Neuroprotective effects of caffeine, erythropoietin, magnesium sulfate, and thyroxine in preterm infants: a network meta-analysis
Original Article

Neuroprotective effects of caffeine, erythropoietin, magnesium sulfate, and thyroxine in preterm infants: a network meta-analysis

Yuqian Wang1, Shanshan Wang1, Haiping Dou1, Yajun Zhang2, Liu Yang1

1Department of Pediatrics, The Second Hospital of Dalian Medical University, Dalian, China; 2Department of Anesthesiology, Dalian Women and Children’s Medical Group, Dalian, China

Contributions: (I) Conception and design: Y Wang, L Yang; (II) Administrative support: L Yang; (III) Provision of study materials or patients: Y Wang, S Wang, H Dou; (IV) Collection and assembly of data: Y Wang, S Wang, Y Zhang; (V) Data analysis and interpretation: Y Wang, L Yang; (VI) Manuscript writing: All authors; (VII) Final approval of manuscript: All authors.

Correspondence to: Dr Liu Yang, MD, PhD. Department of Pediatrics, The Second Hospital of Dalian Medical University, No. 467 Zhongshan Road, Shahekou District, Dalian 116021, China. Email: 439140628@163.com.

Background: Premature infants are prone to having adverse neurodevelopmental outcomes such as cerebral palsy. Some neuroprotective drugs can improve adverse neurological outcomes in premature infants. A network meta-analysis was conducted to compare the effects of four common neuroprotective drugs on neurodevelopmental outcomes in preterm infants.

Methods: To identify eligible randomized clinical trials, three databases (Cochrane Library, PubMed, and EMBASE) were searched from the inception date through March 2024. Neuroprotective drugs included erythropoietin, magnesium sulfate, caffeine, and thyroxine. The primary outcome was whether neurodevelopmental impairment (NDI) or death occurred. Studies that met the eligibility criteria were assessed, and data were extracted. The surface under the cumulative ranking curve (SUCRA) was computed to evaluate and rank the neuroprotective efficacy of various drugs.

Results: Caffeine [relative risk (RR) 0.43, 95% confidence interval (CI): 0.22–0.86] showed the most promising effect in reducing the risk of premature infants with NDI. Caffeine (RR 0.55, 95% CI: 0.43–0.70) and magnesium sulfate (RR 0.66, 95% CI: 0.54–0.80) showed promising effects in reducing the risk of premature infants with cerebral palsy. According to the SUCRA value, caffeine (NDI: 87.9%, cerebral palsy: 91.2%) may be the best drug for neurodevelopmental protection in preterm infants.

Conclusions: The neuroprotective effects of caffeine were observed in the reduction of NDI and cerebral palsy, whereas magnesium sulfate, given to pregnant women at risk of preterm birth, had a protective effect in reducing the risk of cerebral palsy. The analysis of the SUCRA value indicated that caffeine may offer greater benefits compared to other medications in the neuroprotection of premature infants. The findings of this network meta-analysis ought to be regarded as theoretical rather than validated. Additional trials are necessary to validate the existing findings.

Keywords: Neuroprotective effects; caffeine; magnesium sulfate; preterm infants; network meta-analysis

Submitted Apr 19, 2025. Accepted for publication Sep 04, 2025. Published online Oct 29, 2025.

doi: 10.21037/tp-2025-260

Highlight box

Key findings

• Magnesium sulfate given to pregnant women at risk of preterm birth and caffeine given to premature infants demonstrated neuroprotective effects, potentially decreasing the likelihood of adverse neurological outcomes.

What is known and what is new?

• Four drugs, including caffeine, erythropoietin, magnesium sulfate, and thyroxine, have demonstrated neuroprotective potential for preterm infants.

• Given that premature infants are at a heightened risk for adverse neurodevelopmental outcomes, a network meta-analysis was performed to evaluate the impact of four widely used neuroprotective drugs on neurodevelopmental outcomes in preterm infants for the first time. We believe that the results of this study will provide valuable insights that can guide clinical practice and future research directions.

What is the implication, and what should change now?

• Caffeine may have the best neuroprotective effect on preterm infants after birth and may be a better choice for clinical practice.

Introduction

The gradual enhancement of neonatal treatment has significantly increased the survival rate of premature infants born at early gestational ages. However, premature delivery remains the primary cause of perinatal mortality and morbidity, and the long-term outlook for the nervous system continues to be concerning. Up to 50% of very preterm infants have adverse neurodevelopmental outcomes, including long-term neurodevelopmental disorders affecting cognition and learning or motor problems such as cerebral palsy (1,2). The prevention and treatment of poor prognosis of the nervous system in premature infants have gradually attracted wide attention. The susceptibility of immature brain cells to injury, abnormal fluctuation of cerebral blood flow, and impaired cerebral vascular autonomic regulation are the necessary pathological bases of brain injury in premature infants (3). Hypoxia, ischemia, and infection are the main causes of brain injury in premature infants. Treatment measures such as stable cerebral circulation, anti-inflammation, and anti-oxidation before, during, and after birth are currently commonly used in clinical brain protection measures for premature infants (4). At present, the available prevention and treatment options are extremely limited, and the long-term effects on neurodevelopment still need further exploration.

Prior clinical and animal studies have validated the potential advantages of various neuroprotective medications in enhancing neurodevelopment in preterm infants, including erythropoietin (EPO), caffeine, magnesium sulfate, etc. Magnesium sulfate has been used in obstetrics as a tocolytic and to prevent eclampsia. The prenatal application of magnesium sulfate may reduce the risk of cerebral palsy, which has an essential contribution to improving the long-term prognosis of premature infants (5,6). EPO is a pleiotropic cell growth factor widely distributed in various central nervous system cells. EPO plays a central role in erythropoiesis and has been widely used to prevent anemia in preterm infants (7). In addition, EPO is considered to be a neuroprotective or neurorestorative drug for the brain of premature infants because it has been shown experimentally to prevent or reduce white matter damage (8). Caffeine is one of the most widely used drugs in the neonatal intensive care unit. With the wide application of caffeine in the treatment of apnea in premature infants, its effect on improving the prognosis of nervous system in premature infants has gradually attracted the attention of neonatology in recent years. Animal experiments and clinical studies have shown that caffeine has a protective effect on the nervous system of premature infants (9,10). In recent years, it has been found that thyroxine also has a protective effect on the neurological development of premature infants. Studies have shown that early supplementation of levothyroxine 4 may improve the long-term neurodevelopment of infants born under 28 weeks (11,12).

Network meta-analysis can compare multiple treatments by combining direct and indirect evidence to assess the interrelationship between all treatments. Many previous medical studies have demonstrated its usefulness, which can unify and analyze data from randomized clinical trials to compare multiple treatments without destroying the treatments in each trial.

Previous traditional meta-analyses have shown the beneficial effects of EPO and magnesium sulfate in safeguarding the brains of preterm infants. However, there is still a lack of comparison of different neuroprotective drugs to prevent adverse neurological outcomes in preterm infants. Therefore, a network meta-analysis was conducted to compare the effects of various common neuroprotective drugs on enhancing nervous system development in preterm infants. We hope to provide a basis for selecting appropriate drugs for premature infants that exert neuroprotective effects, such as reduced rates of cerebral palsy or improved developmental scores, in clinical practice. We present this article in accordance with the PRISMA-NMA reporting checklist (available at https://tp.amegroups.com/article/view/10.21037/tp-2025-260/rc).

Methods

Search strategy

Two investigators (Y.W., L.Y.) independently searched and identified relevant studies following three data sources from inception date to March 2024, including PubMed, EMBASE and the Cochrane Library database with the terms “infant, preterm”, “brain injury”, “nervous system”, “neurodevelopment”, “neuroprotection”, “erythropoietin”, “EPO”, “magnesium sulfate”, “caffeine” and “thyroxine”. A combination of free and subject words was used when searching the electronic Databases. More details on the search strategies can be found in Figure 1.

Figure 1 Process for identifying studies eligible for the meta-analysis. BSID, Bayley Scales of Infant Development.

Selection criteria and evaluating indicators

This network meta-analysis collected studies of four neuroprotective drugs that can improve neurodevelopmental outcomes in preterm infants. They were EPO, magnesium sulfate, caffeine, and thyroxine. The inclusion criteria for studies were established based on the PICOS principle as follows—(I) participants (P): neonates with a gestational age of less than 37 weeks; (II) interventions (I): the experimental group received EPO, caffeine, or thyroxine for premature infants, or magnesium sulfate for pregnant women at risk of preterm birth. It was essential that each patient was described as using one of the four drugs evaluated in the studies; (III) the control group (C) received treatment with a placebo or other medications exclusively; (IV) outcome indicator (O): the study identified four outcomes: the count of patients exhibiting neurodevelopmental impairment (NDI) at the corrected age of 18–24 months; the mortality rate of patients at the corrected age of 18–24 months; the count of patients diagnosed with cerebral palsy at the corrected age of 18–24 months; the count of patients exhibiting a Mental Development Index (MDI) <70 on the Bayley Scales of Infant Development (BSID) (2nd edition), or a composite cognitive score <85 on the BSID (3rd edition). The primary outcomes included the number of infants with NDI, mortality. The secondary outcomes included the number of infants with cerebral palsy, an MDI <70 on the BSID (2nd edition) or a composite cognitive score <85 on the BSID (3rd edition); (V) study type (S): the design of the study was a randomized controlled trial (RCT).

NDI is characterized by the definitions provided by the authors in the articles, which may include direct testing, clinical record reviews, and parental interviews or surveys. It is also defined as survival to 18–24 months corrected age with the presence of one or more of the following criteria: (I) moderate or severe cerebral palsy; (II) BSID-II and MDI <70 points and psychomotor development index (PDI) <70 points or BSID-III composite cognitive score <85 points; (III) blindness or deafness. Cerebral palsy is a neurodevelopmental disorder characterized by abnormal muscle tone, motor, and motor skills. The diagnosis of cerebral palsy begins with a medical history and includes neuroimaging, standardized neurological, and standardized motor assessments. Cerebral palsy is graded according to the Gross Motor Function Classification System for children 2 years or younger. Moderate or severe cerebral palsy is classified as Gross Motor Function Classification System level III to V. Deafness is characterized as a hearing impairment that necessitates the use of amplification devices. Blindness is characterized by a corrected visual acuity of <20/200. The equivalence of these cognitive outcome measures was established, as research indicated that a composite cognitive score <85 (BSID-III) predicted an MDI <70 (BSID-II) with an overall agreement of 97.3% (13). To incorporate additional articles, the time range for certain articles was extended to 36–42 months or even beyond.

The criteria for excluding studies were outlined as follows: (I) develop various categories of non-RCTs; (II) comprehensive analysis, systematic evaluation, or overview of the study’s characteristics; (III) the experimental content encompasses research involving animals; (IV) investigation into the phenomenon of repeated publication; (V) only a case-reported study; (VI) conference abstracts, comments, or letters presented as literature; (VII) interventions utilizing research on multi-drug combination therapy.

Data extraction

All identified studies were imported into EndNote X9. First, studies that had been conducted multiple times were removed. Subsequently, two independent reviewers screened the titles and abstracts of each article. After that, the full texts of relevant research were reviewed according to the screening criteria. In the event of any disagreement, the third reviewer made the final decision. In instances where the comprehensive information necessary for analysis was lacking, a correspondence was initiated with the author, soliciting the requisite data. The study information (author name, date of publication and sample size, patient characteristics (gestational age, birth weight), intervention details (Intervention type, time point of intervention, specific intervention methods) and clinical neurodevelopmental outcomes were recorded.

Risk of bias assessment

Two review authors independently appraised all included studies using the Cochrane risk of bias tool. The following aspects were investigated: (I) random sequence generation; (II) allocation concealment; (III) blinding of participants and personnel; (IV) blinding of outcome assessment; (V) incomplete outcome data; (VI) selective reporting; (VII) other bias. Finally, RevMan 5.4.1 was used to draw the risk assessment diagram of literature bias.

Statistical analysis

A network meta-analysis was conducted to synthesize evidence related to neuroprotection in preterm infants and to establish a comprehensive ranking of neuroprotective drug treatments. This study employed frequency-based analysis, utilizing Stata 15.0 (STATA CORPORATION, USA) primarily for the creation of a network map, league table, and SUCRA of various interventions. Each dot in the network diagram represented an intervention. The diameter of the dot represented the total sample size of the intervention. The straight line connecting the two interventions indicated that there was evidence of direct comparison between the two interventions. The thickness of the straight line indicated the number of documents included in the comparison of the two interventions. In the absence of a direct connection, there was no evidence to support a direct comparison between the two interventions. All treatment comparisons for each outcome were illustrated using a network graph. The relative risk (RR) was used as the effect analysis statistic and its 95% confidence interval (CI) was provided. If the study formed a closed loop, the node fissure method was used to test the local inconsistency, and the inconsistency between direct evidence and indirect evidence was judged. Treatment strategies were ranked for each outcome by the surface under the cumulative ranking curve (SUCRA) probabilities, and higher SUCRA probabilities in each simulation indicated a higher chance of being the best treatment regimen. Publication bias was evaluated in studies that contributed to primary outcomes and adverse events through a visual inspection of the generated funnel plots.

Results

Characteristics of the included studies

The search strategy employed in this study yielded a total of 2,874 relevant studies during the initial examination. After screening titles or abstracts and removing duplicate studies, the full texts of 71 potentially eligible studies were obtained. Ultimately, 26 randomized clinical trials were included in the quantitative data synthesis. Figure 1 shows the literature retrieval and research selection process of the system. Table 1 shows a summary of the characteristics of the included trials. Among the 26 RCTs included, 17 on NDI, 17 addressed mortality, 19 studies reported on cerebral palsy, and 14 utilized the BSID. The sample size was 24–2,136. These studies were published in journals written in the English language from 2002 to 2021. Only one study control group was aminophylline, and the remaining studies were placebo controls. Different interventions were used in 26 randomized clinical trials, including EPO (n=8), magnesium sulfate (n=8), caffeine (n=6) and thyroxine (n=4).

Table 1

Characteristics of included studies

Study n GA, week BW, g Time points of intervention Intervention Control group Outcome
Ohls et al., 2004 (14) 102 ≤32 0/7 ≤1,000 24–96 h of age rhEPO 400 IU/kg IV or SC, 3 times per week until 35 0/7 weeks’ postmenstrual age Placebo ①③ ④
Ohls et al., 2014 (15) 53a 28 [26–29]b 500–1,250 ≤48 h of age rhEPO 400 IU/kg SC, 3 times per week until 35 0/7 weeks’ postmenstrual age Placebo ① ② ③ ④
Natalucci et al., 2016 (16) 365 26 0/7 to 31 6/7 1,220±327c <3 h of age rhEPO 3,000 IU/kg IV at <3, 12–18, and 36–42 h of age Placebo ① ③ ④
Song et al., 2016 (17) 613 ≤32 0/7 1,372±209c <72 h of age rhEPO 500 IU/kg IV every other day for 2 weeks Placebo ① ② ③ ④
Juul et al., 2020 (18) 628 24 0/7 to 27 6/7 806.4±194.6c ≤24 h of age rhEPO 1,000 IU/kg IV every 48 h for 6 doses, followed by 400 IU SC 3 times per week until 32 6/7 weeks’ postmenstrual age Placebo ① ② ③ ④
Song et al., 2021 (19) 316 ≤32 1,141±230c Within 72 h after birth rhEPO 500 IU/kg IV every other day for 2 weeks Placebo ① ② ③ ④
Vu et al., 2021 (20) 628 24 0/6 to 27 6/7 819.4±191.4c ≤24 h of age rhEPO 1,000 IU/kg IV every 48 h for 6 doses, followed by 400 IU SC 3 times per week until 32 6/7 weeks’ postmenstrual age Placebo ① ④
Peltoniemi et al., 2017 (21) 35 24 0/7 to 30 0/7 1,200 [1,030–1,350]b 1st day of life rhEPO 250 IU/kg IV daily from days 1–6 Placebo ② ③
Crowther et al., 2003 (22) 1,225 <30 Any BW The birth was planned or expected within 24 hours A loading infusion of 8 mL (4 g) of 0.5 g/mL of magnesium sulfate solution for 20 minutes followed by a maintenance infusion of 2 mL/h for up to 24 hours Placebo ① ② ③
Magpie Trial Follow-Up Study Collaborative Group. 2007 (23) 1,593 <37 Any BW The women were eligible for the trial if they had preeclampsia during pregnancy or if the birth was planned or expected within 24 hours Magnesium sulfate dose 4 g intravenously over 10–15 minutes, followed by either 1 g/hour intravenously for 24 hours, or by 5 g every 4 hours intramuscularly for 24 hours Placebo ② ③
Marret et al., 2007 (24) 688 <33 1,350 [1,080–1,670]b The birth was planned or expected within 24 hours A single 40 mL infusion of 0.1 g/mL of MgSO4 (4 g) solution over 30 minutes Placebo ② ③
Mittendorf et al., 2002 (25) 165 25–33 Any BW At risk for preterm
delivery		 MgSO4 was given as a 4 g bolus followed by an infusion of 2–3 g of MgSO4 per hour Placebo ② ③
Rouse et al., 2008 (26) 2,136 29.8±3.1c Any BW At imminent risk for delivery 6 g MgSO4 bolus followed by a constant infusion of 2 g per hour Placebo ② ③ ④
Wolf et al., 2020 (27) 680 24–31 1,498±691c At risk for preterm
delivery		 A loading dose of 5g MgSO4 followed by 1 g/hour in identical volumes Placebo ② ③
Kamyar et al., 2016 (28) 396 29.1±2.8c Any BW At risk for preterm
delivery		 A loading dose infusion of 6 g MgSO4 over 20–30 minutes, followed by a maintenance infusion of 2 g per hour Placebo ① ③ ④
Doyle et al., 2014 (29) 1,254 27.3±2.2c 1,053±389c The birth was planned or expected within 24 hours A loading dose of 4 g MgSO4; 1 g/h maintenance for up to a maximum of 24 hours Placebo ② ③
Mürner-Lavanchy
et al., 2018 (30)		 870 27.3±1.7c 500–1,250 During the first 10 days of life A loading dose of 20 mg of caffeine citrate per kilogram of body weight was followed by a daily maintenance dose of 5 mg per kilogram. If apnea persists, the daily maintenance dose could be increased to a maximum of 10 mg of caffeine citrate per kilogram Placebo ① ③
Schmidt et al., 2017 (31) 920 27.4±1.7c 500–1,250 During the first 10 days of life A loading dose of 20 mg of caffeine citrate per kilogram of body weight was followed by a daily maintenance dose of 5 mg per kilogram. If apnea persists, the daily maintenance dose could be increased to a maximum of 10 mg of caffeine citrate per kilogram Placebo ① ③
Schmidt et al., 2012 (32) 1,640 27.4±1.8c 500–1,250 During the first 10 days of life A loading dose of 20 mg of caffeine citrate per kilogram of body weight was followed by a daily maintenance dose of 5 mg per kilogram. If apnea persists, the daily maintenance dose could be increased to a maximum of 10 mg of caffeine citrate per kilogram Placebo ① ② ③ ④
Schmidt et al., 2007 (33) 1,869 27.4±1.8c 500–1,250 3 days A loading dose of 20 mg of caffeine citrate per kilogram of body weight was followed by a daily maintenance dose of 5 mg per kilogram. If apnea persists, the daily maintenance dose could be increased to a maximum of 10 mg of caffeine citrate per kilogram Placebo ① ② ③ ④
Khurana et al., 2017 (34) 79 ≤34 1,203.84±274.15c The first 36 h of life A loading dose of 20 mg/kg of caffeine citrate
(10 mg/kg caffeine base) diluted in 5% dextrose and given for 30 minutes then continue on a maintenance dose of 5 mg/kg (2.5 mg caffeine base) 24th hourly IV or oral preparation of cafirate solution 20 mg/mL if no adequate response was achieved, then the dose was optimized up to 7.5 mg/kg		 Aminophylline ④
Gupte et al., 2016 (12) 93 26 [23–31]b 800 [410–1,210]b The first 48 h of life Caffeine initial dose Placebo ④
Ng et al., 2020 (35) 59 25.8±1.3c 820.7±183.9c During the first 5 days after birth Thyroxine was given intravenously during the first 5 days after birth and then changed to oral LT4, when enteral feeds were fully established (8 μg/kg birth weight/day as a single daily dose) until 32 weeks corrected gestational age Placebo ① ②
van Wassenaer et al., 2015 (36) 200 197±8c,d 1,078±218c Starting 12–24 hours after birth Thyroxine (8 μg per kilogram of birth weight) was administered daily for six weeks Placebo ① ②
Van Wassenaer et al., 2005 (37) 113 197±9c,d 1,113±225c The 24th hour of life Thyroxine was given during the first 6 weeks of life in a dose of 8 μg/kg birth weight per day Placebo ①
van Wassenaer-Leemhuis et al., 2014 (38) 24 26.1±0.9c,e 863±122c Postnatal day 1 to postnatal day 42 Thyroxine was given 8 μg/(kg·d) either as an intravenous bolus injection (1 mL/kg in 5% dextrose). When converting to oral dosing via gavage, bolus administration was kept as a once-a-day regimen Placebo ① ② ④

① NDI; ② mortality; ③ cerebral palsy; ④ MDI <70 (BSID-II) or a composite cognitive score <85 (BSID-III). Magnesium sulfate is given to pregnant women, caffeine, thyroxine, EPO is given to premature infants. a, this study had three groups: rhEPO (n=29) vs. placebo (n=24) vs. darbepoetin (n=27). The darbepoetin group was not included in the meta-analysis. b, median [interquartile range] GA or BW of included infants. c, mean ± SD GA or BW of included infants. d, the data unit is days. e, the trial’s endpoints were thyroid hormone (thyroxine, T4) and thyrotropin plasma concentrations in eight study arms: six treated with T4 [4, 8, and 16 μg/(kg·d)], bolus or continuous], one treated with iodine only, and one treated with placebo. Only the placebo group and thyroxine 8 μg/(kg·d) bolus group was selected. BSID, Bayley Scales of Infant Development; BW, birth weight; EPO, erythropoietin; GA, gestational age; IU, international units; IV, intravenously; LT4, levothyroxine; MDI, Mental Development Index; NDI, neurodevelopmental impairment; rhEPO, recombinant human erythropoietin; SC, subcutaneously; SD, standard deviation.

Risk of bias

There was no problem of insufficient sequence generation in 26 randomized clinical trials. Regarding allocation concealment, 17 trials (65.4%) used methods such as opaque envelopes or central randomization systems, and the risk was low. In terms of blind method, 13 trials (50.0%) had a blind method for subjects and personnel, and 4 trials (11.5%) had a blind method for outcome evaluation. All randomized clinical trials had a low risk of selective reporting bias, incomplete outcome data, and other biases. Figures 2,3 illustrate the summary assessment of bias risk.

Figure 2 Bias risk assessment chart of included studies.
Figure 3 Bias risk assessment diagram of a single included study.

Network meta-analysis

The qualified comparison network of primary and secondary results is shown in Figure 4. Trial network plots were shown in Figure 4A-4D. The dimensions of the treatment nodes indicated the relative number of patients allocated to the treatment group. It was found that magnesium sulfate had the largest sample size compared with the other three drugs in this network meta-analysis. It was found that the samples of placebo ranked the highest in this network meta-analysis. The thickness of the lines usually reflects the number of studies that directly compare the two interventions. Among the studies that included the outcome of NDI, the studies of EPO compared with placebo were the most (Figure 4A). Among the studies that included the outcome of mortality, the studies of magnesium sulfate compared with placebo were the most (Figure 4B). Among the studies that included the outcome of cerebral palsy, the studies of magnesium sulfate compared with placebo were the most (Figure 4C). Among the studies that included the outcome of an MDI <70 on the BSID (2nd edition) or a composite cognitive score <85 on the BSID (3rd edition), the studies of EPO compared with placebo were the most (Figure 4D). Nonetheless, there was no direct comparison made between any two neuroprotective drugs. This network meta-analysis was carried out to evaluate both direct and indirect comparisons. Since no closed loop was formed in each network graph, the possible inconsistencies in the network meta-analysis were not tested, and only the consistency model was selected.

Figure 4 Network map. (A) NDI; (B) mortality; (C) cerebral palsy; (D) MDI <70 (BSID-II) or composite cognitive score <85 (BSID-III). BSID, Bayley Scales of Infant Development; MDI, Mental Development Index; NDI, neurodevelopmental impairment.

The significant results of this network meta-analysis were NDI and cerebral palsy. Figure 5A,5B summarize the network meta-analysis comparison results of NDI and cerebral palsy. Caffeine (RR 0.43, 95% CI: 0.22–0.86) was more effective than placebo in reducing the risk of NDI (Figure 5A). Caffeine reduced the risk of NDI in premature infants and played a neuroprotective role. Magnesium sulfate (RR 0.66, 95% CI: 0.54–0.80) and caffeine (RR 0.55, 95% CI: 0.43–0.70) were more effective than placebo in reducing the risk of cerebral palsy (Figure 5B). These two drugs reduced the risk of cerebral palsy in premature infants and played a neuroprotective role. The analysis revealed no notable disparity in the occurrence of NDI or cerebral palsy among the various neuroprotective agents, as indicated by the 95% CI encompassing the value of 1.

Figure 5 Summary of results from network meta-analysis on significant outcomes. (A) NDI. (B) Cerebral palsy. Significant results are shown in boldface type. CI, confidence interval; EPO, erythropoietin; NDI, neurodevelopmental impairment; RR, relative risk.

SUCRA was conducted to evaluate the comparative efficacy of various neuroprotective agents. SUCRA indicates the ranking position of the drug in comparison to all other interventions for the same outcome, rather than its absolute effect. The larger the area under the curve, the higher the probability of SUCRA, indicating the greater the possibility of becoming the best treatment plan. The effects of all drugs on NDI were ranked with SUCRA probabilities (Figure 6A). Seventeen studies involving 9,962 patients evaluated the effect of EPO, caffeine, magnesium sulfate, and thyroxine on NDI. Caffeine also ranked first (SUCRA 87.9%) in NDI and showed satisfactory efficacy in reducing NDI. EPO (SUCRA 60.5%) and thyroxine (SUCRA 59.5%) ranked second and third. Magnesium sulfate ranked last (SUCRA 20.6%), and placebo ranked fourth (SUCRA 21.5%). The effects of all drugs on cerebral palsy were ranked with SUCRA probabilities (Figure 6B). Data regarding the effects of EPO, caffeine, and magnesium sulfate on cerebral palsy were available from nineteen trials with 16,348 patients. Caffeine ranked first (SUCRA 91.2%) in cerebral palsy and showed satisfying efficacy in reducing cerebral palsy, followed by magnesium sulfate (SUCRA 57.5%). EPO ranked third (SUCRA 49.7%) and placebo ranked last (SUCRA 1.6%). According to the SUCRA value, caffeine (NDI: 87.9%, cerebral palsy: 91.2%) showed the highest SUCRA value (Table 2). The data analysis results indicated that caffeine appeared to provide the most significant protective effect on the neural development of premature infants among the four drugs studied.

Figure 6 Ranking of treatment strategies based on the probability of their protective effects on outcomes of NDI and cerebral palsy according to the cumulative ranking area (SUCRA). (A) NDI (SUCRA). (B) Cerebral palsy (SUCRA). NDI, neurodevelopmental impairment; SUCRA, surface under the cumulative ranking curve.

Table 2

SUCRA value for treatment ranking

Treatment NDI Cerebral palsy
Placebo 0.215 0.016
EPO 0.605 0.496
Magnesium sulfate 0.206 0.575
Caffeine 0.879 0.912
Thyroxine 0.595

The SUCRA value is a representative number of the overall ranking, and a higher SUCRA value indicates a higher probability. The highest values of SUCRA are in italics. EPO, erythropoietin; NDI, neurodevelopmental impairment; SUCRA, surface under the cumulative ranking curve.

Funnel plots of primary outcomes

STATA drew a funnel plot to identify whether there was a small sample effect in the study and the different colors in the plot. The dots represented the direct comparison between two different drugs, and the number of dots of the same color represented the corresponding. The funnel plot was distributed symmetrically, indicating no publication bias or small sample effect. Figure 7 shows the asymmetric funnel plot of NDI. Five studies fell outside the 95% CI, indicating that the heterogeneity between studies may cause the asymmetry of the funnel plot. Figure 8 shows the funnel plot of cerebral palsy was symmetric.

Figure 7 Funnel plots of NDI. NDI, neurodevelopmental impairment.
Figure 8 Funnel plots of cerebral palsy.

Discussion

Very preterm infants are at risk of death and disability. Despite improvements in the survival rate of this population, evidence indicates that there have not been significant advancements in overall early neurodevelopmental outcomes in recent decades. The primary negative neurological outcomes consist of NDI and cerebral palsy. They seriously affect the quality of life of the children and bring economic and social burden. The aim of the study has consistently been to minimize the incidence of adverse neurological outcomes in premature infants. Neonatal pediatricians have been looking for a suitable neuroprotective drug.

This network meta-analysis aimed to evaluate four drugs in reducing adverse neurodevelopmental outcomes of preterm infants. It demonstrated that caffeine and magnesium sulfate reduced cerebral palsy in premature infants, and caffeine reduced NDI in preterm infants. Caffeine may be more beneficial to the neurodevelopment of premature infants than magnesium sulfate, thyroxine, and EPO, and it has a protective effect. This conclusion will help us to choose caffeine as a neuroprotective drug to prevent adverse neurological outcomes in premature infants. The four neuroprotective drugs examined have a long history of clinical use, demonstrating safety in premature infants without any apparent adverse reactions.

This network meta-analysis constituted the best available evidence for comparing the efficacy of four common neuroprotective drugs in reducing adverse neurological outcomes. This study assessed the main neurodevelopmental outcomes in preterm infants, specifically focusing on NDI and mortality. In addition, the secondary outcomes, including cerebral palsy and MDI <70 (BSID-II) or a composite cognitive score <85 (BSID-III), were also evaluated to identify the efficacy of four different drugs. Finally, 26 RCTs involving 16,744 patients were included.

In this network meta-analysis, there was a decrease in the number of preterm infants with cerebral palsy born to mothers using magnesium sulfate compared to placebo. This result was consistent with the results of a systematic review and meta-analysis in 2020. A systematic review with meta-analysis conducted in 2020 indicated that administering magnesium sulfate to women at risk of preterm birth could lower the risk of cerebral palsy in their offspring (6). There is no traditional meta-analysis to evaluate the effect of caffeine on the development of the nervous system in premature infants. Only some clinical trials or animal experiments have proved that it was beneficial to develop the nervous system in premature infants (39,40). Early caffeine treatment was associated with shorter respiratory support time and reduced cerebral palsy, intracranial hemorrhage, death, and complications (41,42). This network meta-analysis indicated that caffeine therapy in preterm infants reduced the number of NDIs better than placebo. This network meta-analysis indicated that caffeine therapy in preterm infants led to a reduction in the incidence of cerebral palsy compared to placebo. In the context of cerebral palsy and NDI, no significant difference in the incidence of adverse outcomes was observed between any two neuroprotective drugs, potentially attributable to the limited sample size in certain studies.

EPO markedly suppressed microglial activation and diminished the expression of pro-inflammatory factors. In a brain injury model using premature rats induced by lipopolysaccharide, EPO demonstrated a reduction in the inflammatory response within the white matter and facilitated improvements in myelin formation defects through the activation of the JAK2/STAT3 pathway (43,44). A meta-analysis of EPO neuroprotection in very preterm infants in 2021 showed no effect of recombinant human EPO (rhEPO) on mortality or MDI <70 or a composite cognitive score <85 or NDI (45). The exploratory subgroup analyses indicated a positive effect of rhEPO on MDI <70 or a composite score <85, as well as a borderline positive effect on any NDI, specifically within the subgroup of trials that utilized low to moderate-dose rhEPO. However, this network meta-analysis showed that there was no significant difference between the comparison of EPO and placebo in cerebral palsy, NDI, mortality, and MDI <70 (BSID-II) or a composite score <85 (BSID-III). The inconsistency of the results may be related to the inclusion of more literature based on high-dose EPO treatment in this study. The efficacy of high-dose EPO may have a safety upper limit, and it would lose its neuroprotective effect at high concentrations. Therefore, the optimal dose and total course of EPO need more high-quality RCT and long-term follow-up evaluation data to clarify further. In this study, thyroxine had no effect on adverse neurodevelopmental outcomes in preterm infants.

The ranking results of a network meta-analysis on NDI and cerebral palsy suggested that caffeine was most likely to become an effective choice (which was measured by response rate according to the SUCRA value). EPO ranked second after caffeine in reducing the number of patients with NDI. Magnesium sulfate also reduced the number of patients with cerebral palsy. This network meta-analysis indicated that caffeine may be a more practical option for neuroprotection in preterm infants compared to the three other neuroprotective drugs, thereby assisting the study in selecting the suitable neuroprotective drugs. Early caffeine therapy was more conducive to protecting the neurodevelopment of premature infants and improving prognosis.

The mechanism of caffeine’s protective effect on the brain of premature infants may be as follows: as an adenosine receptor blocker, caffeine may play a brain-protective role by blocking A1 receptors and improving the myelination disorder of nerve cells stimulated by undesirable factors; caffeine can increase the synthesis of cAMP in neurons and affect the development and function of neurons by acting on Ca2+/cAMP response element binding protein; as a free radical scavenger, it can reduce the release of oxidative stress reactants, down-regulate inflammatory factors, reduce the degradation of extracellular matrix and the generation of pro-apoptotic substances, and promote the generation of antioxidant substances (46-49). Caffeine is currently mainly used in the prevention and treatment of apnea in premature infants. Neonatal caffeine treatment of apnea in premature infants improved visual movement, visual perception, and visual spatial ability at the age of 11 years (29). The beneficial effects of caffeine on cardiopulmonary physiology in stabilizing systemic circulation and cerebral hemodynamics and its ability to alleviate hypoxia and reduce respiratory depression, may play a role in neuroprotection. In addition, caffeine also has lung protection compared with other neuroprotective drugs (50), which can reduce the incidence of bronchopulmonary dysplasia in premature infants. Bronchopulmonary dysplasia is also a common and severe complication for premature infants, which seriously affects the quality of life. Early caffeine use can bring significant and long-term benefits to premature infants. An animal study showed that caffeine improved long-term cognitive function in neonatal rats with hypoxic-ischemic white matter damage (51). The mechanism was that caffeine inhibited the activation of NOD-like receptor thermal protein domain-associated protein 3 through A2aR, reduced the activation of microglia, regulated the phenotypic polarization of microglia and the release of inflammatory factors, and improved myelin development. Caffeine improved mitochondrial dysfunction in the white matter of hypoxic-ischemic neonatal rats by deacetylation (52).

The smaller the gestational age, the better the brain-protective effect of caffeine. It is recommended that very preterm infants should be given caffeine as early as possible after birth (preferably within 3 days) until the corrected gestational age of 33–35 weeks. Caffeine had no adverse effect on general intelligence, attention and behavior, which highlighted the long-term safety of caffeine in very low birth weight premature infants (29). The incidence of adverse reactions in average doses is relatively low. Through rigorous clinical trials and economic evaluations, neonatal pediatricians can still ensure that caffeine is effective and represents a good use of social resources (53). If a higher maintenance dose is used clinically, therapeutic drug monitoring is recommended, and the side effects of caffeine, such as irritability, hyperglycemia, and tachycardia, are closely monitored (54).

The findings of our study further validated that magnesium sulfate prenatal administration was advantageous for their neurodevelopment in preterm infants. The treatment is safe for pregnant women. It seems more reasonable to use magnesium sulfate in pregnant women at risk of preterm birth. Animal research proved that magnesium pre-treatment reduced neuronal apoptosis in newborn rats in hypoxia-ischemia (55). The mean blood flow velocity of the middle cerebral artery in preterm infants who were exposed to magnesium sulfate showed a significant increase, suggesting an enhancement in cerebral blood flow. This enhancement had the potential to lower the likelihood of ischemic brain injury through the dilation of cerebral vessels or the optimization of perfusion (56). Exposure to magnesium sulfate was linked to a reduction in cerebral oxygen extraction efficiency, indicating a potential decrease in brain oxygen demand or an enhancement in oxygen utilization efficiency. This effect operated independently of the mechanisms involved in the redistribution of cerebral blood flow in preeclampsia, potentially linked to the direct regulation of cerebrovascular tension or neuronal metabolism (57). Magnesium sulfate neuroprotection mechanism was placental mediated by inhibition of inflammation, apoptosis and oxidative stress (58). A cost-effectiveness analysis of magnesium sulfate for fetal neuroprotection found that the administration of magnesium sulfate to patients in whom preterm birth is a dominant and cost-effective strategy (59). However, in developing countries, the use of magnesium sulfate is often limited. In addition, further exploration is needed to optimize the dosage regimen. Magnesium sulfate reduced the incidence of cerebral palsy but had little neuroprotective effect on NDI. Magnesium sulfate is thought to be suitable for prenatal administration only, whereas brain development related to NDI extends from the fetus into the neonatal period and necessitates ongoing intervention. A clinical study found that prenatal magnesium sulfate was beneficial to the cognitive outcomes of female premature infants, but led to poor prognosis of male premature infants (60). We need to pay more attention to the role of gender differences in the prevention of adverse neurodevelopmental outcomes in the neonatal population. Future studies need to consider the impact of gender factors to ensure the best care for high-risk neonatal populations.

The findings of the network meta-analysis revealed variability across the studies, which may account for the asymmetry noted in the funnel plot. The analysis revealed that clinical heterogeneity could be linked to differences in gestational age, timing of intervention, and dosage. The literature reviewed did not address whether it was used in conjunction with other neuroprotective strategies (such as melatonin, prenatal steroid hormones, or hypothermia therapy in premature infants older than 36 weeks), which could also contribute to the observed clinical heterogeneity. Data regarding “male potential risk” and “regulation of inflammatory state” is currently inadequate, and future research will prioritize the investigation of these subgroups. The neuroprotective effects of caffeine and magnesium sulfate may exhibit subgroup heterogeneity, necessitating individual evaluation of exposure dose and associated risks. There was a certain overlap and statistical repetition in our study, which may lead to the overall effect of intervention on NDI being exaggerated. Future research will conduct a comprehensive hierarchical analysis of NDI.

The limitations of this study are as follows: (I) there were few studies on the thyroxine drug group, and the number was relatively small. This limited the ability to draw reliable conclusions about its neuroprotective effects. In addition, some infants may receive more than one treatment in clinical practice. Due to the heterogeneity of the included study design, we failed to fully separate the independent effects of each intervention on the neurodevelopmental outcomes of preterm infants, and did not report the overlapping medications in a stratified manner. This may lead us to overestimate or underestimate the true role of the drug and fail to identify potential synergistic or antagonistic effects; (II) the timing, dosage, frequency of use, course of treatment, and follow-up time of some included studies were not wholly consistent; (III) the collected studies did not offer a direct comparison among the four drugs, especially caffeine and EPO; thus, the consistency model was deemed the suitable approach. Additional large-scale clinical studies may be necessary to enhance robustness in the future; (IV) no side effects were compared. More clinical large-sample randomized controlled studies will be done to explore related side effects; (V) it was too early to assess neurological development at 2 years of age, and it was more meaningful to evaluate the cognitive ability of preschool children to help comprehensively evaluate the long-term neurodevelopmental outcomes of children. Missing data prevented us from including this study in a network meta-analysis. It will be explored in the following clinical research in the future. The combined outcome of “NDI and/or cerebral palsy” may be utilized in future investigations to evaluate the risk of neurodevelopment in greater detail; (VI) our study did not consider the differences in geographical location or medical system; (VII) sex-specific treatment differences were not analyzed due to a lack of stratified data.

In summary, the use of neuroprotective drugs to improve the neurodevelopmental outcome of premature infants has been confirmed. According to the results of this study, caffeine can be used as an effective and safe drug for the neurodevelopmental protection of premature infants. Whether it can become the first choice to improve the neurodevelopmental outcome of premature infants requires more high-quality clinical evidence to further confirm.

Conclusions

Caffeine demonstrated a neuroprotective effect by mitigating cerebral palsy and NDI, whereas magnesium sulfate, given to pregnant women at risk of preterm birth, had a protective effect in reducing the risk of cerebral palsy. Caffeine had neuroprotective effects on premature infants and reduced the occurrence of adverse neurological outcomes. The SUCRA value suggested that caffeine could offer greater benefits compared to other medications in protecting the neurological health of premature infants. In the future, it is necessary to further design high-quality, large-sample, multicenter RCT studies to explore the protective effects of caffeine, magnesium sulfate, EPO, and other neuroprotective treatments alone or in combination on the nervous system of premature infants and to explore the optimal time, effective dose, and intake time of drug use.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the PRISMA-NMA reporting checklist. Available at https://tp.amegroups.com/article/view/10.21037/tp-2025-260/rc

Peer Review File: Available at https://tp.amegroups.com/article/view/10.21037/tp-2025-260/prf

Funding: This study received funding from the National Key R&D Program of China (No. 2024YFC2707700, sub-project No. 2024YFC2707704) and the National Natural Science Foundation of China (No. 82201896).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tp.amegroups.com/article/view/10.21037/tp-2025-260/coif). All authors report that this study received funding from the National Key R&D Program of China (No. 2024YFC2707700, sub-project No. 2024YFC2707704) and the National Natural Science Foundation of China (No. 82201896). The authors have no other conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

  1. Anderson PJ, De Luca CR, Hutchinson E, et al. Attention problems in a representative sample of extremely preterm/extremely low birth weight children. Dev Neuropsychol 2011;36:57-73. [Crossref] [PubMed]
  2. Glass HC, Fujimoto S, Ceppi-Cozzio C, et al. White-matter injury is associated with impaired gaze in premature infants. Pediatr Neurol 2008;38:10-5. [Crossref] [PubMed]
  3. Bano S, Chaudhary V, Garga UC. Neonatal Hypoxic-ischemic Encephalopathy: A Radiological Review. J Pediatr Neurosci 2017;12:1-6. [Crossref] [PubMed]
  4. Fleiss B, Van Steenwinckel J, Bokobza C, et al. Microglia-Mediated Neurodegeneration in Perinatal Brain Injuries. Biomolecules 2021;11:99. [Crossref] [PubMed]
  5. Wolf HT, Hegaard HK, Greisen G, et al. Treatment with magnesium sulphate in pre-term birth: a systematic review and meta-analysis of observational studies. J Obstet Gynaecol 2012;32:135-40. [Crossref] [PubMed]
  6. Wolf HT, Huusom LD, Henriksen TB, et al. Magnesium sulphate for fetal neuroprotection at imminent risk for preterm delivery: a systematic review with meta-analysis and trial sequential analysis. BJOG 2020;127:1180-8. [Crossref] [PubMed]
  7. Ohlsson A, Aher SM. Early erythropoietin for preventing red blood cell transfusion in preterm and/or low birth weight infants. Cochrane Database Syst Rev 2014;CD004863.
  8. Juul SE, Pet GC. Erythropoietin and Neonatal Neuroprotection. Clin Perinatol 2015;42:469-81. [Crossref] [PubMed]
  9. Back SA, Craig A, Luo NL, et al. Protective effects of caffeine on chronic hypoxia-induced perinatal white matter injury. Ann Neurol 2006;60:696-705. [Crossref] [PubMed]
  10. Endesfelder S, Weichelt U, Strauß E, et al. Neuroprotection by Caffeine in Hyperoxia-Induced Neonatal Brain Injury. Int J Mol Sci 2017;18:187. [Crossref] [PubMed]
  11. Ng SM, Turner MA, Gamble C, et al. Effect of thyroxine on brain microstructure in extremely premature babies: magnetic resonance imaging findings in the TIPIT study. Pediatr Radiol 2014;44:987-96. [Crossref] [PubMed]
  12. Gupte AS, Gupta D, Ravichandran S, et al. Effect of early caffeine on neurodevelopmental outcome of very low-birth weight newborns. J Matern Fetal Neonatal Med 2016;29:1233-7. [Crossref] [PubMed]
  13. Johnson S, Moore T, Marlow N. Using the Bayley-III to assess neurodevelopmental delay: which cut-off should be used? Pediatr Res 2014;75:670-4. [Crossref] [PubMed]
  14. Ohls RK, Ehrenkranz RA, Das A, et al. Neurodevelopmental outcome and growth at 18 to 22 months' corrected age in extremely low birth weight infants treated with early erythropoietin and iron. Pediatrics 2004;114:1287-91. [Crossref] [PubMed]
  15. Ohls RK, Kamath-Rayne BD, Christensen RD, et al. Cognitive outcomes of preterm infants randomized to darbepoetin, erythropoietin, or placebo. Pediatrics 2014;133:1023-30. [Crossref] [PubMed]
  16. Natalucci G, Latal B, Koller B, et al. Effect of Early Prophylactic High-Dose Recombinant Human Erythropoietin in Very Preterm Infants on Neurodevelopmental Outcome at 2 Years: A Randomized Clinical Trial. JAMA 2016;315:2079-85. [Crossref] [PubMed]
  17. Song J, Sun H, Xu F, et al. Recombinant human erythropoietin improves neurological outcomes in very preterm infants. Ann Neurol 2016;80:24-34. [Crossref] [PubMed]
  18. Juul SE, Vu PT, Comstock BA, et al. Effect of High-Dose Erythropoietin on Blood Transfusions in Extremely Low Gestational Age Neonates: Post Hoc Analysis of a Randomized Clinical Trial. JAMA Pediatr 2020;174:933-43. [Crossref] [PubMed]
  19. Song J, Wang Y, Xu F, et al. Erythropoietin Improves Poor Outcomes in Preterm Infants with Intraventricular Hemorrhage. CNS Drugs 2021;35:681-90. [Crossref] [PubMed]
  20. Vu PT, Ohls RK, Mayock DE, et al. Transfusions and neurodevelopmental outcomes in extremely low gestation neonates enrolled in the PENUT Trial: a randomized clinical trial. Pediatr Res 2021;90:109-16. [Crossref] [PubMed]
  21. Peltoniemi OM, Anttila E, Kaukola T, et al. Randomized trial of early erythropoietin supplementation after preterm birth: Iron metabolism and outcome. Early Hum Dev 2017;109:44-9. [Crossref] [PubMed]
  22. Crowther CA, Hiller JE, Doyle LW, et al. Effect of magnesium sulfate given for neuroprotection before preterm birth: a randomized controlled trial. JAMA 2003;290:2669-76. [Crossref] [PubMed]
  23. Magpie Trial Follow-Up Study Collaborative Group. The Magpie Trial: a randomised trial comparing magnesium sulphate with placebo for pre-eclampsia. Outcome for women at 2 years. BJOG 2007;114:300-9. [Crossref] [PubMed]
  24. Marret S, Marpeau L, Zupan-Simunek V, et al. Magnesium sulphate given before very-preterm birth to protect infant brain: the randomised controlled PREMAG trial*. BJOG 2007;114:310-8. [Crossref] [PubMed]
  25. Mittendorf R, Dambrosia J, Pryde PG, et al. Association between the use of antenatal magnesium sulfate in preterm labor and adverse health outcomes in infants. Am J Obstet Gynecol 2002;186:1111-8. [Crossref] [PubMed]
  26. Rouse DJ, Hirtz DG, Thom E, et al. A randomized, controlled trial of magnesium sulfate for the prevention of cerebral palsy. N Engl J Med 2008;359:895-905. [Crossref] [PubMed]
  27. Wolf HT, Brok J, Henriksen TB, et al. Antenatal magnesium sulphate for the prevention of cerebral palsy in infants born preterm: a double-blind, randomised, placebo-controlled, multi-centre trial. BJOG 2020;127:1217-25. [Crossref] [PubMed]
  28. Kamyar M, Manuck TA, Stoddard GJ, et al. Magnesium sulfate, chorioamnionitis, and neurodevelopment after preterm birth. BJOG 2016;123:1161-6. [Crossref] [PubMed]
  29. Doyle LW, Anderson PJ, Haslam R, et al. School-age outcomes of very preterm infants after antenatal treatment with magnesium sulfate vs placebo. JAMA 2014;312:1105-13. [Crossref] [PubMed]
  30. Mürner-Lavanchy IM, Doyle LW, Schmidt B, et al. Neurobehavioral Outcomes 11 Years After Neonatal Caffeine Therapy for Apnea of Prematurity. Pediatrics 2018;141:e20174047. [Crossref] [PubMed]
  31. Schmidt B, Roberts RS, Anderson PJ, et al. Academic Performance, Motor Function, and Behavior 11 Years After Neonatal Caffeine Citrate Therapy for Apnea of Prematurity: An 11-Year Follow-up of the CAP Randomized Clinical Trial. JAMA Pediatr 2017;171:564-72. [Crossref] [PubMed]
  32. Schmidt B, Anderson PJ, Doyle LW, et al. Survival without disability to age 5 years after neonatal caffeine therapy for apnea of prematurity. JAMA 2012;307:275-82. [Crossref] [PubMed]
  33. Schmidt B, Roberts RS, Davis P, et al. Long-term effects of caffeine therapy for apnea of prematurity. N Engl J Med 2007;357:1893-902. [Crossref] [PubMed]
  34. Khurana S, Shivakumar M, Sujith Kumar Reddy GV, et al. Long-term neurodevelopment outcome of caffeine versus aminophylline therapy for apnea of prematurity. J Neonatal Perinatal Med 2017;10:355-62. [Crossref] [PubMed]
  35. Ng SM, Turner MA, Weindling AM. Neurodevelopmental Outcomes at 42 Months After Thyroxine Supplementation in Infants Below 28 Weeks' Gestation: A Randomized Controlled Trial. Thyroid 2020;30:948-54. [Crossref] [PubMed]
  36. van Wassenaer AG, Kok JH, de Vijlder JJM, et al. Effects of thyroxine supplementation on neurologic development in infants born at less than 30 weeks gestation. N Engl J Med 1997;336:21-6. [Crossref] [PubMed]
  37. van Wassenaer AG, Westera J, Houtzager BA, et al. Ten-year follow-up of children born at <30 weeks' gestational age supplemented with thyroxine in the neonatal period in a randomized, controlled trial. Pediatrics 2005;116:e613-8. [Crossref] [PubMed]
  38. van Wassenaer-Leemhuis A, Ares S, Golombek S, et al. Thyroid hormone supplementation in preterm infants born before 28 weeks gestational age and neurodevelopmental outcome at age 36 months. Thyroid 2014;24:1162-9. [Crossref] [PubMed]
  39. Liu S, Zhang X, Liu Y, et al. Early application of caffeine improves white matter development in very preterm infants. Respir Physiol Neurobiol 2020;281:103495. [Crossref] [PubMed]
  40. Kilicdag H, Daglioglu YK, Erdogan S, et al. Effects of caffeine on neuronal apoptosis in neonatal hypoxic-ischemic brain injury. J Matern Fetal Neonatal Med 2014;27:1470-5. [Crossref] [PubMed]
  41. Dobson NR, Patel RM, Smith PB, et al. Trends in caffeine use and association between clinical outcomes and timing of therapy in very low birth weight infants. J Pediatr 2014;164:992-998.e3. [Crossref] [PubMed]
  42. Patel RM, Leong T, Carlton DP, et al. Early caffeine therapy and clinical outcomes in extremely preterm infants. J Perinatol 2013;33:134-40. [Crossref] [PubMed]
  43. Kumral A, Baskin H, Yesilirmak DC, et al. Erythropoietin attenuates lipopolysaccharide-induced white matter injury in the neonatal rat brain. Neonatology 2007;92:269-78. [Crossref] [PubMed]
  44. Rees S, Hale N, De Matteo R, et al. Erythropoietin is neuroprotective in a preterm ovine model of endotoxin-induced brain injury. J Neuropathol Exp Neurol 2010;69:306-19. [Crossref] [PubMed]
  45. Fischer HS, Reibel NJ, Bührer C, et al. Prophylactic Erythropoietin for Neuroprotection in Very Preterm Infants: A Meta-Analysis Update. Front Pediatr 2021;9:657228. [Crossref] [PubMed]
  46. Connolly S, Kingsbury TJ. Caffeine modulates CREB-dependent gene expression in developing cortical neurons. Biochem Biophys Res Commun 2010;397:152-6. [Crossref] [PubMed]
  47. Di Martino E, Bocchetta E, Tsuji S, et al. Defining a Time Window for Neuroprotection and Glia Modulation by Caffeine After Neonatal Hypoxia-Ischaemia. Mol Neurobiol 2020;57:2194-205. [Crossref] [PubMed]
  48. Kumar VHS, Lipshultz SE. Caffeine and Clinical Outcomes in Premature Neonates. Children (Basel) 2019;6:118. [Crossref] [PubMed]
  49. Yang L, Yu X, Zhang Y, et al. Encephalopathy in Preterm Infants: Advances in Neuroprotection With Caffeine. Front Pediatr 2021;9:724161. [Crossref] [PubMed]
  50. Yuan Y, Yang Y, Lei X, et al. Caffeine and bronchopulmonary dysplasia: Clinical benefits and the mechanisms involved. Pediatr Pulmonol 2022;57:1392-400. [Crossref] [PubMed]
  51. Yang L, Yu X, Zhang Y, et al. Caffeine treatment started before injury reduces hypoxic-ischemic white-matter damage in neonatal rats by regulating phenotypic microglia polarization. Pediatr Res 2022;92:1543-54. [Crossref] [PubMed]
  52. Zhang Y, Wang Y, Dou H, et al. Caffeine improves mitochondrial dysfunction in the white matter of neonatal rats with hypoxia-ischemia through deacetylation: a proteomic analysis of lysine acetylation. Front Mol Neurosci 2024;17:1394886. [Crossref] [PubMed]
  53. Zupancic JAF. Cost-effectiveness and pricing of caffeine. Semin Fetal Neonatal Med 2020;25:101179. [Crossref] [PubMed]
  54. Saroha V, Patel RM. Caffeine for preterm infants: Fixed standard dose, adjustments for age or high dose? Semin Fetal Neonatal Med 2020;25:101178. [Crossref] [PubMed]
  55. Türkyilmaz C, Türkyilmaz Z, Atalay Y, et al. Magnesium pre-treatment reduces neuronal apoptosis in newborn rats in hypoxia-ischemia. Brain Res 2002;955:133-7. [Crossref] [PubMed]
  56. Imamoglu EY, Gursoy T, Karatekin G, et al. Effects of antenatal magnesium sulfate treatment on cerebral blood flow velocities in preterm neonates. J Perinatol 2014;34:192-6. [Crossref] [PubMed]
  57. Richter AE, Scherjon SA, Dikkers R, et al. Antenatal Magnesium Sulfate and Preeclampsia Differentially Affect Neonatal Cerebral Oxygenation. Neonatology 2020;117:331-40. [Crossref] [PubMed]
  58. Khatib N, Ginsberg Y, Ben David C, et al. Magnesium sulphate neuroprotection mechanism is placental mediated by inhibition of inflammation, apoptosis and oxidative stress. Placenta 2022;127:29-36. [Crossref] [PubMed]
  59. Bickford CD, Magee LA, Mitton C, et al. Magnesium sulphate for fetal neuroprotection: a cost-effectiveness analysis. BMC Health Serv Res 2013;13:527. [Crossref] [PubMed]
  60. McLeod RM, Rosenkrantz TS, Fitch RH. Antenatal Magnesium Sulfate Benefits Female Preterm Infants but Results in Poor Male Outcomes. Pharmaceuticals (Basel) 2024;17:218. [Crossref] [PubMed]
Cite this article as: Wang Y, Wang S, Dou H, Zhang Y, Yang L. Neuroprotective effects of caffeine, erythropoietin, magnesium sulfate, and thyroxine in preterm infants: a network meta-analysis. Transl Pediatr 2025;14(10):2640-2656. doi: 10.21037/tp-2025-260

Article Options

Download Citation

Share