It is difficult to precisely determine the production capacity and circulating levels of melatonin in PTNB due to the need for studies on this group of children [12, 21, 22]. Melatonin levels can vary significantly due to factors that increase oxidative stress in PTEN. These factors include episodes of pain that increase free radicals, the requirement for supplemental oxygen, and an imbalance between melatonin consumption and production [3, 10, 14, 18, 29].

In our study, we found that preterm infants aged 27–34 weeks (GA, 30.7+/−2 weeks) on the third day of life had plasma melatonin levels of 33.8 ± 12.01 pg/ml (95% CI, 30.5–37.2 pg/ml). Our results agree with those observed by Marseglia et al. [12]. Table 1 shows that plasma melatonin levels are practically identical in preterm infants < 30 weeks and those born between 31 and 32 weeks. However, preterm infants < 32 weeks have lower plasma melatonin levels than those born between 32 and 34 weeks. Other authors have reported lower melatonin levels in preterm infants, with a more significant relationship between lower gestational age and melatonin levels in larger infants [21, 30].

The lower association of melatonin with gestational age and higher plasma melatonin levels in our study may be due to two aspects: (1) Our premature infants who did not have perinatal asphyxia and did not require oxygen had lower levels of oxidative stress and consumed less melatonin [1, 2, 10, 12, 14]. This aspect suggests that few preterm infants in our study had deficient melatonin levels. 9.8% had plasma melatonin levels < 10 pg/ml, whereas in the prospective multicenter study by Biran et al., 81% of their preterm infants < 34 weeks had plasma melatonin < 7 pg/ml on the third day of life. (2) In our study, there were very few preterm infants < 28 weeks or weighing < 1000 g (n = 7, 11.4%), which were those with the most significant oxidative damage and consequently would have a higher melatonin consumption [1, 10,11,12].

Concerning birth weight, we observed that preterm infants under 1250 g had lower melatonin levels than those over 1250 g (p = 0.05), and preterm infants < 1500 g have slightly lower melatonin levels (p = 0.06) (Table 2) with a decrease of 6.1 pg/ml of their melatonin levels (0.08). At lower weights, there were usually lower melatonin levels (r = 0.22, p = 0.07), as shown in Tables 3 and 4. Muñoz-Hoyos et al. [22] analyzed plasma melatonin levels in newborns with respiratory distress. They described newborns < 1500 g had significantly lower levels, probably because their study included term newborns, who usually have higher melatonin levels than PTNBs [31].

Preterm infants in the intensive care unit (ICU) undergo numerous painful procedures, increasing their oxidative stress levels (6.36). Although pain stimulates melatonin production in a physiological attempt to eliminate the free radicals generated, melatonin is often consumed to neutralize them [1, 32, 33]. Moderate or severe pain (PPIPP > 5) is present in 50.8% of preterm infants without asphyxia and oxygen needs, being especially frequent in preterm infants < 32 weeks (Table 2). We have observed that preterm infants with moderate–severe pain (PIPP > 5) have lower plasma melatonin levels (p = 0.01) (Table 3) and being preterm with PIPP > 5 was associated with lower plasma melatonin levels (p = 0.03) (Table 4).

The low levels of melatonin in PTNB with moderate to severe pain may be due to an imbalance between melatonin production and consumption. This imbalance results in higher consumption that cannot be counteracted by the immature antioxidant system of these children [6, 34]. This consideration may explain why melatonin administration has been shown to decrease pain and oxidative stress markers during painful procedures. Melatonin administration has also been observed to decrease the PIPP score in preterm infants [7, 33].

It has been reported that certain drugs, like caffeine, can elevate melatonin levels in the blood by competing for the same metabolic pathway [35]. Caffeine treatment was received by 43.5% of our preterm infants, and no differences in melatonin levels were observed between preterm infants who received caffeine and those who did not, likely due to the significant individual variability in the bioavailability of this drug and the hepatic metabolism in PTNB [29].

Parenteral nutrition is an exogenous source of free radicals, where oxidative stress can come from in vitro nutrient oxidation of solutions and in vivo reactions when intravenous prooxidant molecules are infused [5, 36]. Our study found that receiving parenteral nutrition did not impact melatonin levels in preterm infants who did not experience hypoxia. This finding may be explained because these infants are less exposed to free radicals, and their natural melatonin production is sufficient to counteract any exogenous free radicals. Additionally, protecting parenteral nutrition from light and adding mono- and polyunsaturated fatty acids (PUFA) minimizes free radical formation [37].

Low melatonin levels are one of the determinant factors in the development of free radical diseases (BPD, ROP, HIV, NEC, and sepsis) in PTNB [16, 30]. In preterm infants without hypoxia, free radical diseases also occur in 23.7% of cases [37]. We observed that most preterm infants can produce melatonin levels above 25 pg/ml at 72 h of life [12]. On the third day of life, there were no differences between the melatonin levels at third day of life of preterm infants who developed free radical disease (sepsis, ROP, BPD, HIV, or NEC) probably because many factors associated with free radical disease occurred after the third-day melatonin was measured (late sepsis in 18% of cases, enterocolitis, etc.). Melatonin was not associated with developing these diseases. These diseases were associated with lower gestational age, lower birth weight, lower Apgar test score at birth, and requiring mechanical ventilation or moderate–severe pain on the third day of life, as shown in Tables 3 and 4 [1, 8, 9, 37].

Deficiency of melatonin at the mitochondrial level may be the primary cause of free radical diseases, rather than plasma melatonin levels, as melatonin synthesis and metabolization occur significantly at the mitochondria [18,19,20]. Mitochondria contain much higher concentrations of melatonin in their granules compared to the melatonin levels found in the bloodstream [38]. It has been demonstrated that free radicals at the mitochondrial level can cause cellular injury implicated in the pathogenesis of free radical diseases due to oxidative self-injury and mitochondrial dysfunction [39, 40]. This reasoning is supported by the fact that premature infants treated with melatonin have a decreased incidence of free radical diseases [10, 12, 30, 41] when plasma melatonin levels 500–1000 times higher than physiological levels are reached. Since melatonin has a great facility to cross cell membranes [17], these doses may increase melatonin at the mitochondrial level, which could explain the clinical improvement and prevention of free radical disease seen with its administration.

Ongoing clinical trials, such as the one registered on Clinical Trial.gov (NCT04235773) [42], will help us better understand the effects of administering melatonin to preterm infants below 30 weeks of age. These trials will evaluate the pharmacokinetic and metabolic properties of melatonin at these ages and its effectiveness in preventing free radical diseases that commonly cause neurological morbidity and high mortality [43,44,45,46].

The study has some strengths: it followed a prospective longitudinal design, and the pain scale assessment was performed exclusively by two trained nurses, which reduced inter-subject variability. However, some limitations also exist. The most relevant is the presence of other factors that may affect melatonin levels. These factors include light stress due to light intensity in the neonatal ICU, acoustic stress, emotional stress from the kangaroo program, antibiotic treatment, free radicals present in parenteral nutrition, or other factors not accounted for in this study.