] nmol/L. According to Lp(a) levels, most children (69.3%) were at low risk for CVD (Lp(a) <75 nmol/L), while 171 (21.6%) fell into the high-risk category (Lp(a) >125 nmol/L). A total of 72 (9.1%) patients were in the grey zone and considered to be at an intermediate risk for CVD (Lp(a): 75–125 nmol/L). A small percentage of patients (n=7, 0.9%) had serum Lp(a) levels over 430 nmol/L.
No significant association between Lp(a) levels and age was found (R=0.029, P=0.413) (Supplementary Fig. 2). When evaluating the distribution of patients in each age group by Lp(a) class, the results also showed no statistically significant association between the 2 variables (τ=0.055, P=0.081) (Fig. 2). Of relevance was the similar frequency of children aged 0–4 years with Lp(a) levels >125 nmol/L (24.2%) to that of the oldest patients (over 22.9%).
The demographic and clinical characteristics of patients by Lp(a) CVD risk classes are summarized in Supplementary Table 2 and Supplementary Fig. 3. Sex and age were equally distributed across Lp(a) CVD risk classes, with a median age of 12 years. Higher lipid values, such as total cholesterol, non-HDL-C, calculated LDL-C and ApoB were generally associated with higher Lp(a) levels. In addition, a higher prevalence of early-onset family history of CVD, dyslipidemia and hypertension was observed in higher Lp(a) risk classes, while low-grade inflammation was more common in the intermediate risk class group. Still, the distribution of patients by each Lp(a) group considering their medical history was statistically homogeneous between patients with and without a given condition (Fig. 3). For instance, the distribution of patients by Lp(a) class was not significantly different between those with early-onset family history of CVD and those without (P=0.566). The same result applied to dyslipidemia (P=0.113), family history of dyslipidemia (P=0.519), malnutrition (P=0.299), obesity (P=0.205), and low-grade inflammation (P=0.235).
4. Correlation of Lp(a) with lipid profile
To investigate whether Lp(a) correlated with the lipid profile of pediatric patients, an analysis was performed comparing the levels of Lp(a) and of total cholesterol, non-HDL-C, LDL-C, triglycerides, and HDL-C (Supplementary Fig. 4). The results revealed a significant weak positive correlation between Lp(a) and total cholesterol (R=0.12, P<0.001), non-HDL-C (R=0.12, P<0.001), and LDL-C (R=0.15, P<0.001). On the opposite, there was no statistically significant correlation between Lp(a) and triglycerides (R=-0.034, P=0.35) and HDL-C (R=0.037, P=0.29).
Accordingly, the median levels of total cholesterol, non-HDL-C, and LDL-C were significantly higher in the Lp(a) class >125 nmol/L compared to the <75 nmol/L group (P<0.001). Conversely, these values were similar between the Lp(a) groups 75–125 and >125 nmol/L, and groups <75 and 75–125 nmol/L (Supplementary Table 2). Triglycerides and HDL-C did not significantly differ between any of the Lp(a) groups.
5. Predictive factors for elevated Lp(a) levels
Next, it was assessed which factors present in this sample were associated with a higher probability of having Lp(a) levels ≥75 nmol/L. The multivariate analysis was adjusted for age group, sex, BMI, familiar history of dyslipidemia, early-onset family history of CVD, levels of fasting glucose, Apo A1/ApoB100, and levels of LDL-C, and included data from 465 patients with information on all variables. As shown in Table 2, elevated levels of LDL-C significantly increased the probability of having Lp(a) levels ≥75 nmol/L (OR, 1.011; P<0.001). That is, by increasing 1 mg/L of LDL-C, the chance of having Lp(a) levels ≥75 nmol/L is increased by 1.1%.
Discussion
This cross-sectional and retrospective study describes the prevalence trends of Lp(a) levels in a Portuguese real-world pediatric sample referred to a clinical nutrition appointment, due to cardiovascular risk-related conditions, and its association with CVD risk factors. By showing that over 30% of children had Lp(a) levels of 75 nmol/L or higher, correlating with increasing lipid values, namely LDL-C, this study’s results emphasize the utmost importance of measuring Lp(a) early in life along with other potential CVD risk factors, so that strategies can be optimized to prevent and delay the onset of CVD later in life.
In this sample, 9.1% and 21.6% of the children had intermediate (75–125 nmol/L) and high (>125 nmol/L) Lp(a) levels, respectively, meaning 21.6% of patients were at high risk for CVD. This surpassed prevalences reported by others. Besides, a small percentage of patients (n=7, 0.9%) had serum Lp(a) levels over 430 nmol/L, implying a very high-risk for CVD, similar to familial hypercholesterolemia. A Korean retrospective study including pediatric patients aged 2 to 17 years reported 11.3% of patients with Lp(a) ≥110 nmol/L [12]. Such discrepancy possibly reflects differences in the sample characteristics, especially considering that Choi et al. [12] analyzed data from patients who attended local clinics and hospitals regardless of the reason for Lp(a) testing. In the Netherlands, 24.4% of a pediatric sample attending a lipid clinic and having suspected or definite familial hypercholesterolemia had Lp(a) levels of >30 mg/dL, whereas 12.2% exceeded 50 mg/dL [27]. The authors also demonstrated that patients with a clinical presentation of familial hypercholesterolemia had significantly higher and more frequently elevated Lp(a) levels [27], highlighting the importance of incorporating the clinical background of patients when assessing Lp(a) prevalence.
While the 75 and 125 nmol/L Lp(a) thresholds reflect the approach for adult cardiovascular risk stratification guidelines, their application in pediatric populations is supported by evidence that Lp(a) levels stabilize in early childhood. In fact, Lp(a) trends in children were comparable to those of a Portuguese sample of adult patients who were being followed in the same hospital and whose Lp(a) testing was conducted by the same laboratory [28]. In that sample, 10.4% and 28.7% of adults had intermediate and high Lp(a) levels, respectively, suggesting that the levels measured in childhood appear to be stable into adulthood, reinforcing the genetic contribution and the stability of Lp(a) values. On the other hand, emerging evidence suggests that intra-individual variability can occur, with changes in Lp(a) levels of up to 70% from childhood to adulthood [15] and up to 10 mg/dL or more than 25% in a significant proportion of adults, especially those with intermediate-risk levels [29-31]. Such variability may impact clinical decision-making, particularly in borderline cases, and could be better addressed through repeated measurements. Furthermore, both Portuguese studies identified a subset of patients with exceptionally high Lp(a) levels (>430 nmol/L), who carry an increased lifetime risk of atherosclerotic CVD [32], emphasizing the need for vigilant monitoring and early intervention in this group.
Nevertheless, comparisons of Lp(a) trends between studies must be made with caution. Adding to the different study designs and sample characteristics, the lack of a standardized method for measuring Lp(a) poses a further challenge to the comparison of results. For instance, a Dutsch study found a 11.3% difference in pediatric Lp(a) concentration obtained from 2 different assays and a 70% intra-individual variation arising from the different methods used [15]. The concentration of Lp(a) is currently measured by a variety of laboratory assays and is expressed either in moles, corresponding to the number of particles, or in mass, which varies with the different sizes of apo (a) molecules and does not correspond exactly to the number of particles [32-35]. These assays differ on their apo (a) isoform antibody specificity and calibration material. In an attempt to standardize the Lp(a) quantification method, the International Federation of Clinical Chemistry (IFCC) developed an international reference material – the IFCC SRM 2B – that has been accepted as the reference reagent for Lp(a) immunoassays [36]. The use of this reagent in combination with a monoclonal isoform-sensitive antibody against an unrepeated region of apo (a), as employed in this study, should resolve the inconsistency of results obtained from different assays and allow direct comparison of Lp(a) concentrations [35].
In agreement with others, the median Lp(a) levels were similar between the different age groups [12]. However, and despite the low number of patients, the distribution of Lp(a) in those aged 4 years and younger was similar to that in older patients (9 years and older). The youngest group of patients also appeared to have the highest levels of triglycerides, total cholesterol, and ApoB, as well as the highest rates of obesity and of early-onset family history of CVD, the latter being a known positive genetic determinant of Lp(a) levels [8-11] that has been linked to elevated Lp(a) levels [27,37]. Considering this profile, this group of patients warrants continued vigilance and would likely benefit from optimized measures to promote a healthy lifestyle and prevent CVD.
It is important to note that the sample comprises patients referred to the center due to specific clinical concerns, such as obesity, eating disorders, dyslipidemia, kidney or cardiac transplants, nephrotic syndrome, or a family history of premature CVD or dyslipidemia. This referral pattern indicates that the youngest group (<4 years) represents a highly selected and at-risk population, which may not be representative of the general pediatric population of that age.
Importantly, we did not find an association between elevated Lp(a) levels and an early-onset family history of CVD, contrary to previous descriptions [27,37], which could be a reflection of the considerable missing information. Still, we observed a significant yet weak positive correlation between Lp(a) levels and total cholesterol, non-HDL-C, and LDL-C values (P<0.001), all of which have been associated with an increased risk for adult cardiovascular events in 2 prospective multicenter studies that followed patients from childhood to adulthood [7,38]. It is therefore important to consider both patient characteristics and family history from the outset of a patient's follow-up to properly assess and manage their cardiovascular risk profile. Given the influential role of Lp(a) in atherosclerosis and CVD [2-7], both of which have a substantial clinical and economic impact in Portugal, Lp(a) testing during childhood could help identify individuals at increased lifetime cardiovascular risk and provide an additional reason for clinicians to encourage the adoption of a healthy lifestyle and implement early monitoring of CVD risk factors. Lp(a) has also been shown as a useful marker for cardiovascular risk reclassification, particularly in low-moderate risk patients [39], further reinforcing its early determination. Altogether, Lp(a) testing, combined with increased awareness and proactive management of cardiovascular risk factors, can provide significant clinical and economic benefits to both patients and the healthcare system, outweighing the cost of testing [40]. However, for Lp(a) testing to be used effectively in routine clinical practice, it is crucial to establish clear guidelines for its implementation, interpretation and follow-up. In particular, standardization of Lp(a) measurement assays and focusing screening strategies on high-risk populations will help to optimize resource allocation to individuals most likely to benefit from early detection and management.
The major strengths of this study include the use of a consistent and highly specific assay for quantifying Lp(a) that was performed by the same laboratory to reduce bias, as well as the relatively large sample size. Still, this study has limitations that need to be acknowledged. It was limited to the patients of a single center who were referred to a clinical nutrition appointment due to conditions that encompass cardiovascular risk. This introduced a selection bias toward a high-risk referred population, limiting the generalizability of the results to the broader Portuguese pediatric population, including the prevalence evaluation. The considerable amount of missing data due to its retrospective design also affected the analysis of risk factors associated with elevated Lp(a). Finally, its cross-sectional nature does not allow for causal inferences and relationships, limiting conclusions about their stability across different age groups. A nationwide and longitudinal analysis is desirable to elucidate the prevalence of Lp(a) in the Portuguese pediatric population, as well as its significance for assessing the cardiovascular risk, providing clinicians with important insights for patient management.
Altogether, the findings of this real-world study show a significant prevalence of children with high Lp(a) levels, similar to that of a Portuguese adult sample, underscoring the importance of incorporating this marker into the cardiovascular risk assessment profile early in childhood and at least once in life. However, it is essential to validate pediatric Lp(a) thresholds for early cardiovascular risk assessment, which can guide healthcare professionals to implement close monitoring and early strategies to prevent future cardiovascular risk.
PDF Links
PubReader
ePub Link
XML Download
PubMed
Download Citation
