This retrospective case-control study was carried out at Chaozhou Central Hospital affiliated to Southern Medical University, China. In-hospital neonates from Department of Neonatology or General Pediatrics from February 2012 to November 2019 were selected; 175 cases (110 males and 65 females) were in the prolonged jaundice group, while 149 cases (78 males and 71 females) were set as the control group. All infants were full-term, had gestational age greater than 37 weeks, and weighed more than 2,500 g.

The diagnostic criteria for neonatal unconjugated hyperbilirubinemia were formulated based on previous studies (6, 7). Prolonged jaundice was defined as the presence of unconjugated hyperbilirubinemia beyond 14 days with total serum bilirubin (TSB) levels >150 µmol/L, or >100 µmol/L beyond 28 days of age (6). Severe hyperbilirubinemia was defined as prolonged jaundice newborns with TSB ≥ 342 µmol/L. The inclusion criteria for the control group were infants more than 14 days old without pathological jaundice, and hospitalized for mild pneumonia, umbilical infection, or other non-jaundice causes. The exclusion criteria were neonates with hemolysis due to incompatibility of the ABO or Rh blood group, sepsis, thalassemia, extravascular hemolysis, serious maternal diseases, organ dysfunction, and congenital malformation with clear etiology, or liver disease that might affect the total serum bilirubin level. These conditions were evaluated based on the medical history of the patients, clinical and laboratory tests such as whole blood cell count, liver function tests, reticulocyte count, blood groups and subgroups, and Coombs test.

All infants received standard clinical care. After conducting a routine clinical test, the remaining ethylenediaminetetraacetic acid (EDTA) anticoagulant blood was obtained. We stored the remaining blood from the neonates in the biobank of our hospital at −40°C for further analysis. Information on sex, age, birth weight, gestational age, delivery mode, feeding pattern, and bilirubin level was recorded at admission before phototherapy. The phototherapy protocol, reviewed from medical records, was based on the American Academy of Pediatrics guidelines (2004) (7).

The Medical Ethics Committee of Chaozhou Central Hospital approved this study (Approval No. 2011021 and No. 2015001). The samples of this retrospective case-control investigation were collected after standardized diagnosis and treatment; Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

The G6PD-deficient samples, determined by the clinically approved G6PD enzyme quantification assay (Guangzhou Micky Medical Instrument Co., Guangdong, China), were used for molecular analysis (8). Extraction of the genomic DNA and detection of the common G6PD mutation, including C.95A > G (Gaohe), C.871G > A (Viangchan), C.1004C > T (Foshan), C.1024C > T (Chinese-5), C.1376G > T (Canton), and C.1388G > A (Kaiping), were performed using a kit (Yaneng Biotechnology Limited Corp., Shenzhen, China) (8). For samples highly suspected of having G6PD gene mutations but not detected with this kit, 10 pairs of primers were designed for amplifying and sequencing exons 2–13, and the sequences were compared to those in NCBI BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) (8).

Genomic DNA from peripheral blood samples was extracted using the blood genomic DNA extraction kits (Yaneng Biotechnology Limited Corp., Shenzhen, China). The primers were designed for exon 1 and the flanking sequence of the UGT1A1 gene; PCR amplification was performed using the TaKaRa Taq enzyme. The reaction conditions were 95°C for 5 min, followed by 94°C for 30 s, 60°C for 40 s, 72°C for 1 min, 35 cycles, and 72°C extension for 10 min. Finally, the sequences of the samples were analyzed using an ABI Genetic Analyzer (9).

Bilirubin levels at admission, gestational age, and birth weight were compared between two groups by conducting T-tests. To assess factors such as the ratio of male to female, the delivery method, the feeding methods, and the variation frequency of the G6PD and UGT1A1 genes, the χ² test or Mann–Whitney U test was performed. The odds and odds ratios of prolonged jaundice and severe hyperbilirubinemia (TSB ≥ 342 µmol/L), associated with UGT1A1, G6PD, sex, weight at birth, delivery mode, gestational age, and feeding methods, were analyzed with multiple logistic regression performance. SPSS16.0 was used to conduct all statistical analyses, and all differences between groups were considered to be statistically significant at P < 0.05.

In total, data from 324 infants were included in the study, 175 cases (110 males and 65 females) were classified in the prolonged jaundice group and 149 cases (78 males and 71 females) were in the control group. The clinical characteristics of the infants are presented in Table 1. The sex ratio, gestational age, birth weight, and delivery mode ratio between the two groups were not significantly different (P > 0.05); however, feeding methods differed significantly (P < 0.001). In addition, the bilirubin levels of the infants in the jaundice group were higher than those of the infants in the control group (Table 1 and Figure 1).

The infants with prolonged jaundice (n = 175) were divided into mild-medium jaundice group 1 (TSB < 342 µmol/L) and severe jaundice group 2 (TSB ≥ 342 µmol/L) (Figure 1). The clinical characteristics, including sex, age, gestational age, birth weight, delivery mode, and feeding method, did not differ significantly between these two groups (P > 0.05) (Table 2). Some infants (52/107, 48.5%) in group 1 and all infants in group 2 (68/68, 100%) received phototherapy (χ² = 50.974, P < 0.001). Three infants in group 2 developed bilirubin encephalopathy, with the highest total bilirubin levels of 469.9 µmol/L, 527.9 µmol/L, and 561.9 µmol/L, respectively.

The genotype frequencies of homozygous and heterozygous c.211G > A variant of UGT1A1*6 were 12.0% (21/175) and 46.9% (82/175) in the case group, respectively, which were higher than those of the control group (Table 1). The allele frequency of the c.211G > A variant of UGT1A1*6 in the prolonged jaundice group was 58.9% while it was 21.5% in the control group (χ² = 50.15, P < 0.001). Furthermore, the allele frequencies of c.211G > A in groups 1 and 2 were 46% and 79%, respectively (χ² = 21.66, P < 0.001) (Table 2). The prevalence of homozygous and heterozygous UGT1A1*6 in group 1 was 6.5% (7/107) and 39.3% (42/107), respectively; and in group 2, their prevalence was 20.6% (14/68) and 58.8% (40/68), respectively (Table 2).

G6PD deficiency was identified in 6 cases of prolonged jaundice group (175 cases) by enzyme assay and no G6PD deficiency case was identified in the control group. The G6PD genotypes in the prolonged jaundice group were c.1388 G > A (2/6), c.1376 G > T (2/6), c.517 T > C (NanKang) (1/6), and c.95 A > G (1/6), respectively. The frequency of G6PD deficiency differed significantly between the prolonged jaundice group (3.4%, 6/175) and the control group (0%, 0/149) (χ² = 5.21, P = 0.033) (Table 1).

Of the 68 infants with severe prolonged jaundice, three suffered from bilirubin encephalopathy, with peak bilirubin values of 469–561 µmol/L, two cases were homozygous 211G > A with G6PD mutation (c.1376 G > T and c.1388 G > A, respectively), and one was heterozygous 211G > A with G6PD deficiency (c.95 A > G).

Infants with prolonged jaundice, who were only breastfed, had a higher risk than formula-fed infants (OR = 2.27, 95% CI: 1.19–4.34, P = 0.012). Additionally, both homozygous (OR = 63.10, 95% CI: 5.30–751.63, P = 0.001) and heterozygous forms (OR = 4.43, 95% CI: 2.53–7.74, P = 0.000) of c.211G > A were significant risk assessment indicators for prolonged jaundice (Table 1). As summarized in Table 2, homozygous (OR = 8.80, 95% CI, 2.84–27.23, P < 0.001) and heterozygous (OR = 4.01.95% CI, 1.91–8.45, P < 0.001) UGT1A1*6 (c.211G > A) was the most significant predictor of severe prolonged jaundice (TSB ≥ 342 µmol/L).