Chorioamnionitis and Cerebral Palsy in Term and Near-Term Infants FREE

Cerebral palsy (CP), a group of nonprogressive motor impairment syndromes caused by lesions of the brain arising early in development,¹ occurs in 1 to 2.4 per 1000 live births.^2- 5 Although premature infants are at particularly high risk, more than half of all children with CP are born at term.⁵ Population-based studies suggest that despite advancements in obstetrical and neonatal care, the prevalence of CP in term infants has remained constant in recent decades.^2- 4,6- 7

Chorioamnionitis, or inflammation of the placental membranes, may increase the risk of CP by 2- to 12-fold in term infants.^8- 12 However, previous studies are limited by small size,^9- 10 and adjusted relative risk estimates are either not significant¹⁰ or not presented for term infants.^8,12 Whereas birth asphyxia was previously thought to be the most common etiology of CP, clinical signs attributed to birth asphyxia, such as low Apgar scores and neonatal encephalopathy, may in some cases be due to chorioamnionitis.^9,13- 15 Previous studies lacking neuroimaging data^8- 10 have been unable to elucidate the complex relationships between chorioamnionitis, birth asphyxia, and CP. Therefore, we examined the association between chorioamnionitis and CP, as well as the role that "birth asphyxia" plays in this relationship, in a large cohort of term infants in the Northern California Kaiser Permanente Medical Care Program (KPMCP), most of whom had undergone neuroimaging studies.

This case-control study is nested within the cohort of all singleton term and near-term (36 or more weeks' gestation) infants born in 1991 to 1998 at the KPMCP. All study procedures were approved by the institutional review boards at the KPMCP and at the University of California, San Francisco.

The KPMCP is a large managed care organization that provides care for more than 3 million residents (30% of the population) in northern California. The members of the KPMCP are demographically similar to the general California population, except that the very poor and very wealthy are underrepresented.¹⁶ Of its 33 facilities, 12 have delivery rooms and 6 have level III neonatal intensive care units.

Among all singleton live births of 36 or more weeks' gestation born between January 1, 1991, and December 31, 1998, we electronically searched the KPMCP patient database for inpatient and outpatient physician diagnoses of CP (International Classification of Diseases, Ninth Revision, Clinical Modification¹⁷ [ICD-9-CM] codes 343.0-343.9), paresis (ICD-9-CM codes 342.1, 342.8, 342.9, 344.0, 344.1, 344.30-344.32, and 344.5), or gait abnormality (ICD-9-CM code 781.2). A single pediatric neurologist (Y.W.) who was blinded to the presence of chorioamnionitis then reviewed the outpatient medical records to confirm the diagnosis of CP.

We defined CP as a nonprogressive congenital motor dysfunction with findings of spasticity, rigidity, or choreoathetosis. Inclusion criteria were 36 or more weeks' singleton gestation, nonprogressive motor dysfunction, examination findings of increased tone or choreoathetosis, and moderate to severe disability. Moderate disability was defined as diminished use of the most affected limb, and severe disability referred to the lack of any functional use of the most affected limb.⁵ Hypotonic CP in the absence of increased tone was not included because this entity is most likely etiologically distinct from spastic and dyskinetic CP.

All children were older than 2 years at the onset of the study. Since CP can be definitively diagnosed at age 2 years,¹⁸ those children who met case eligibility criteria after age 2 years were considered "definite" cases. Children with CP but who were lost to follow-up between ages 15 and 24 months were considered "probable" cases, and also included in the study. Infants who met inclusion criteria but who were lost to follow-up before age 15 months were excluded, since early motor abnormalities suggestive of CP may resolve by childhood.¹⁹

Because we were interested in perinatal risk factors for unexplained CP, children with postnatal brain injury or known developmental abnormalities were excluded. Criteria for exclusion were a central nervous system insult occurring after 1 week of age; a known developmental or genetic abnormality, such as a brain malformation or chromosomal anomaly; evidence of a congenital viral infection; and presence of a condition that is not considered to be CP,²⁰ including neural tube defects and arthrogryposis (Box).

237.7x Neurofibromatosis
275.1 Wilson disease
277.2 Lesch-Nyhan syndrome
277.5 Mucopolysaccharidosis
331.8 Reye syndrome
331.9 Cerebral degeneration, unspecified
333.6 Idiopathic torsion dystonia
334.x Spinocerebellar disease, including ataxia telangiectasia
335.x Anterior horn cell disease
336.x Other diseases of spinal cord
340 Multiple sclerosis
349.82 Toxic encephalopathy
358.x Myoneural disorders (eg, myasthenia gravis)
359.x Muscular dystrophies and other myopathies
741.xx Spina bifida
742.5x Spinal cord anomalies
754.59 Arthrogryposis
755.55 Apert syndrome
756.16 Klippel-Feil disease
757.33 Bloch-Sulzberger disease (incontinentia pigmenti), xeroderma pigmentosum
758.x Chromosomal anomalies
759.5 Tuberous sclerosis
759.81 Prader-Willi Syndrome
759.89 Other specified anomalies, including Cornelia de Lange syndrome, Lawrence Moon Biedl syndrome, Rubenstein-Taybi syndrome, Carpenter syndrome, cerebrohepatorenal syndrome, Cockayne syndrome, Menkes kinky hair disease

We randomly selected 2 controls per case from the study population. We used electronic records to determine the length of follow-up at the KPMC for all control children.

A trained medical record abstractor reviewed obstetric and neonatal charts without knowledge of neurologic outcome. Chorioamnionitis was considered present if a treating physician made a diagnosis of chorioamnionitis or endometritis based on clinical symptoms. Histologic chorioamnionitis was defined as microscopic evidence of inflammation in the placental membranes or umbilical cord. Intrauterine growth restriction was defined as birth weight less than the 10th percentile for gestational age based on race- and sex-specific normative data compiled from California births.²¹

"Birth asphyxia" is a vague and controversial term that denotes a clinical diagnosis lacking specificity for any single underlying pathological condition. To study the significance of this diagnosis as it is used in practice, we abstracted this terminology as it was used by treating physicians. That is, we considered birth asphyxia to be present if a treating physician diagnosed an infant clinically with either birth asphyxia or hypoxic-ischemic encephalopathy, regardless of whether true hypoxia-ischemia was present.

In contrast, we use the term "hypoxic-ischemic brain injury" to refer either to the neuroradiological findings of brain injury in the parasagittal watershed distribution, basal ganglia, or thalami, or to diffuse brain edema observed in the first days of life. Although the specificity of these findings for hypoxia-ischemia is unknown because there is no way to directly measure intrapartum blood flow and oxygen delivery to the brain,²² these findings are often taken to imply the presence of perinatal hypoxia-ischemia in neonates with encephalopathy.^23- 26

We calculated univariate odds ratios (ORs) and 95% confidence intervals (CIs) using the exact method, and compared univariate ORs stratified by type of imaging abnormality using polytomous logistic regression.²⁷ We calculated multivariable ORs by performing a backward stepwise logistic regression, with P<.10 used as the cutoff for retention in the model. Odds ratios are close approximations of the relative risk since the outcome of CP is rare in term infants.

Potential confounders found to be significantly associated with CP on univariate analysis were included in the multivariable model, as were risk factors that were considered a priori to be potential confounders. Factors that were not considered to be etiologic, such as low Apgar scores, emergent cesarean section, neonatal seizures, and a diagnosis of "birth asphyxia," were not included in the model. These abnormalities are likely to be markers of the underlying brain injury leading to CP, and adjusting for them could falsely diminish the association between chorioamnionitis and CP. Instead, we tested the hypothesis that these factors may in some cases be due to chorioamnionitis by looking for associations between chorioamnionitis and these clinical markers in infants with CP. STATA version 7 (StataCorp LP, College Station, Tex) was used for all analyses; P<.05 was used to determine statistical significance.

Of the 231 582 infants in the study population, the electronic search for relevant physician diagnoses revealed 618 children with possible CP (Figure 1). Chart review confirmed the diagnosis of CP in 206 children. The prevalence of CP was 0.9 per 1000 live births (approximately 1/1100), which is consistent with previous population-based estimates of CP in term infants.^3,5- 6,28

After excluding 97 children with CP for reasons such as mild disability or the presence of a developmental anomaly (Figure 1), the remaining 109 children with unexplained moderate to severe CP comprised the final cases. The vast majority (93%) of these children had been followed up for at least 2 years, and 87% had been diagnosed by a neurologist as having CP (Table 1). Most cases had hemiparesis (40%) or quadriparesis (38%). Two hundred eighteen controls were selected.

Magnetic resonance imaging or computed tomography of the head had been performed in 83% of children with CP (n = 75 and n = 16, respectively). The most common abnormalities were focal infarct, white matter abnormalities sparing the cortex, hypoxic-ischemic brain injury, and generalized atrophy (Table 1).

Twenty-six of the 35 infants (74%) with hemiplegic CP who received an imaging study of the head were diagnosed with a focal infarct. Patients who underwent a magnetic resonance imaging scan were more likely to be diagnosed with a perinatal stroke than were those who underwent only a computed tomography scan (36% vs 13%, respectively; P = .07). Only 6 of 24 infants (25%) diagnosed clinically with birth asphyxia who underwent neuroimaging studies demonstrated findings specific for hypoxic-ischemic brain injury. All children with hypoxic-ischemic brain injury had spastic quadriplegia.

Chorioamnionitis was diagnosed in 14% of cases and 4% of controls (OR, 3.8; 95% CI, 1.5-10.1; P = .001). Maternal fever, prolonged rupture of membranes (≥36 hours compared with <24 hours), nulliparity, and IUGR also were associated with CP (Table 2). The OR of chorioamnionitis and CP did not differ significantly between mothers who received intrapartum antibiotics and those who were not given antibiotics (OR, 6.8 [95% CI, 1.2-70] vs 2.6 [95% CI, 0.5-13.0]; P = .37), though our numbers were small.

A histologically confirmed diagnosis of chorioamnionitis was made in 5 of 19 children (26%) with CP vs 1 of 9 (11%) control children (P = .36). Five of 6 mothers with histologic evidence of placental inflammation lacked any clinical signs of chorioamnionitis, while 6 of 8 mothers diagnosed clinically with chorioamnionitis had no placental inflammation. Since placental examinations were not performed in the vast majority of study participants, histologic chorioamnionitis was not included in further analyses.

Several factors that we considered to be clinical markers of fetal distress and brain dysfunction, rather than causes, were strongly associated with CP, including 5-minute Apgar score less than 7, the clinical diagnosis of "birth asphyxia," and neonatal seizures (Table 2). As hypothesized, however, these same findings were associated with chorioamnionitis among the children with CP (5-minute Apgar score <7 [OR, 5.3; 95% CI, 1.4-19.5], birth asphyxia [OR, 4.9; 95% CI, 1.3-18.1], and neonatal seizures [OR, 5.6; 95% CI, 1.5-21.3]). None of the 9 control children born to mothers with chorioamnionitis demonstrated these or other neonatal complications, although 2 infants had brief temperature elevations immediately after birth.

Chorioamnionitis remained an independent risk factor for CP in the multivariable model (OR, 4.1; 95% CI, 1.6-10.1). Other risk factors included intrauterine growth restriction (OR, 4.0; 95% CI, 1.3-12.0), maternal age older than 25 years (OR, 2.6; 95% CI, 1.3-5.2), black race (OR, 3.6; 95% CI, 1.4-9.3), and nulliparity (OR, 1.8; 95% CI, 1.0-3.0) (Table 3). Although black race did not represent a significant risk factor in univariate analysis, it was found to be independently associated with CP after adjustment for confounders. This discrepancy may be explained in part by the fact that black mothers were on average younger than their white counterparts. Younger maternal age conferred a protective effect, and without adjustment for this confounder, maternal age would act to suppress the effect of ethnicity on risk of CP. Of note, none of the black women in the case or control groups had chorioamnionitis.

The relationship between chorioamnionitis and CP varied according to subtype of CP. In addition, the OR of chorioamnionitis for CP was higher among children with a neuroimaging diagnosis of hypoxic-ischemic brain injury than among those with other neuroimaging findings (univariate OR, 17.2 [95% CI, 3.3-88] vs 3.2 [95% CI, 1.2-8.1]; P = .04).

Multivariable analyses stratified by CP subtypes were performed when numbers were adequate (Table 4). The adjusted OR of chorioamnionitis for CP was particularly high among children with quadriplegic CP, and among those with neuroimaging findings of hypoxic-ischemic brain injury. Chorioamnionitis was not an independent risk factor for CP resulting from focal infarction. Instead, when we performed the multivariable analysis restricted to the 29 children with focal infarction, preeclampsia (OR, 3.4; 95% CI, 1.0-11.8) and intrauterine growth restriction (OR, 4.3; 95% CI, 1.0-18.3) were the only significant risk factors retained in the model.

Children with apparent CP but who were followed up for less than 15 months were excluded from the study. When the 13% of control children who were lost to follow-up prior to 15 months were also excluded, the relationship between chorioamnionitis and CP did not change appreciably.

We found that clinical chorioamnionitis is independently associated with a 4-fold increased risk of CP in term infants. A history of birth asphyxia was also strongly predictive of CP, a finding that is supported by recent reports.^9,29- 31 However, the diagnosis of birth asphyxia often was made in the setting of clinical chorioamnionitis.

The association between maternal fever and CP was first reported in 1955.³² Subsequently, 2 of the 3 available studies were unable to confirm this association in term infants after adjusting for confounders.^10,12 However, children with CP of all causes were analyzed together in these studies. We found that children with CP due to a developmental or genetic abnormality represented 19% of all children with CP. Our data suggest that when these children are excluded from the analysis, chorioamnionitis is indeed associated with an increased risk of CP.

How chorioamnionitis might lead to CP is poorly understood; most likely, several mechanisms act together to cause fetal brain injury. Hypothesized mechanisms include (1) elevated levels of fetal cytokines in the presence of maternal infection cause direct injury to the fetal brain (ie, the fetal inflammatory response)^33- 34; (2) inflammation of the placental membranes leads to interruption of placental gas exchange and blood flow, resulting in hypoxic-ischemic brain injury in the fetus³⁵; (3) maternal fever raises the core temperature of the fetus, which in turn may be harmful to the developing brain, especially in the setting of cerebral ischemia^35- 37; and (4) maternal intrauterine infection leads to direct infection of the fetal brain or meninges, although this is seen rarely.^9,14

The finding of a particularly strong association between chorioamnionitis and CP among the 6% of children who sustained hypoxic-ischemic brain injury was unexpected and based on small numbers. However, this finding supports the second hypothesis, ie, that chorioamnionitis plays a role in initiating or exacerbating brain injury from hypoxia-ischemia. Chorioamnionitis leads to an elevation of inflammatory cytokines in the fetus.³³ In animal models, the inhibition of these cytokines (interleukin 1 and tumor necrosis factor α) has been shown to reduce the extent of brain injury following hypoxia-ischemia.^38- 40 Although chorioamnionitis may potentiate hypoxic-ischemic brain injury in neonates, it also is possible that inflammation produces a pattern of brain injury that mimics the neuroradiological findings of hypoxic-ischemic brain injury in term infants.

The term "birth asphyxia" is imprecise and misleading, as it implies the existence of hypoxic-ischemic brain injury. Our results suggest that chorioamnionitis may be a cause of the neonatal syndrome that often is referred to as birth asphyxia. Indeed, intrapartum fever has been found to represent a significant risk factor for signs of birth asphyxia, such as low Apgar scores and neonatal encephalopathy.^9,13- 15 In our study, only 25% of children diagnosed with birth asphyxia demonstrated neuroradiological findings suggesting hypoxic-ischemic injury. Although the diagnosis of birth asphyxia was a strong predictor of CP in our population, it is important to emphasize that apparent birth asphyxia may be a marker of multiple potential pathways leading to injury of the developing brain. For this reason, "neonatal encephalopathy" is emerging as the preferred term for describing the status of neurologically depressed newborns,^41- 43 given that it does not imply a specific underlying pathogenesis.

Focal infarction was a common cause of hemiplegic CP in our patients, a finding that is consistent with other reports.^44- 45 It is hypothesized that the naturally occurring prothrombotic state of pregnancy, combined with a hereditary or acquired thrombophilia and the presence of a right-to-left shunt, all contribute to the relatively high incidence of thromboembolic stroke in the newborn period.⁴⁶ Among children with focal infarction in our study, preeclampsia was identified as a risk factor in the multivariable model. It is unclear whether preeclampsia and perinatal thromboembolic stroke represent outcomes that both stem from a common underlying prothrombotic condition^47- 49 or whether they are causally related via vasculopathy and clot formation within the placenta.

The incidence of CP in term infants has not previously been reported to vary by ethnicity, although few studies are available.^3,5,50 In contrast, we found that among singleton term births, black ethnicity was associated with a 3.6-fold increased risk of CP, after adjusting for confounders. Our results also support previous findings that chorioamnionitis may increase the risk of CP among whites but not blacks.⁵¹ Whether these ethnic differences are due to genetic or other factors deserves further investigation.

Our study is limited by multiple comparisons and by the possibility that other confounders may have been overlooked. In addition, due to small subgroup sample sizes, our stratified analyses produced ORs with wide CIs. Children with CP were not examined by study investigators to confirm their diagnoses, and the lack of standardization in both the type and timing of neuroimaging studies could lead to potential bias in the ascertainment of brain imaging abnormalities. Selection bias also may have occurred given the incomplete follow-up, though restricting both case and control infants to those with 15 or more months' follow-up did not alter the results.

Assuming that chorioamnionitis plays a role in causing CP, the population-attributable fraction of chorioamnionitis for CP without a known developmental cause is about 11% among singleton term infants. This number is even higher for spastic quadriplegia (27%), and also may increase if we improve our ability to diagnose chorioamnionitis, because the current set of criteria for chorioamnionitis is likely to have low specificity as well as sensitivity for diagnosing the true underlying etiologic agent.^52- 55

There are currently no evidence-based strategies for preventing CP in term infants. Recent studies have found that neonatal serologic markers of inflammation are associated with future CP.^56- 57 If these inflammatory markers are mediators of perinatal brain injury associated with chorioamnionitis, strategies for protecting the developing brain from such perinatal inflammation may be beneficial for preventing CP.

We conclude that chorioamnionitis confers a 4-fold overall increased risk of CP in term infants, and that future etiologic studies of CP will be strengthened if children are stratified by type of CP and neuroimaging findings.