Analysis of risk factors for parenteral nutrition-associated cholestasis in preterm infants: a multicenter observational study

The clinical data were obtained from a prospective multicenter study on the effect of SMOF in preterm infants, which was conducted in five tertiary level A hospitals with NICU of level IIIB or above in China between January, 2021 and December, 2021. The study was approved by the Ethics Committee of the Women and Children’s Hospital of Xiamen University (Grant No. KY-2020–0086–01) and it was registered in the China Clinical Trials Registry (Registration No. ChiCTR2100041910) (10/01/2021). Written informed consent was obtained from the parents or guardians of preterm infants included in the study.

A total of 465 preterm infants were included in the study (Fig. 1). Based on the inclusion criteria, exclusion criteria and PNAC status, preterm infants were assigned to the PNAC group (n = 81) and the non-PNAC group (n = 384).

Collected data included general information on preterm infants and clinical data on early complications in preterm infants, including respiratory distress syndrome (RDS), hemodynamically significant patent ductus arteriosus (hsPDA), early-onset sepsis (EOS), stage II or higher NEC, surgically treated NEC, LOS, brain injury (including intraventricular hemorrhage and periventricular white matter softening), retinopathy of prematurity (ROP), moderate and severe BPD, metabolic bone disease of prematurity (MBDP ), EUGR, nutritional information related to the intake of amino acids and fat emulsions, duration of PN, time to initiate enteral feeding, breastfeeding, feeding intolerance (FI), time to reach total enteral nutrition and the number of days of total calorie intaked up to 110 kcal/kg/day.

SPSS 22.0 software was used for statistical analysis. Enumeration data were expressed as rates (%) and comparisons between groups were made using the χ² test or Fisher’s exact test. The Kolmogorov-Smirnov test was used to evaluate whether the measures conformed to a normal distribution. Normally distributed measures were expressed as x ± s and independent samples t-test was used for comparison between the groups. When the measurement data belong to abnormal distribution, data were presented as median (interquartile range) and the rank sum test was used for comparison between the groups. The correlations between birth weight, gestational age at birth, duration of PN and PNAC were calculated by Pearson’s correlation coefficient (r). Multivariate analysis was performed using binary logistic regression analysis. p < 0.05 was considered a statistically significant difference.

Compared to the non-PNAC group, the PNAC group had a lower mean gestational age and weight at birth and longer duration of invasive and non-invasive mechanical ventilation, longer total oxygen support and longer total hospital stay (p < 0.001 for all). There were no significant differences in the incidence of low gestational age, sex, cesarean section, five-minute Apgar score ≤ 7, premature rupture of membranes > 18 h, etc. between the two groups (P > 0.05 for all) (Table 1).

The incidence of PNAC in infants born at gestational age < 28w, 28–29w and > 30w were 34.2%, 16.7% and 12.7%, respectively. The incidence of PNAC in infants born at weight < 1000 g, 1000–1499 g and ≥ 1500 g were 42.3%, 13.8% and 5.8%, respectively. Lower gestational age and lower birth weight were associated with a higher incidence of PNAC (r values were − 0.180 and − 0.277, respectively) (Table 2).

Compared to the non-PNAC group, preterm infants in the PNAC group received more medium/long-chain fat emulsion and less SMOF. The incidence of PNAC was 13.7% in the preterm infant group receiving SMOF compared to 21.2% in the preterm infant group receiving medium/long-chain fat emulsion (P = 0.002). The duration of PN was longer for patients with a higher incidence of PNAC. The incidence of PNAC was 9.0%, 13.0% and 29.6% in preterm infants with duration of PN < 3w, 3w-4w and > 4w, respectively. The duration of PN was positively associated with the incidence of PNAC (r = 0.235, P = 0.044). In addition, the PNAC group had higher maximum doses of amino acids and fat emulsion, lower rates of breastfeeding, higher incidence of FI, more accumulated days to achieve total enteral nutrition, a slower velocity of weight growth and fewer accumulated days of total calories up to standard 110 kcal/kg/day (P < 0.05 for all). There were no significant differences in the initial dose of amino acid, fat emulsion, etc. between the groups (Table 3).

The incidence of RDS (P = 0.004), hsPDA (P = 0.015), stage II or more, surgically treated NEC (P = 0.004), LOS (P = 0.037), MBDP (P < 0.001) and EUGR (P < 0.001) was higher in the PNAC group compared to the non-PNAC group. There was no statistically significant difference in the incidence of EOS, brain injury, ROP with intervention and moderate to severe BPD between the groups (P > 0.05 for all) (Table 4).

All univariate results at p < 0.05 were included in logistic regression analysis, which showed that higher maximum doses of amino acids (OR, 5.352; 95% CI, 2.355 to 12.161), EUGR (OR, 2.396; 95% CI, 1.255 to 4.572), FI (OR, 2.581; 95% CI, 1.395 to 4.775), surgically treated NEC (OR, 11.300; 95% CI, 2.127 to 60.035) and longer total hospital stay (OR, 1.030; 95% CI, 1.014 to 1.046) were independent risk factors for the development of PNAC. SMOF (OR, 0.358; 95% CI, 0.193 to 0.663) and breastfeeding (OR, 0.297; 95% CI, 0.157 ~ 0.559) were protective factors for PNAC (Table 5).

The present study showed that both gestational age at birth and birth weight were lower in the PNAC group (p < 0.001), and that the development of PNAC was negatively correlated with both gestational age at birth and birth weight (r values were − 0.180 and − 0.277, respectively). Lee et al. [6] showed that lower birth weight was an independent risk factor for PNAC. Zhang et al. [21] also reported that low gestational age at birth and low birth weight were the main risk factors for the progression of PNAC. However, in the present study, multivariate regression analysis showed that gestational age at birth and low birth weight were not independent risk factors for PNAC. A retrospective study of risk factors associated with PNAC in preterm infants showed that birth weight and gestational age at birth were not independent risk factors for PNAC in preterm infants (P were 0.219 and 0.998, respectively) [22]. Another study of 387 infants examining risk factors for PNAC in VLBWI also showed that birth weight and gestational age at birth were not independent risk factors for the development of PNAC (P 0.773 and 0.181, respectively) [23], which is consistent with the results of this study. However, lower gestational age at birth and lower birth weight were associated with a higher incidence of early complications in preterm infants, which tend to have more severe symptoms and inevitably lead to a longer hospital stay. Multivariate logistic regression analysis in this study showed that longer hospital stay was an independent risk factor for PNAC (OR, 1.030; 95% CI, 1.014 to 1.046). Indirectly, this data suggested that lower gestational age at birth and lower birth weight were strongly associated with the development of PNAC.

The maximum dose of amino acids in the infants of the PNAC and non-PNAC groups in this study was 4.0 g/kg/d and 3.5 g/kg/d, respectively, and this study showed that the maximum dose of amino acids was an independent risk factor for PNAC (OR, 5.352; 95% CI, 2.355–12.16). A study by Sung et al. [24] showed that high doses of amino acids intake in the early postnatal period (3.0 g/kg/d) increased the risk of refeeding syndrome, such as electrolyte disturbances in extremely low birth weight infants. Vileisis et al. [25] performed a prospective controlled study comparing the effects of two different levels of amino acids intake (2.3 g/kg/d vs. 3.6 g/kg/d) on liver function, which showed that increased levels of amino acids intake in PN was associated with the early onset of PNAC and higher bilirubin levels in infants, suggesting that higher level of amino acids intake during PN can directly damage hepatocytes and affect bile excretion from the damaged liver, leading to hepatocellular cholestasis. Data from another retrospective study on factors influencing the progression of PNAC showed a higher incidence of PNAC in infants with an intake of parenteral amino acids > 3.5 g/kg/d compared to those with an intake of < 3.5 g/kg/d (15.1% vs. 6.5%) [22]. This may be attributed to the fact that some of the amino acids in PN, such as methionine, tryptophan and phenylalanine, which considered hepatotoxic, causing cholestasis by directly damaging the tubular membranes of hepatocytes [26]. SMOF is well tolerated in patients consisting of 30% medium chain triglycerides, 30% soybean oil, 25% olive oil and 15% fish oil. ω-6 to ω-3 ratio of long-chain polyunsaturated fatty acids (LCPUFAs) is 2.5:1, which is lower than the 7:1 ratio of medium and long chain fat emulsions. The phytosterol content is lower, while vitamin E and α-tocopherol content is higher in SMOF. SMOF combines the advantages of a variety of oils and is currently considered to be a more balanced fat emulsion. The results of this study showed 13.7% incidence of PNAC in infants receiving SMOF compared to 21.2% in infants receiving medium and long-chain fat emulsions (P = 0.002). The results of the multifactorial logistic regression analysis in this study showed that SMOF was a protective factor for PNAC (OR, 0.358; 95% CI, 0.193 to 0.663). Jiang et al. [27] performed a randomized controlled trial on the effect of SMOF and medium/long chain fat emulsion in 160 neonates after intestinal surgery, which showed that infants receiving SMOF tolerated lipids better and were better protected against PNAC compared to those receiving medium/long-chain fat emulsion, and this protection was more pronounced in infants on intravenous nutrition lasting more than 4 weeks. A retrospective study by Kasirer et al. [28] also found a lower incidence of PNAC in VLBWI with SMOF compared to medium/long-chain fat emulsion (6% vs. 13%; P = 0.022). These results are consistent with the present study. It is suggested that SMOF has a better hepatoprotective effect and reduces the incidence of PNAC in infants mainly due to the high content of ω-3 LCPUFAs, which possess anti-inflammatory properties and confer a potent hepatoprotective effect. Olive oil, one of the main components of SOMF, is rich in oleic acid, an ω-9 monounsaturated fatty acid, which not only reduces oxidative stress but also promotes the biosynthesis of ω-3 LCPUFAs in the liver [29]. SMOF contains less phytosterol, which has been shown to inhibit the expression of proteins involved in bile acid and bilirubin excretion in animal studies, leading to cholestasis [30]. SMOF also contains high doses of vitamin E and ɑ-tocopherol, which act as anti-inflammatory and antioxidant agents. It appeared that many components of SMOF are protective factors against hepatotoxicity, which possibly inhibit the development of PNAC in infants.

Studies have shown that breastfeeding promotes growth and development, improves feeding tolerance, reduces oxidative stress and promotes anti-inflammatory responses in preterm infants [31]. It can effectively reduce the incidence of complications, such as BPD, NEC, ROP and PNAC in preterm infants. In this study, multifactorial logistic regression analysis confirmed that breastfeeding (OR, 0.297; 95% CI, 0.157 to 0.559) was a protective factor for PNAC, and FI (OR, 2.581; 95% CI, 1.395 to 4.775) was a risk factor for PNAC. The results showed a higher breastfeeding rate in the non-PNAC group compared to the PNAC group (54.4% vs. 27.2%, P < 0.001), resulting in a lower incidence of FI, shorter time to achieve total enteral feeding and faster average velocity of weight growth. However, the fewer accumulated days of total calories up to standard 110 kcal/kg/day suggested that energy supply in the early postnatal period for preterm infants in the PNAC group was mainly through PN, particularly via increased amino acid and fat intake, while enteral nutrition was relatively inadequate, leading to a longer duration of PN with an increased risk of PNAC.

Therefore, the management of postnatal enteral and parenteral nutrition of VPI or VLBWI should be balanced by increasing breastfeeding rates, promoting feeding tolerance, avoiding excessive amino acid intake and infusing SMOF, thus, reducing the incidence of PNAC.

Lower gestational age at birth and birth weight of preterm infants were associated with poorer neurological development, poorer organ and tissue development and higher incidence of EUGR [32]. In this study, a multifactorial logistic regression analysis revealed that EUGR (OR, 2.396; 95% CI, 1.255 to 4.572) was an independent risk factor for PNAC, and the PNAC group required longer duration of invasive and non-invasive ventilation, total oxygen support, and had a higher incidence of RDS, hsPDA and LOS (p < 0.05 for all). It is suggested that the infants in the PNAC group were severely ill, and the management of enteral nutrition during the early postnatal period is more likely to be neglected, resulting in inadequate enteral nutritional support and a longer time to reach total enteral feeding, yet enteral nutrition was essential to avoid EUGR in preterm infants [33]. Moreover, the duration of PN was prolonged due to inadequate enteral nutrition support, and a longer duration of PN infusion increased the risk of developing PNAC, forming a vicious circle of EUGR and PNAC. Multifactorial logistic regression analysis in this study also found that surgically treated NEC (OR, 11.300; 95% CI, 2.127 to 60.035) was an independent risk factor for PNAC and contributed most to the progression of PNAC. A retrospective study on factors influencing PNAC in 114 cases of VLBWI (including 27 cases in the PNAC group and 87 cases in the non-PNAC group) found that NEC was an independent risk factor for PNAC [34]. A retrospective study by Kim et al. [35] examining factors influencing cholestasis in VLBWI showed that both stage II or higher NEC and bowel surgery were independent risk factors for PNAC, which is consistent with the results of this study. One study was conducted on NEC concerning postnatal growth at hospital discharge and 2 years later in extremely low birth weight infants by Hong et al. [36] in a total of 9,171 infants, which showed that the incidence of severe growth retardation at discharge was higher in both infants with NEC and those with surgically treated NEC compared to those without NEC (56% vs. 36% and 61% vs. 36%). A study reports a 5-fold increased risk of EUGR once NEC is diagnosed, regardless of surgical treatment [37]. This is related to the fact that prolonged fasting after the occurrence of NEC and clinicians tend to be very cautious about the indication for reopening the breastfeeding after fasting. Surgically treated NEC can lead to short bowel syndrome, disruption of the intestinal mucosal barrier, dysbiosis, intestinal bacterial translocation and absorption of toxins [38, 39], and these factors combined with EUGR collectively increased the risk of PNAC. Therefore, enteral nutrition in preterm infants (VPI or VLBWI) should be prioritized without increasing the incidence of NEC, especially avoiding surgical NEC and reducing the incidence of EUGR [4]. Tailored management of enteral nutrition in VPI play an important role in reducing the incidence of PNAC.

This study collected data from a multicenter observational survey and analyzed the risk factors associated with the development of PNAC in VPI or VLBWI. With a large sample size and more convincing evidence, we provided a further scientific basis for the prevention of PNAC. This study had some limitations. Firstly, the study missed several parameters, including the duration days of fasting of preterm infants in the two groups. However, analyzing the incidence of FI and the time to achieve full enteral feeding could have compensated for the missing parameters. Secondly, the study did not explore the effect of dose of dextrose on PNAC, in the literature, it has been reported that excessive intravenous dextrose intake also increases the risk of developing PNAC [40]. Third, the factors affecting the onset of PNAC may be different in different PN durations, and this study did not in-depth analysis of the effect of different PN durations on the onset of PNAC. The next step will be to conduct a follow-up study to clarify the impact of PNAC on the long-term prognosis of preterm infants.

In summary, the incidence of PNAC can be reduced by increasing breastfeeding for VPI or VLBWI, using SOMF, avoiding excessive amino acid intake, promoting feeding tolerance, preventing severe NEC requiring surgery and reducing the incidence of EUGR.