Selection of Neonates for Neuroprotective Therapies: Title and subTitle BreakOne Set of Criteria Applied to a Population FREE

SEVERAL AUTHORS have proposed criteria for early recognition of serious birth asphyxia.^{1 - 4} Infants with such characteristics might be reasonable candidates for clinical trials of proposed neuroprotective therapies.^{5 - 8} Perlman and Risser² have suggested 3 early characteristics as criteria for serious neonatal asphyxia, as tested by their association with neonatal seizures: a 5-minute Apgar score of less than 6, blood pH of 7.00 or less, and need for intubation for ventilation in the delivery room.

The adverse neurodevelopmental outcome that has been consistently linked with birth asphyxia is spastic cerebral palsy (CP), and especially spastic quadriplegia,^{9 - 10} sometimes accompanied by dyskinesia. Mental retardation, epilepsy, and pure hypotonia or pure ataxia in children without spasticity have not been linked with birth asphyxia. Pure dyskinesia, whose relationship with birth events has not been demonstrated in controlled studies, is too uncommon to be a target for tests of therapies in the newborn period. Spastic CP is therefore the plausible outcome for trials of neuroprotective treatment in surviving infants.

Data on the characteristics proposed by Perlman and Risser² for recognition of serious birth asphyxia are available in a population-based case-control study of CP in 4 northern California counties in a 3-year period.¹¹ We examined these criteria among full-term children surviving to age 3 years with disabling CP in an effort to estimate the proportion of CP that might be prevented by using these criteria as a basis for selecting children for neuroprotective treatment, assuming treatment to be perfectly successful. Such numbers might also give some indication of the feasibility of randomized trials for birth asphyxia in the newborn period.

Low Apgar scores, low blood pH, and need for ventilatory support are not specific to acute asphyxial states.^{12 - 16} Therefore, we also examined other perinatal and postneonatal characteristics of children with disabling spastic CP. The purpose of this effort was to identify other conditions that might have influenced outcome in children who met the stated criteria.

The outcome of this study was spastic CP, which is more likely than purely ataxic or dyskinetic forms to be related to birth events. Cerebral palsy was defined as a chronic disability originating in the central nervous system, characterized by aberrant control of movement or posture, appearing early in life and not the result of progressive disease. Children with mild motor dysfunction and those in whom disability was acquired after the first 28 days of life or through nonaccidental head trauma in the first month were excluded. Children with hypotonia, ataxia, athetosis, or other nonspastic motor dysfunctions were included only if they also had limb spasticity.

Case patients were singleton children born in 1983 through 1985 to residents of 4 San Francisco Bay area counties and included all children who met the following criteria: gestational age 37 weeks or more or, if gestational age was unknown, weighing 2500 g or more at birth; survival to age 3 years; residents of California for the entire period; and moderate or severe congenital CP. For initial ascertainment of cases we used records of 2 state agencies known to enroll virtually all eligible children; we determined final case status based on standardized clinical examination or extensive record review. Detailed information on ascertainment procedures is available elsewhere.¹¹ Case children were a median age of 4.9 years at examination. Controls were randomly selected from singleton children who met all the case criteria except for CP.

Demographic and clinical data were obtained from birth certificates and medical records at more than 40 hospitals in the 4 counties. Maternal labor and delivery and neonatal records were abstracted by nurses who did not know the purpose of the study or whether the records were those of case or control children. Nurse abstractors noted diagnoses and descriptions recorded by medical caregivers and did not infer diagnoses from clinical signs and symptoms. Gestational age was derived from measures recorded in mothers' medical records before delivery, with precedence given to dates established early in pregnancy and to estimates based on ultrasound examinations before 19 weeks of gestation. Level of care of the hospital of birth was determined according to criteria of the California Children's Services program. Level I hospitals were those without specialized services for sick or premature infants; level II hospitals were those providing care for sick neonates who did not require intensive care; and level III hospitals provided a full range of services, including neonatal intensive care. Distinctions between levels II and III were often unclear, and these were grouped for analysis.

Diagnoses of birth complications, including maternal infections during the admission for delivery, were as recorded in maternal records. Information on neonatal characteristics came from the hospital of birth and all other inpatient admissions before infants' first discharge to their homes. For children with CP, we examined postneonatal medical and service agency records for documentation of conditions outside the intrapartum period that might have been relevant to outcome.

As part of the effort to identify other causes of illness in symptomatic neonates, we reviewed ultramicroimmunoassay findings in a small subgroup of children with spastic CP whose archived neonatal blood was supplied by the Genetic Disease Branch of the California State Department of Health and tested for coagulation factors, as recently described.¹⁶ Consent materials were mailed to the parents or guardians of a selected group of 59 children, most of them full-term infants without known malformations or syndromes, who were considered clinically interesting by a child neurologist or medical geneticist. Consent was obtained for 42 children, refusal for 4, and no response for the remainder. Of those consenting, blood specimens were missing or insufficient for 11. The final sample consisted of 31 children with CP for whom blood samples were available. Control children were selected from the larger series of 482 controls who were born weighing 1500 g or more in the same study years and counties as the cases and who survived to age 3 years. Individual identifiers were not available for control children but their racial and ethnic distribution approximated the larger series of controls.

Whole-blood samples, dried on filter paper and provided without information as to case or control status, were processed for analysis. A 5-mm-diameter circle (equivalent to 15 mL of whole blood) was removed from each sample. Reactive antibodies to lupus anticoagulant, anticardiolipin, antithrombin III, and the translation product of the factor V Leiden mutation were initially isolated as total IgG and IgM fractions by recycling immunoaffinity chromatography, then measured by capillary electrophoresis with chemiluminescence-enhanced immunoassay.¹⁷ Proteins C and S antigens were isolated and measured by recycling immunoaffinity chromatography with laser-enhanced fluorescence detection. The results were compared with known standards processed under identical conditions, and the final concentrations were compared with normal standards for adult humans¹⁸ and neonatal controls. For identification of a group abnormal for coagulation factors, we used recursive partitioning to identify the value that best discriminated between children with CP and control children.¹⁹

In a population of 155,636 survivors to age 3 years, there were 172 singleton children with moderate or severe congenital CP and 494 singleton control children. Of these, there were 87 singleton children with spastic CP whose gestational age was 37 weeks or greater (or, if gestational age was unknown, weighed 2500 g or more at birth) and 375 controls whose gestational age was 37 weeks or greater (or, if gestational age was unknown, weighed 2500 g or more at birth) for whom information on most delivery room variables was available. In addition, there were 11 singleton children with pure ataxia and 2 with pure dyskinesia who were not included. From these 100 cases and 375 controls, there were 3 children with CP and 9 controls whose delivery records were incomplete, resulting in 84 cases and 366 control children for whom information was sufficient for analysis.

Blood pH was not routinely measured in these full-term infants but was examined for most infants with signs of neonatal depression (see below). Thirty-four children with CP and 19 controls had pH determinations.

Among 84 full-term children with moderate or severe spastic CP, there were 18 with 5-minute Apgar scores below 6 (Table 1). Twenty children required intubation for ventilation in the delivery room, and 5 had initial blood pH of 7.00 or less. Three (3.6%) of the 84 children with CP had all of these signs, and all 3 experienced neonatal seizures (Table 2)—2 within 3 hours of birth and 1 at 55 hours after birth.

Two of 366 full-term control children had low Apgar scores and 2 others needed ventilation; only 1 of these 4 had pH testing. Of the 19 control children whose pH was measured, none had a value of 7.00 or less and no control experienced a neonatal seizure.

We expanded the cited criteria to evaluate the number of additional children with CP and controls who would thereby be included. Requiring low Apgar scores and need for resuscitation but broadening the pH criterion to 7.10 or less would lead to the inclusion of 5 additional children with CP, of whom 3 had neonatal seizures (Table 2), 2 of these within 2 hours of birth and 1 at 6.6 hours after birth.

If the criteria were expanded to require any 2 of the 3 original elements but used the broader pH cutoff of 7.10, then 9 additional children with CP would be included, 6 of whom had neonatal seizures. Two additional children had low Apgar scores and needed resuscitation but did not have pH determinations; both had neonatal seizures.

Inclusion of these 11 children who met expanded criteria would mean that 19 (23%) of 84 children with CP would be identified, 11 of whom had neonatal seizures. Eight of the 19 were born in level I facilities and 14 of the 19 had the spastic quadriplegic subtype of CP.

No control child met the original or the expanded criteria. For the 2 excluded children with pure dyskinesia, Apgar scores were good (8/9 and 9/9 at 1 and 5 minutes, respectively), neither required ventilation, and blood pH testing performed in 1 of these children yielded a value of 7.39.

We examined other clinical characteristics of children with either the strict or expanded criteria to identify other conditions that might have influenced outcome. The available clinical information varied widely, both for perinatal and postneonatal information, and few postneonatal records contained complete and recent workups for all potentially relevant conditions.

Of the the 3 children with CP who met the original criteria, 2 had tight nuchal cords reported in the maternal record. For 1 of these children, chorioamnionitis and severe funisitis were also reported (Table 2); the other was described as dysmorphic by a geneticist on the basis of a deformed nose with thick alae nasae and frontal hair whorl. The third child, whose blood pH was extraordinarily low, had hydronephrosis recognized on prenatal ultrasonography, was born to a preeclamptic mother by difficult vacuum extraction, and had a maternal aunt with CP and a father with first myocardial infarction at age 44 years. This child's archived neonatal blood, tested by immunoassay in a study finding that high values of proteins C and S were associated with CP (fully described elsewhere¹⁶ ), revealed the highest concentrations of protein C and protein S of any of the 96 case and control children tested.

Many of the other children with CP who met the original or the expanded criteria had complicated perinatal histories. Four infants had evidence of in utero exposure to maternal infection. In another, severe neonatal infection was identified soon after delivery; this was a neonate with tight nuchal cord, pneumonia, initial white blood cell count of 33.9 × 10³/µL, high level of antithrombin III, and dental enamel hypoplasia suggestive of perturbation at a gestational age of 4 to 5 months.²⁰ Another infant had cystic fibrosis and at birth had massive ascites related to intrauterine ileal perforation and meconium peritonitis. Thrombocytopenia was reported in 3 children; an unspecified coagulation disorder in another; and growth retardation, preeclampsia, and cocaine use were features in the histories of others. Coagulation factors, tested in 5 infants, revealed an abnormality associated with CP¹⁶ in all 5; whether these were constitutive or reactive is not clear.

Many authors propose neuroprotective therapies for asphyxiated newborn infants,^{5 - 8} but only a few discuss criteria that might govern the selection of infants for clinical trials of administration of those therapies. One article² that did propose criteria chose early characteristics about which data were available in this population-based study of CP, permitting examination of the frequency of these findings in children with CP and controls, and assessment of other characteristics of infants who met these criteria soon after birth.

Our purpose in this study was to estimate the proportion of full-term children with spastic CP who had low Apgar scores, low blood pH, and respiratory depression. The California Cerebral Palsy Project is one of the few US population-based sources for such data. As part of the effort to assess the potential effectiveness of neuroprotective therapies in the newborn, we also examined the nature of the illnesses in infants who met these criteria, some variant of which are commonly used to identify "birth asphyxia." There is reason (see below) for thinking that these criteria are not specific for acute asphyxia. It is an important strength of this study, therefore, that it incorporated a wide range of perinatal and postneonatal information on neurologically depressed full-term neonates with a later diagnosis of CP.

An important limitation of this study is its restriction to survivors to age 3 years, meaning that this study is not informative about the association of the examined criteria with mortality, nor is it an appropriate basis for predicting the joint outcome (death or CP). In addition, blood pH determinations were not standard practice; pH values were obtained on the basis of clinical indication, so that performance of testing was itself a marker of clinician concern. Only 2 children with CP who had low Apgar scores and needed resuscitation did not have pH testing, however, and only 1 control with a low Apgar score (who had not required resuscitation) did not have a measured pH. It appears unlikely, then, that many children, cases or controls, were excluded from consideration because pH was not tested.

There were too few controls to permit examination of indicators that were rare in controls, such as low blood pH or neonatal seizures, so a false-positive rate cannot be offered. Nonetheless, given the 2 children with low Apgar scores and 2 others who needed early ventilation among the 366 full-term controls, a false-positive rate related to these 2 markers would be nontrivial.

This birth cohort of 155,636 neonates produced only 3 surviving full-term singleton infants with spastic CP who met the studied criteria, a prevalence of 1 in 51,879 births or 0.019 per 1000, one tenth the 0.2 per 1000 births estimated in a recent Canadian study.³ Inclusion of the 2 infants in this study who met 2 of the examined criteria but lacked a blood pH determination would bring the prevalence to 1 in 31,127 births, still well below the rate suggested by the Canadian study. Prevalence is obviously important in questions regarding the feasibility of randomized trials of neuroprotective agents in full-term infants.

The tested criteria did identify a sick group of neonates, and none of the 366 control children had more than 1 of these criteria. However, neuroprotective therapy, even if perfectly effective, offered only to children with the neonatal characteristics examined here would be unlikely to prevent most CP. Noting this, it is tempting to consider further broadening the criteria. But prior studies show that most children who survive even severe depression have a good chance for later neurological normality.^{21 - 22} Neither Perlman and Risser² nor we^{10 ,23} found that presence of meconium in the amniotic fluid or electronic fetal monitoring data gave additional information that was helpful in recognizing level of risk. The experimental therapies are unlikely to be free of risk, and risk cannot be judged entirely on the basis of studies in animals. The probable false-positive rate using Apgar score and ventilatory requirement would have to be weighed against potential benefit in considering exposure to risk of infants who might do well without experimental intervention.

Signs commonly attributed to birth asphyxia, including low Apgar scores, need for resuscitation, and neonatal seizures, may occur in neonatal illnesses not due to uncomplicated interruption of oxygen supply. We¹⁶ and others^{12 - 14} have shown that intrauterine exposure to infection is associated with low Apgar scores and need for ventilatory support in the delivery room. In addition, low Apgar scores and low pH tended to be more frequent in children with CP whose blood, tested in the neonatal period, contained markers of autoimmune disorders or abnormalities of coagulation factors.¹⁶ Neonatal seizures can be early manifestations of fetal or perinatal strokes. Such strokes may be associated with anticardiolipin antibodies in blood of mothers^{24 - 27} or infants,²⁸ or with the factor V Leiden mutation.^{29 - 31} Low neonatal blood pH in children with CP might be due to lactic acidosis related to infection, or to placental vasculopathy or thrombosis resulting from coagulopathy or autoimmune disorder.^{32 - 33} In the presence of placental vasculopathy or thrombosis, embolization, inflammation, or vasculopathy in the fetus might contribute to illness observed in the early hours and days of life. It is apparent, then, that low Apgar scores, need for resuscitation, and low pH, as well as neonatal seizures, are signs of neonatal illness that are not specific to acute asphyxial states. Most of the animal models on which neuroprotective strategies are based involve uncomplicated insults to previously healthy young experimental animals; such models probably do not well reflect the complexities encountered in the human experience.

Indeed, most of the infants who met either the strict or relaxed criteria evaluated here had complicated perinatal histories and a variety of conditions other than primarily asphyxial illnesses that were potentially relevant to outcome. This important observation was made despite the incompleteness of postneonatal clinical information available and the small number who underwent neonatal blood testing, undoubtedly resulting in underascertainment of nonasphyxial illnesses in these children. Intrauterine exposure to infection,¹⁵ neonatal alloimmune thrombocytopenia³⁴ and other coagulation factor abnormalities,¹⁶ for example, are themselves associated with risk of neurological disability. The presence of such conditions may alter expectations of the responses of such children to neuroprotective therapies.

Would neuroprotective therapies directed at hypoxic or ischemia be likely to protect against brain injury caused by other disorders? Despite hypotheses about convergence of causal pathways resulting in cerebral injury—involving excitatory or oxidative mechanisms of injury, for example—optimal therapy might be different depending on the cause of neonatal depression and encephalopathy.

We observed that several infants with marked early depression and acidosis in this population had abnormalities of coagulation factors in neonatal blood. Hypothermia, recently advanced as a promising approach to neuroprotection in the neonate,^{35 - 36} can be complicated by coagulopathy if severe and prolonged.^{37 - 38} Might infants with coagulation defects be vulnerable to degrees and durations of hypothermia that would not injure other infants?

Two striking findings in this study are the small proportion of full-term children with spastic CP who met the criteria evaluated, or an expanded version of these, and the complicated medical backgrounds of these children.

Controlled studies examining the cause of neonatal depression and encephalopathy in full-term and near-term infants are scarce.^{39 - 40} We need more of such studies, and inclusion of more of the diagnostic possibilities now emerging as potentially relevant. Neuroprotective therapies may make significant contributions in the future, but before these are undertaken it would be prudent to seek a better understanding of the causes of, and needs for specific therapy in, encephalopathy in full-term neonates.

Accepted for publication September 9, 1998.

Terry M. Phillips, DSc, PhD, performed the microassays described here. James Dambrosia, PhD, did the statistical analysis determining the optimal levels of coagulation factors for separation of case from control children. Cynthia Curry, MD, called to our attention her work on coagulation factor abnormalities in a group of children with cerebral palsy. We are also indebted to the Genetic Disease Laboratory and Genetic Disease Branch of the California Department of Health Services, who supplied the neonatal blood samples, and to the parents and children who participated in this study.

Reprints: Karin B. Nelson, MD, 7550 Wisconsin Ave, Room 714, Bethesda, MD 20892-9130 (e-mail: knelson@helix.nih.gov).

Editor's Note: The provocative study of a select population of children with cerebral palsy should stimulate some interesting discussions. What will the lawyers say?—Catherine D. DeAngelis, MD