Background

Primary carnitine deficiency (PCD, OMIM #212,140) is a rare inherited autosomal recessive disorder of fatty acid oxidation caused by mutations of the solute carrier family 22 member 5 (SLC22A5, MIM:603,377) gene. SLC22A5 encodes the organic cation/carnitine transporter type 2 (OCTN2), which is prominently expressed in the heart, skeletal muscle, kidney, and placenta [1, 2]. Defect in OCTN2 synthesis hinders the reabsorption of carnitine, resulting in low levels of serum carnitine, impairing the transport of long-chain fatty acids from the cytosol to the mitochondria for beta oxidation.

The clinical manifestations of PCD vary widely, ranging from asymptomatic to acute metabolic decompensation early in life, progressive hypertrophic cardiomyopathy, myopathy, and encephalopathy later in life [3,4,5]. However, untreated patients with PCD may experience sudden death [6], thereby highlighting the importance of timely and continuous carnitine supplementation to prevent metabolic decompensation and ensure favorable long-term outcomes. Hence, early diagnosis is crucial. Newborn screening (NBS) for PCD is performed to measure free carnitine (C0) levels in dried blood spot samples using tandem mass spectrometry (MS/MS). The widespread implementation of NBS for PCD in China has enabled early diagnosis and timely treatment of patients with PCD.

The estimated incidence of PCD varies widely across countries due to region and race, ranging from 1:120,000 to 1:300 newborns [7,8,9,10]. Additionally, significant differences in the incidence of PCD have been reported among different regions within China, with rates ranging from 1:100,000 to 1:3000 newborns [11,12,13,14,15]. It is important to clarify the overall prevalence of PCD among the Chinese population, as well as potential regional disparities, notably across the Qinling Mountains-Huaihe River Line that divides China into northern and southern regions. This geographic division introduces variations in the natural environment, geographical landscape, and residents' lifestyles, which may influence disease prevalence. It remains unclear whether there are differences in the prevalence of PCD between the northern and southern regions.

SLC22A5, located on chromosome 5q31.1, includes ten exons and three introns and encodes 557 amino acids. Over 180 pathogenic SLC22A5 gene variants have been identified, and variations have been observed among different racial and regional populations. (http://www.hgmd.cf.ac.uk; data collected on December 15, 2023). Some studies have identified c.51C > G (p.F17L), c.760C > T (p.R254*), and c.1400C > G (p.S467C) as the three most common variants in the Chinese population [16,17,18,19]. However, the frequency of the most prevalent variant varies across regions.

To elucidate the epidemiological characteristics of PCD in Chinese populations, a comprehensive meta-analysis was conducted to analyze the nationwide incidence of PCD and clarify the differences in the prevalence of PCD and SLC22A5 gene variants between northern and southern regions.

Methods

Literature search

The systematic review and meta-analysis were conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [20], with the protocol registered in PROSPERO (ID: CRD42024526722). Three independent researchers (ZJF, LJL, and ZYL) systematically searched databases from 2000 to June 2023 for observational studies on PCD, including English databases encompassing the PubMed, Embase, and Web of Science, and Chinese databases encompassing the China National Knowledge Infrastructure (CNKI), Veipu (VIP), and Wanfang. Search terms comprised (“primary carnitine deficiency” OR “carnitine uptake defect” OR “carnitine transport defect”) AND (“SLC22A5” OR “OCTN2”) AND (“mutation” OR “variant”) AND (“neonate” OR “newborn” OR “neonate”). Relevant studies on human participants published in English and Chinese were included, and reference lists of relevant reviews and articles were manually examined.

Eligibility and exclusion criteria

Studies were included if they met the following criteria: (1) original observational studies; (2) studies reporting results of PCD screening for newborns in different provinces, cities, and autonomous regions of China; (3) studies featuring main indicators such as the prevalence of PCD, information on SLC22A5 gene variants and other relevant PCD-related characteristics, and (4) Studies with relatively high quality.

Studies failing to meet these criteria were excluded from the analysis. In addition, studies with overlap** screening regions or times, low quality, or those not published in English or Chinese languages were excluded.

Data extraction

Two researchers (ZJF and LJL) extracted the information independently. All relevant data were compiled into a data extraction table based on the specified eligibility and exclusion criteria. Information obtained from the original publications included the first author, publication year, screening year, region, number of NBS participants, number of diagnosed PCD cases, and details on SLC22A5 gene variants. The frequencies of SLC22A5 gene variants were calculated from the extraction data. Any discrepancies were resolved through discussion with another investigator (ZYL). Since all analyses relied on previously published studies, ethical approval or patient consent was not required.

Quality assessment

Two investigators (ZJF and LJL) independently evaluated the quality of included studies using the Agency for Healthcare Research and Quality (AHRQ) criteria. Studies were assigned a score of 1 for compliance with each criterion and a score of 0 for non-compliance or uncertainty, yielding a total score ranging from 0 to 11. Higher scores indicated superior quality, with the included studies categorized as having high, moderate, or low quality, corresponding to scores of 8–11, 4–7, and 0–3 points, respectively. Any discrepancies were thoroughly discussed and resolved by another reviewer (ZYL) if necessary.

Statistical analyses

All statistical analyses were performed using RevMan version 5.3 (Update Software Ltd., Oxford, Oxon, UK). The chi-square test and I2 were used to evaluate heterogeneity. Heterogeneity was deemed small when the I2 value was less than 50% with a P value > 0.10, in which case the fixed-effects model (the Mantel–Haenszel method) was used to estimate the pooled prevalence (OR) and 95% confidence intervals (CIs). Otherwise, a random-effects model (the Der Simonian and Laird method) was employed. Subgroup analysis was conducted to assess the effect of region across the studies. Sensitivity analyses were performed by individually removing studies to evaluate their impact on pooled ORs. Publication bias was visually assessed using a funnel plot and quantitatively evaluated using Begg’s test. A P value < 0.1 indicated the existence of publication bias.

Results

Study selection

In our initial data search, 1,889 articles (797 in Chinese and 1,092 in English) were identified. Among these, 1,067 articles were excluded after screening duplicates. After reviewing the titles and abstracts of the remaining articles, 781 were further excluded. Subsequently, 41 articles were considered potentially eligible. Upon a thorough review of the full texts, 19 articles were excluded, including those with overlap** screening regions or screening time eligibility (n = 15), meta-analyses (n = 1), and low-quality studies (n = 3). Finally, 30 eligible studies were included in the meta-analysis. A flowchart illustrating the literature search process is shown in Fig. 1.

Fig. 1
figure 1

Flow chart of the study selection process

Full size image

Study characteristics

Twenty-two studies involving 9,958,380 newborns were included in our analysis, among which 476 patients were diagnosed with PCD, as shown in Table 1. Notably, 79.53% of the screened newborns resided in southern China. Of the 476 patients with PCD, 469 underwent genetic diagnosis, revealing 890 variants of 934 alleles of SLC22A5, among which 107 different variants were detected. The five most prevalent variants accounted for 71.13% of the total number, including c.1400C > G, p.Ser467Cys (29.78%, 265/890), c.51C > G, p.Phe17Leu (19.89%, 177/890), c.760C > T, p.Arg254Ter (14.27%, 127/890), c.428C > T, p.Pro143Leu (4.16%, 37/890), and c.338G > A, p.Cys113Tyr (4.04%, 36/890), as shown in Table 1.

Table 1 Characteristics of studies included in the meta-analysis
Full size table

Assessment of quality

The evaluation results of the AHRQ quality assessment items are also presented in Table 1, indicating that articles with scores four or more were classified as having moderate or high quality. The average score of 7.68 indicated minimal risk of bias.

Results of meta-analysis

Incidence of PCD

All included studies reported the incidence of PCD. Due to significant heterogeneity among included studies (I2 = 83%, P < 0.05), a random-effects model was employed to analyze the incidence of PCD in China. The meta-analysis showed that the prevalence of PCD was 0.05‰ [95%CI, (0.04‰, 0.06‰)] or 1/20 000 [95%CI, (1/16 667, 1/25 000)] in China (Fig. 2). Subgroup analyses of regional incidence revealed that the incidence in southern China [0.07‰, 95%CI, (0.05‰, 0.08‰)] was greatly higher than that in north China [0.02‰, 95%CI, (0.02‰, 0.03‰)] (P < 0.001), as shown in Figs. 2 and 3.

Fig. 2
figure 2

Meta-analysis of the incidence of PCD between southern and northern China

Full size image
Fig. 3
figure 3

The schematic diagram shows the incidence of PCD in different provinces of China

Full size image

Frequency of SLC22A5 gene variants

Among the included studies, 21 reported variant spectra of SLC22A5 in patients with PCD. We performed a meta-analysis of the frequencies of the five most prevalent gene variants. Since significant heterogeneity was observed among the variant frequencies of c.1400C > G and c.760C > T (I2 = 66% and 76%, P < 0.05), a random-effects model was used for the analysis. No significant heterogeneity was identified among the variant frequencies of c.51C > G, c.428C > T, and c.338G > A (I2 = 29%, 0%, and 0%, P > 0.05), and a fixed-effects model was used for the analysis. The result of the meta-analysis showed that the frequency of the variant with c.1400C > G, c.51C > G, c.760C > T, c.338G > A, and c.428C > T were 45% [95%CI, (34%, 59%)], 26% [95%CI, (22%, 31%)], 14% [95%CI, (10%, 20%)], 6% [95%CI, (4%, 8%)], and 5% [95%CI, (4%, 8%)], respectively. Among the subgroup analyses, the variant frequency of c.1400C > G in southern China [39%, 95%CI, (29%, 53%)] was significantly lower than that in northern China [79‰, 95%CI, (47‰, 135‰)] (P < 0.05) (Fig. 4), while there was no statistically significant difference in the frequency of other variants between southern and northern China (Fig. S1-4).

Fig. 4
figure 4

Meta-analysis of the frequency of the c.1400C > G variant of the SLC22A5 gene between southern and northern China

Full size image

Publication bias analysis

Funnel plots were utilized to evaluate publication bias for the incidence of PCD and the frequency of SLC22A5 gene variants. The plots were approximately symmetrically distributed, indicating no significant publication bias (Fig. 5).

Fig. 5
figure 5

Funnel plots for publication bias. A Funnel plot of the incidence of PCD. B Funnel plot of the frequency of the c.1400C > G variant of the SLC22A5 gene. C Funnel plot of the frequency of the c.51C > G variant of the SLC22A5 gene. D Funnel plot of the frequency of the c.760C > T variant of the SLC22A5 gene. E Funnel plot of the frequency of the c.428C > T variant of the SLC22A5 gene. F Funnel plot of the frequency of the c.338G > A variant of SLC22A5 gene

Full size image

Discussion

Our meta-analysis included 22 studies on PCD screening for newborns conducted in 12 provinces (municipalities) across China over the past two decades, including approximately 10 million newborns. This analysis is currently the most comprehensive and systematic review of PCD screening worldwide. Our findings revealed a prevalence of PCD of approximately 1 in 20,000 Chinese newborns, with a significantly higher prevalence noted in southern China than in northern China. Benefiting from the extensive sample size and the relative representativeness of regional population distribution, along with the absence of publication bias in the literature, the results of our meta-analysis offer an objective and reliable assessment.

Timely detection, diagnosis, and intervention for PCD through NBS are crucial to mitigate severe clinical outcomes in affected individuals [21,22,23]. With the widespread application of MS/MS and genetic testing, PCD can be promptly diagnosed and managed.

Globally, the Faroe Islands exhibit the highest incidence of PCD, with a prevalence of up to 1:297 [7]. This ratio varies across different regions, estimated at 1:50,000 in the United States [9], 1:25,000 in Egypt [24], 1:30,000 in Thailand [10], and 1:40,000 in Japan [25]. It is worth noting that the estimated prevalence of PCD is 1:17,641 in China based on the carrier frequency of SLC22A5 pathogenic or likely pathogenic (P/LP) variants from the Chinese Newborn Genome Project [26]. Additionally, a national cross-sectional survey included 7 million newborns and reported a prevalence of 1:20,284 in mainland China [27], consistent with our study findings. Subgroup analysis revealed a significantly higher prevalence of PCD in southern China, particularly in the Fujian [17, 28, 29], Jiangxi [30], and Guangxi Provinces [31, 32], indicating a geographical trend with higher prevalence in southern China and lower prevalence in northern China.

Uncovering the mutation spectrums of SLC22A5 gene plays an important role in clarifying the correlation between genotype and phenotype, genetic counseling, and management of PCD. Several ethnic-specific variant spectra of SLC22A5 have been identified in different populations. For example, c.95A > G (p.N32S) is predominant in the Faroe Islands [7], c.136C > T (p.P46S) in the United States [33], c.51C > G (p.Phe17Leu) in Thailand [10], c.760C > T (p.R254X) and c.454G > C (p.G152R) in Turkey [34], and c.1400C > G (p.S467C) in Japan [25].

Our analysis revealed that the top five mutations collectively accounted for 71.13% of the total number, providing convincing evidence for the rapid detection of targeted variants of the SLC22A5 gene in the Chinese population. While studies have consistently identified c.1400C > G (p.S467C), c.51C > G (p.F17L), and c.760C > T (p.R254*) as the top three prevalent variants in the Chinese population, the variant with the highest frequency varies across different geographic regions. Subgroup analysis indicated that the c.760C > T mutation, which results in very low residual OCTN2 transporter activity and obvious clinical manifestations, was most frequent in the Fujian Province [17], while c.1400C > G, with residual OCTN2 transporter activity that may lead to mild phenotypes, predominated in Jiangsu [35], Shandong [11], and Henan [19] Provinces. In contrast, c.51C > G was the most prevalent variant in Shanghai [36] and Guangxi [30,31,32] Provinces. Our findings underscore regional disparities, with c.1400C > G exhibiting a higher frequency in northern China than in southern China.

Conclusions

Our systematic review and meta-analysis of NBS results for PCD in China yielded a comparatively accurate prevalence of 1/20 000. Our findings highlight a significantly higher incidence of PCD in southern China than in northern China. Additionally, we confirmed the three most common variants of SLC22A5 in the Chinese population, noting a higher frequency of c. 1400C > G variant in northern China than in southern China. However, the coverage rate of MS/MS-based newborn screening, being a voluntary chargeable program in China, remains relatively low in northwest China due to poor medical and economic conditions. Consequently, the number of included studies in northern China is relatively limited compared to that in southern China. Overall, our study provides valuable epidemiological insights into PCD in the Chinese population, guiding future endeavors in NBS for PCD.