New Criteria for Extrauterine Growth Restriction in Very Low-birth-weight Preterm Children and Neuropsychological Development at the Beginning of Primary School.

Controversy between short-term neonatal growth of very low birth-weight preterm (VLBW) and neurodevelopment may be affected by new criteria of intra- and extrauterine growth restriction (IUGR and EUGR). Objective: To determine if IUGR and IUGR classic and new criteria are related to the neuropsychological development in VLBW. Patients and methods: 87 VLBW were studied at 5–7 years. Neuropsychological assessment included RIST test and NEPSY-II tests. Results of these tests were related to IUGR and EUGR classication using Fenton and INTERGROWTH-21 (IGW-21) graphs and standards. Results: Weight IUGR by Fenton and IGW-21 was 37.9 and 39.1% (Kappa 0.879). Classic (static) EUGR was 89.7% and 75.9%, respectively (Kappa 0.532). “True” EUGR was 52.9% and 36.8% (Kappa 0.683). Result of the RIST index was correlated with Fenton´s z-score weight (0.034), length (< 0.001) and head circumference (HC) (< 0.001) at birth; with IGW-21´s length (0.002) at birth; with Fenton´s weight (0.004) and length (0.003) at neonatal discharge; and with IGW-21´s weight (0.004) and length (0.003) at neonatal discharge. We found statistically signicant differences in IQ when comparing cases with and without IUGR of weight, length and HC by Fenton; and length and HC by IGW-21; also we found differences in static EURG for length and HC for Fenton and IGW-21. In NEPSY-II subtasks we found some relationship in inhibitory control and visuospatial abilities. Conclusion: IUGR and EUGR classic and new concepts show different classication percentages of VLBW. This should be considered when assessing relationships with risk of alterations in neuropsychological tests at school initiation, due to the different results found depending on the classication criteria chosen.


Background
Brain development of newborn with very low birth weight (VLBW) depends largely on correct nutritional intake during immediate postnatal period. Nutrition and, therefore, body and brain growth in rst weeks of life may be of great importance for neuropsychological development of these children (1).
Postnatal growth rate of VLBW and its relationship with neuropsychological development has been studied by many authors with different conclusions (2)(3)(4)(5)(6)(7)(8)(9). Results variability are given by different growth parameters analysed, different neurodevelopmental tests used, and different ages studied. Of course, neurodevelopment is in uenced by many other factors, pre-and postnatal, organic and social, that can confuse or modify this relationship (10)(11)(12).
When we analyze growth of premature infants in rst weeks of life, two important questions arise: what ideal growth of these infants is? and what measure is the most opportune to verify it? (13)(14)(15).
American Academy of Pediatrics recommended since 1977 use fetal growth as ideal of postnatal growth in preterm infants (16). For this reason, most hospitals used reference curves based on cross-sectional measurements of weights of neonates at birth with different gestational ages (GA). In last 20 years, Olsen and Fenton´s curves (17)(18)(19), with their progressive modi cations, were basics in calculations and classi cations of preterm infants, especially in developed countries.
With these curves neonates were classi ed in percentiles or z-scores, in order to label neonates as small (IUGR), normal or large for their GA, in weight, length and cranial perimeter. But, in addition, they were used to label their growth from birth, during their neonatal stay, until 28 days, 36 weeks of GA or discharge, in order to assess so-called extrauterine growth restriction (EUGR).
However, in recent years, several changes in these concepts have been postulated. It is advocated not to use curves created with data from cross-sections at birth, but rather ones that de ne how a healthy premature baby should grow, without prenatal or postnatal complications throughout their rst months of life (15,20). This reasoning has changed 1997 AAP´s paradigm to a new approach based on standards such as those of INTERGROWTH-21st Project (IGW-21), which publishes data on the growth of preterm infants from birth to week 64 of postmenstrual age (21,22), then coupling to WHO curves (WHO Child Growth Standards (23)).
Articles have been published comparing both strategies, recalculating percentages of IUGR preterm infants (weight below 10th percentile at birth) and static EUGR (weight below 10th percentile at discharge, or at 36 weeks of corrected GA), differing according to graphs / standards used (24).
Another change that arises on this publications is whether we should call EUGR to all neonates who are below 10 th percentile at discharge or at 36 weeks of corrected GA, or only those who born above this percentile (non-IUGR) and are below it at discharge, what some authors call true EUGR (25). To end this discussion, other authors prefer to use dynamic concept of loss of 1 or 2 standard deviation (SD) in the zscore from birth to discharge, to speak of EUGR, what we could call dynamic EUGR.
In any case, debate remains open and discussion on the classi cation of premature infants, with static or dynamic criteria, creates new fronts in the effect of growth on the neuropsychological development of these children (2). For all the above, we believe it is of interest to analyse neuropsychological functions at 5-7 years, (beginning of primary education) of a cohort of children under 1500 g at birth, relating their results with the somatometric and neonatal growth criteria used and modi ed in last years in literature.

Patients And Methods
An observational study was conducted, including 87 VLBW (less than 1500 g) who were born in our Neonatology Service between 1/1/2009 and 1/1/2012.
In this period, 181 initial cases were born in our hospital, of which 34 died before discharge and 147 survivors were invited to participate in this study, responding a rmatively 87 cases (51 men and 36 women). All of them went to hospital for a clinical assessment and to carry out a neuropsychological study between February 2016 and May 2017. Comparing studied cases (87) and those that were not studied (60 lost cases), we did not nd statistically signi cant differences neither in neonatal variables, nor in neurological evolution at medium-long term (Table 1).
In 87 studied cases, protocol was completed with somatometry at birth, somatometry at discharge from neonatal admission, and at 2 years of chronological age corrected for GA. Birth and discharge data were analysed with Fenton and IGW-21 graphs and standards, and 2 years data with WHO graphs.
We de ne IUGR as cases born below 10th percentile for GA. We de ned static (or cross-sectional) EUGR as cases that were below 10th percentile for GA at neonatal discharge. We de ne as true EUGR the cases that, born above 10th percentile (not IUGR), were discharged below 10th percentile for their GA. And we de ne dynamic (or longitudinal) EUGR cases that lost more than 1 or more than 2 SD from birth to discharge, regardless of their initial or nal percentiles.
During the study visit, a neuropsychological evaluation was performed by psychologists, including an estimation of IQ, inhibitory control, verbal and visuospatial memory and visuospatial abilities, using RIST test and NEPSY-II battery.
RIST test (Reynols Intellectual Screening Test) is a screening test that provides a measure of an estimated IQ through two tasks, Riddles, a verbal task, and Categories, a non-verbal task. Its reliability coe cient varies between 0.89 and 0.91 depending on the age of child (5-7 years) and if it is the verbal or non-verbal subtest (26). For this study, the IQ values (mean 100, SD 15) are reported.
From NEPSY-II battery (27), the following subtasks were employed: Inhibition, as measure of inhibitory control; Memory for names subtask, which measures child's verbal memory capacity; Memory for designs subtask, that assess the ability to retain visual and spatial information; and Route nding, which measures visuospatial abilities, speci cally, knowledge of visuospatial relationships and directionality.
For this study, scalar scores from all these test are reported.
Usual tests were carried out in the statistical analysis, initially determining the normality of the variables with the Kolmogorov test. Variables with normal distribution were compared with Student's t test. The variables whose distribution was not normal, and those variables score type, were compared using nonparametric statistics test (Mann Whitney U test). Comparisons of percentages were made with the Chisquare test (using Fisher's exact test if expected were less than 5). To establish correlations between zscores, Spearman Rho test was used. Finally, Kappa concordance test was used to compare Fenton and IWG-21 classi cations.
Study was approved by the Research Ethics Committee of the Principality of Asturias.

Results
1.-Description of the growth of 87 cases studied Somatometric data at birth, at discharge from neonatal admission and at 2 years of age can be seen in Table 2, while in Table 3 we can read the classi cations at birth, at neonatal discharge and at 2 years according Fenton, IGW-21 and WHO graphs and standars.
Percentage of low weight for their GA (less than 10th percentile or IUGR) for Fenton and IGW-21 were 37.9 and 39.1%, respectively (Kappa 0.879). Percentage of cases weighing less than 10th percentile at neonatal discharge (static or cross-sectional EUGR) were 89.7% for Fenton and 75.9% for IGW-21 (Kappa 0.532). Percentage of cases that being born with weight above 10th percentile for Fenton and IGW-21, at discharge were below this percentile (true EUGR) were 85 and 60.4%, corresponding to 52.8% (46 cases) and 36.8% (32 cases) of total, respectively for both Fenton and IGW-21 (Kappa 0.683). All true IGW-21 weight EUGRs were recognized by Fenton as true EUGRs, but 30% of Fenton's true weight EUGRs (14 cases) escape IGW-21 as true EUGRs.

2.-Description of results of neuropsychological tests
In Tables 4 and 5 are shown the global neuropsychological results carried out in 87 studied cases.
The RIST test, performed in 86 cases, was scored as high in 9 (10.3%), as medium in 44 (50.6%), as a deviation below mean in 23 (26.4%) and as two deviations below mean in 10 cases (11.5%). In the other tests, a level below or well below that expected was obtained in 11 cases (12.6%) for the Inhibition task, in 8 cases (9.2%) for Memory for names, in 27 cases (31%) for Route nding test and in no case for Memory for designs.
Comparing scalar scores or IQ scores of different tests between cases with and without clinical neurological alterations diagnosed during evolution, we found statistically signi cant differences in Inhibition and in Route nding (Table 6).

3.-Relationship between growth and neurological clinical evolution and neuropsychological tests
There are not signi cant associations between growth disturbance and to have a clinical neurological alteration, except for the loss of 2 SD in length between birth and neonatal discharge for Fenton (Table 7).
We found statistically signi cant correlations between RIST index and z-score for Fenton at birth (weight, length and HC), for IGW-21 at birth (length); for Fenton at discharge (weight and length); for IGW-21 at discharge (weight and length). We did not nd any statistically signi cant correlation between RIST and z-scores at 2 years, nor between any of other tests analysed with any of measures studied (Table 8).
Comparing scalar or IQ values of different tests between established cut-off points we found some statistically signi cant differences that can be read in detail in Table IX. IQ scores from RIST index showed statistically signi cant differences for the IUGR of weight, length and HC; for static EUGR of length and HC; and for HC less than 2SD at 2 years.
We did not nd any statistically signi cant difference for Memory for names nor for Memory for designs. For Inhibition task, we found statistically signi cant differences for IUGR for Fenton and IGW-21, for the classi cation of true EUGR of weight for IGW-21 (-1SD between birth and discharge) and for height and HC at 2 years. In Route nding task we found differences for static HC EUGR <-2DS for Fenton and IGW-21, and for dynamic HC IGW-21 EUGR at -2DS. All differences found were in favour of normal data (not IUGR or not EUGR), except for inhibition task and true EUGR, in which we found better performance in those cases with true EUGR. As seen in Table 9, clinical importance of statistically signi cant differences found is highly variable. Discussion VLBW may be due to prematurity and / or IUGR of different etiologies. Later, growth of these neonates can be in uenced by multiple genetic, epigenetic, neonatal (respiratory, cardiac, infectious, nutritional, etc.) and post-neonatal (nutrition, morbidity, culture, etc.) factors. On the other hand, neurodevelopment of these children is also in uenced by different factors, coinciding some of them with the above mentioned and being some of them related to other matters (parental intelligence, socio-family level, education, etc.).
In published observational studies, it has been analyzed whether growth is related to neurodevelopment in short, medium and long term, although the conclusions are not uniform and debate remains open (2-9).
Beyond neurological alterations diagnosed clinically in follow-up (motor, sensory, cognitive, behavioral, ...), VLBW may present neuropsychological alterations at school age, only detectable by speci c tests for this purpose, such RIST test and NEPSY-II battery at 5-7 years (11,28). For this reason, our study aimed to know relationship between result of these tests and initial growth of VLBW, comparing different de nitions of neonatal growth, without entering into the in uence of other neonatal factors that are already present in previous publications (11). Growth.
One of the major problems that arises when assessing relationship between neonatal growth and subsequent neurodevelopment is which growth parameters and somatometry to use. Classically, Fentoń s 10th percentile was cut-off at birth and neonatal discharge for classifying children as IUGR and EUGR, but this classi cation is now obsolete. First, IGW-21 standards seem much more interesting as basic data, given the way they are obtained (longitudinal with healthy preterm infants), to classify children (21), although it is still not clear whether we should call all those under 10th percentile at discharge or if we should only use this term for those who, born above 10th percentile, go from discharge below it. Other authors go further, proposing that EUGR concept should be based on the reduction of standard deviations, rather than static percentiles, adding a dynamic and critical look to this discussion. This is how the concept of static (or transverse) EUGR and dynamic (or longitudinal) EUGR arose, being this latter the currently preferred (2,4,16,29,30).
Our small series aimed to evaluate these aspects, analysing static somatometric data (weight, length and HC at birth and discharge), but also some dynamic ones (true EUGR and changes less or greater than -1 and -2 SD between birth and discharge), both for Fenton and IGW-21. Our results support that the percentage of VLBW classi ed as IUGR and EUGR vary greatly depending on criteria used.
Thus, IGW-21 is slightly less selective at birth in IUGR classi cation (most cases than Fenton) but, at the same time, stricter at discharge in the static EUGR classi cation for weight (less cases less than Fenton), which makes some true Fenton-tagged EUGRs slip away (30.4% in our series), while Fenton doesn't miss any of the IGW-21-tagged ones. The same does not happen with length and with CP, which do not present a uniform trend, contrary to what appears in other series (31).

Neuropsychological test
In our 87 VLBW, we found no differences in somatometry at birth and in their growth up to 2 years of age between children with and without neurological alteration clinically diagnosed at 5-7 years, but in neuropsychological tests performed at this age. With RIST index and NEPSY-II battery we have veri ed that, in addition to static somatometry at birth, certain parameters of neonatal growth and during the rst 2 years of life in uence the performance of these children at study age.
RIST is a screening test that provides us with a measure of estimated IQ, which, in our series, was correlated with z-score at birth (Fenton´s weight, length and HC, and IGW-21´s length) and at neonatal discharge (Fenton and IGW-21 weight and length), but not with z-score differences between birth and neonatal discharge, nor with the z-score at 2 years. In addition, IUGRs in weight, length and HC, and static EUGRs in length and HC show lower scores, statistically signi cant, compared to neonates who were not, with differences of 10-20 points between them. Same occurs with neonates who present HC score less than 2SD for WHO at 2 years. These results coincide partially with the previous literature, nding that some neonatal and early postnatal measures which indicated a lower growth status could be related with later and lower IQ achievements (2).
However, it is also worth mentioning major methodological differences of our study with some previous publications, such as the use of different de nitions of EUGR, with Fenton and IGW-21 graphs and standars. Therefore, it seems that the use of these new approaches may be of interest, as a possible predictor of subsequent neuropsychological development.
Another issue to consider would be the different importance given to the IQ in these previous publications. As it has been said before, while previous articles are focused on IQ assessment as the only measurement of cognitive development, our aim was to consider a wider neuropsychological pro le, and due to that, RIST test was included only as a screening measure of IQ. With these results, we must admit that RIST index is mathematically related with static somatometric values at birth and discharge, but not with dynamic ones. In any case, this relationship, as we discussed previously, can be in uenced by many other prior or intermediate factors that surely play important roles.
In NEPSY-II tests, we found some interesting relationships, although we did not nd a clear pattern between growth and the different subtasks. It is striking, for example, that Memory for names and Memory for design tasks, related to verbal and visuospatial memory respectively, did not present any type of relationship (neither correlation nor statistically signi cant differences) with any growth parameter. This may be due to the fact that the memory scores reported in the present study are composite indexes of both short-and long-term memory, which are related but independent cognitive functions and their addition could decrease their ability to differentiate between growth statuses. Besides, it should be noticed that in Memory for designs, any of our participants scored below expected, and then, this can limit the discriminatory capacity of such task.
We also did not nd correlations between scalar scores of other studied NEPSY-II subtasks (Inhibition and Route nding subtasks), although we were able to determine some statistically signi cant differences for some growth items, but without a clear pattern of behavior or clinical differences, important enough to be highlighted. These ndings could point to the possibility that slower growth does not affect speci c neuropsychological functions, but rather a more global pattern of cognitive development.
In summary, classi cation IUGR and EUGR (static, dynamic and true) with Fenton and IGW-21 show very varied percentages according to the criteria used in the classi cation. Besides, growth in uence on neuropsychological tests does not seem robust, except for IQ measurement, which is clearly correlated with some static measures and which shows clinically important differences between some groups classi ed according to these measurements.
New IUGR and EUGR concepts should be evaluated in future studies in this eld since, probably, they catalog neuropsychological risk groups differently and modify ideas that were maintained up to that moment. Reviewing growth importance in rst weeks of life in later neurodevelopment is a topic of great interest to neonatologists, endocrinologists, neurologists, and psychologists.  Tables  Table 1 Comparison between studied cases and survivors not studied (lost) for the perinatal and neonatal variables.          Table 9 Differences found in the scalar score (statistically signi cant) of analyzed tests in the comparison of different somatometric groups, for all cases (n = 87).