Cord gas parameters in infants born to women with sickle cell disease: a retrospective matched cohort study.

Abstract

Women with sickle cell disease (SCD) often experience adverse fetal events (Villers et al, 2008), probably placentally-mediated (Rathod et al, 2007). Given the placental role in fetal gas-exchange, umbilical cord blood gas (UCBG) analysis provides insight into intrapartum placental function and fetal acid-base status (American College of Obstetricians and Gynecologists [ACOG] Committee on Obstetric Practice 2006), enabling detection of metabolic acidosis, with established links to adverse outcomes (Armstrong & Stenson, 2007) but has not been studied in SCD. We aimed to determine whether UCBGs in infants of SCD-affected women (cases) differ from infants of non-affected women (controls). This was a retrospective cohort study of SCD-affected pregnant women attending the Special Pregnancy Program at Mount Sinai Hospital (MSH), Canada (January 2004–December 2015). Inclusion criteria comprised SCD-diagnosis (HbSS, HbSC, HbS/Beta-Thalassaemia). Exclusion encompassed fetal malformations, intrauterine fetal death and multiple gestations. Controls were identified through MSH delivery records in a 2:1 ratio matched by delivery-date (within 1 year), then by maternal age (within 5 years), parity (nulliparous versus multiparous), gestational age at delivery (within 1 week) and delivery type (elective or emergent C-Section, spontaneous or assisted vaginal delivery). Exclusion criteria for controls were as for cases plus chronic conditions (e.g. diabetes mellitus, hypertensive disorders, autoimmune conditions, renal or hepatic dysfunction). Primary outcomes included: arterial and venous cord pH and base excess (BE). Secondary outcomes included: arterial and venous cord oxygen saturation (aSaO2 and vSaO2, respectively), fetal oxygen extraction (vSaO2 − aSAO2)/vSaO2), 5-min Apgar score <7, and Neonatal Intensive Care Unit (NICU) admission. Given that normal venous results do not exclude large arterio-venous differences (possibly overlooking metabolic acidosis), and a single result is typically venous owing to ease of sampling (Westgate et al, 1994; Armstrong & Stenson, 2007), collection from both vessels was required. Samples with pH difference <0·02 or pCO2 difference <3·8 mmHg were excluded assuming single vessel, while arterial samples were differentiated from venous by lower pH and higher pCO2 in the former (Westgate et al, 1994; Armstrong & Stenson, 2007). Significance was assessed by repeated measures ANOVA for continuous variables and conditional logistic regression for categorical variables, accounting for case-control matching. P < 0·05 denoted significance. Multivariate linear regression assessed influence of potential confounding variables (Table 1), where differences between cases and controls were identified. Pearson correlation assessed pre-delivery haemoglobin and UCBG values. Of 198 cases, 93 were excluded (39 sickle trait, 5 multiple gestation, 1 fetal malformation, 7 IUFD, 14 UCBG unavailability, 9 single-vessel, 7 insufficient controls, 11 non-pregnant), yielding 105 cases, for which 210 matched-controls were selected. The groups did not differ on matched characteristics, except maternal age (Table 1). Groups did not differ for most potential confounders, except body mass index (BMI), birthweight, small for gestational age (SGA) and maternal haemoglobin. Cases differed from controls on all primary outcomes, but none of the secondary outcomes (Table 2). Differences remained significant following multivariate linear regression, accounting for confounding by maternal age, BMI and SGA. Whilst maternal pre-delivery haemoglobin positively correlated with primary outcomes, it did not correlate with fetal aSaO2 or vSaO2 (Table SI). UCBG analysis offers an impartial benchmark of fetal metabolic status and response to peripartum events (ACOG Committee on Obstetric Practice 2006). While arterial pH reflects acidaemia, BE exposes its metabolic component, with pH <7·0 and BE ≥12 mmol/l heightening risk of persistent neurological deficit (MacLennan, 1999). Physiologically, BE progressively declines through the active first-stage of labour, accelerating in second-stage (Ross & Gala, 2002), placing fetuses entering labour with lower pH/higher BE at risk of exhausting their reserves to the point of pathological acidaemia. Where fetal compensatory responses remain sufficient despite antenatal challenges, the tipping point of irreversible cerebral damage might occur intrapartum, as compensatory responses fail (MacLennan, 1999). Lack of between-group differences for NICU-admission or Apgar scores in our study is not surprising, as few UCBGs reached arterial pH <7·1, associated with such outcomes (Georgieva et al, 2013). It is however worth noting that consequences of lower UCBGs occur in context of a “significant inverse curvilinear relationship between cord pH and BE” (Victory et al, 2004), suggesting progressively accelerated risk of adverse neurological events at lower ranges of deteriorating values (Victory et al, 2004). There were no between-group differences in fetal oxygen saturation and extraction. This relative stability of fetal SaO2 in infants of SCD-affected women may be secondary to high fetal-blood oxygen-affinity (promoting transplacental oxygen transfer) (Carter, 2015) and lower maternal-blood oxygen affinity of sickle haemoglobin compared to normal haemoglobin (altered Bohr Effect), resulting in marked liberation of oxygen from maternal cells, particularly at pH<7·4 (Ueda et al, 1979). Yet, preservation of fetal oxygenation in this manner may come at the expense of increased maternal sickling, given the propensity for sickle haemoglobin formation in de-oxygenated states (Ueda et al, 1979). Whilst we saw a correlation between maternal pre-delivery haemoglobin and primary outcomes, no such correlation existed for aSaO2 or vSaO2. This parallels prior observations of placental capacity for histological adaptation, with consequential thinning of the placental barrier to maintain fetal oxygen-transfer in hypoxic states stemming from reduction in oxygen content of maternal blood (e.g. maternal anaemia) (Kingdom & Kaufmann, 1997). Although during short-term response to maternal anaemia, uterine blood flow remains unaltered, uterine oxygen delivery is hardly reduced, and uterine/placental/fetal oxygen-consumption is negligibly lower (Carter, 2015), thus maintaining high placental oxygen-consumption; during long-term hypoxic conditions, placental oxygen-consumption falls, promoting fetal oxygen availability (Carter, 2015). Thus, the lack of difference in oxygen saturation/extraction between study groups potentially stems from the relative maintenance of adaptive capacities of feto-placental oxygen exchange and absence of acute-onset, severe maternal anaemia, while diminished fetal growth velocities in SCD-affected women probably reflect subtle, ongoing, long-term hypoxic insults (Carter, 2015). The retrospective nature of our study is limiting; however, meticulous accounting for confounding factors during design and analysis strengthens our findings. Notably, matching for delivery date and delivery mode removes confounding potentially introduced by changes in practice over time. As oxygen supplementation is part of intrapartum care for SCD-affected women, we were unable to control for maternal oxygen-use during labour; however, an even greater effect on the observed differences would be expected with elimination of oxygen. Our findings of lower cord pH and higher BE in infants of SCD-affected women underscore that these pregnancies remain at high-risk of adverse events peripartum. The importance of consideration of the duration of labour and vigilance in monitoring of fetal well-being, particularly during prolonged labour, cannot be overemphasized. The complex nature of these pregnancies calls for collaborative management by Maternal-Fetal Medicine and Haematology specialists, in centres with the requisite expertise. AKM: study concept, data collection, analysis and interpretation of data, drafting of manuscript, manuscript revision, approval of final version. PC: data collection, analysis and interpretation of data, manuscript revision, approval of final version. JY: statistical analysis and interpretation of data, manuscript revision, approval of final version. RD: study concept, interpretation of data, manuscript revision, approval of final version. NS: interpretation of data, manuscript revision, approval of final version. RW: interpretation of data, manuscript revision, approval of final version. KHMK: interpretation of data, manuscript revision, approval of final version. KEM: study concept, interpretation of data, manuscript revision, approval of final version. The authors report no conflict of interest. This work was supported by the Comprehensive Research Experience for Medical Students (CREMS) Summer Program at the University of Toronto. The funder played no role in conducting the research or writing the paper. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.