Evaluation of Metabolic Acid-base Abnormalities in Pre-eclamptic Women Applying the Stewart-Fencl Approach

Studies evaluating acid/base imbalances are scarce in pre-eclamptics, and tools to thoroughly evaluate them are lacking. The detection of metabolic acidosis has been proposed as a diagnostic criterion for a severe disease state8, 11, however other studies entirely reject the presence of any metabolic acid-base disorder in pre-eclampsia10. One reason for this controversy may lie in the fact that no study performed a full comprehensive acid base analysis. Whereas some measured plasma pH, HCO3 and BE8, 11 others focused on AG1 or AG and HCO32, 10 only. Merely one study corrected the AG for low albumin1, and no study focused on independent variables. While some of the cited reports were based on retrospective chart review without differentiating mild and severe disease state1, 2, the few prospective studies available were of low sample size including a total number of 311 or 98women with severe pre-eclampsia. As indicated above, it is totally legitimate to use the traditional approach to describe changes in hydrogen ion concentration and HCO3. The Henderson-Hasselbach equation does differentiate well between respiratory and metabolic acid-base changes. However, it is unable to quantify metabolic causes15. SBE quantifies net change in metabolic acid-base status in vivo, but does not tell us about mechanisms16. The AG is insufficient to detect unmeasured anions17, 18, but when corrected for low albumin it does not detect all unmeasured cations16.

Intravascular volume depletion and disorders of electrolytes and albumin are common findings in pre-eclampsia1, 2 and the impact of such disorders on acid-base homeostasis has been increasingly acknowledged in the last decade3, 19. Decreased serum albumin has an alkalizing effect on plasma, resulting in hypoalbuminemic alkalosis. Furthermore, derangements in volume, sodium and chloride homeostasis, as well as the accumulation of unmeasured anions, are common findings in pre-eclampsia7 and may result in multiple acid-base disorders. The presence of hyponatremia or hyperchloremia has an acidifying effect on blood pH, whereas hypernatremia or hypochloremia lead to alkalosis. The accumulation of unmeasured anions, such as lactate or uric acid, will lead to acidemia, if not otherwise compensated3, 20.

We hypothesize that with comprehensive analysis, clinically significant and complex metabolic derangements might be discovered in pre-eclamptics that will potentially impact their care plan. To our knowledge, this will be the first time that the Stewart-Fencl methodology will be used to determine causes of acid-base status in pre-eclamptics, which takes into account the multiple, potentially opposing, electrochemical disturbances observed in this patient population.

Specific Aims

Aim 1: Prevalence

The primary aim is a description of the prevalence of acid-base disorders in pregnant women with pre-eclampsia at three time points:

1. at time of diagnosis with mild pre-eclampsia,
2. at time of progression to severe pre-eclampsia, and
3. at time of delivery. Pregnant women without pre-eclampsia will be used as a comparison group for (a) and (b) for testing whether the.prevalence of acid-base disorders is higher in pre-eclampsia than in pregnancy without pre-eclampsia.

Aim 2: Changes over time in acid base status This aim is to describe and test change over three time points in the prevalence of acid-base disorders among those women who progress from mild to severe pre-eclampsia, and change from initial diagnosis to delivery among women whose pre-eclampsia remains mild.

Aim 3: Acid-base as clinical predictor This aim will test whether acid-base variables at time of diagnosis with mild pre-eclampsia predict which women will progress to severe pre-eclampsia and/or need an urgent delivery.

Significance (Note-include significance in relation to translational health science)

Pre-eclampsia complicates 2-8% of pregnancies, and is a major contributor to fetal and maternal morbidity/mortality worldwide5, 21, 22.

The only cure for pre-eclampsia is delivery. However, especially remote from term, the premise that delaying delivery to allow for the fetus to continue to grow and thrive in utero is a big part of what expectant management is trying to achieve. The challenge remains to this day to balance the benefits of a prompt delivery (undeniably benefiting the mother) versus allowing the fetus to remain in utero (although if placental insufficiency is severe and endothelial dysfunction critical, the baby may well be better off delivered). Unfortunately, there are no strict criteria (hemodynamic, metabolic or other) or any predictive model that may guide obstetricians in the decision to proceed with a prompt delivery or allow prudent expectant management. Therefore, management is mostly guided by expert opinion-based guidelines, which do not accurately predict adverse maternal events23, 24. The fullPIERS-model, recently published results from a large clinical trial, considers gestational age, chest pain or dyspnea, oxygen saturation, platelet count, serum creatinine levels and aspartate transaminase concentration, to predict adverse maternal outcome with a sensitivity of 75% and a specificity of 87%25. However, for most of these predictors the worst values recorded during the first 48 hours of study were used to predict complications within the same 48 hour time frame 26. In practice, a model based exclusively on information available at admission is still needed. The data evaluated by the proposed Stewart-method would add to the physiologic understanding of the disease, would be easy to obtain at admission and possibly discover new predictors for adverse maternal and perinatal events.

Research Approach

In this prospective case-control study, previously normotensive women diagnosed with mild pre-eclampsia as defined by the American Association of Hypertension 27 will be enrolled. Blood samples from women with pre-eclampsia and healthy controls will be gathered, and plasma pH, HCO3- and BE will be measured. According to the Stewart-Fencl methodology these parameters are dependent variables of the following four factors 12, 28:

• The difference of strong cation and anion concentration, or 'strong ion difference' (SID)
• The pCO2
• The concentration of weak acids, which are mainly albumin and phosphate (Atot)
• The presence of other acids, so called unmeasured anions (UMA) or strong ion group (SIG)

Based on that concept, BE can be defined by:

(A) plasma dilution/concentration (=changes in Na+-concentration, (BENa+)) (B) changes in chloride concentration (BECl-) (C) changes in albumin concentration (BEAlb) (D) changes in lactate (BELac) (E) changes in unmeasured anions (BEUMA). BE = BENa+ + BECl- + BEAlb + BELac + BEUMA

Each component will be identified and quantified as described by Gilfix et al.17. (A detailed explanation can be found in the Appendix of this application).

Statistical Analysis Because of the descriptive and exploratory nature of this pilot study, no controls for multiple comparisons will be made. Any results from this study will be interpreted as hypothesis generating, to justify a larger further study.

Aim 1: Prevalence The primary aim is a description of the prevalence of acid-base disorders in pregnant women with pre-eclampsia at three time points: (a) at time of diagnosis with mild pre-eclampsia, (b) at time of progression to severe pre-eclampsia, and (c) at time of delivery, in comparison to pregnant women without pre-eclampsia.

For each of the BE measures, abnormal will be defined as differing from +/-2mEq/l or more. If any one of the BE measures is abnormal, the woman will be classified as having an acid-base disorder. Descriptive displays (tables and/or bar charts) will be used to describe the percent of women with an acid-disorder at each of (a) to (b) and in control women. In addition, we will describe how many of these cases have offsetting disorders so that the total BE is normal, i.e. they would not have been identified by the traditional approach. Further displays will describe the prevalence of abnormal values for each of the 5 components separately. In addition, we will explore the utility of displaying means and SDs of each of the 5 BE components.

Logistic regression will be used to test for differences in prevalence of acid-base disorder in pre-eclampsia versus controls, at each of (a) and (b) time points, while controlling for gestational age as a covariate. Linear regression will be used to test for differences in the means of each of the BE components between pre-eclampsia and controls, controlling for gestational age.

Aim 2: Changes over time in acid base status This aim is to describe and test change over three time points in the prevalence of acid-base disorders among those women who progress from mild to severe pre-eclampsia, and change from initial diagnosis to delivery among women whose pre-eclampsia remains mild.

Mixed models for binary outcome data, using for example xtlogit in Stata, will be used to test for changes over time in prevalence. Further analyses will use mixed models for continuous outcomes to test for changes in means of the individual BE scores over time.

Aim 3: Acid-base as clinical predictor This aim will test whether acid-base variables at time of diagnosis with mild pre-eclampsia predict which women will progress to severe pre-eclampsia and/or need an urgent delivery. A chi-square test can be used to test whether the fraction of women progressing to sever pre-eclampsia is different in those with abnormal versus normal acid-base at time of diagnosis of mild pre-eclampsia. Further analysis will use logistic regression analyses to test whether the 5 BE components predict probability of progression. Potential confounding variables such as medications (e.g furosemide) used to control symptoms or severity of the disease will also be evaluated to see whether controlling for these variables changes the association of acid-base variables with risk of progression.

We will further explore a variety of neonatal and maternal outcome variables for a possible association with acid-base predictors. Please find those secondary outcome variables described in detail in the appendix of the revised application. Please find also clarified if variables are categorical or continuous.

• Sample size

Sample size is based on the observation in a pilot sample of 25 paired arterial and venous blood specimens in women with severe pre-eclampsia of mean venous BE(Alb) = + 4.4 (± 0.75) mEq/l. Considering a power of 1.0 and an alpha of 0.05, a sample size of 25 women in each group was needed to show a difference of 20% in BE(Alb) between cases and control groups.