Aspects to Consider in Adopting Pregnancy-Specific Reference Intervals

Introduction

This commentary will focus on the changes in maternal physiology which result in changes in electrolyte and acid-base status during pregnancy. These are also the biochemical parameters being considered for the first round of harmonisation of pregnancy reference intervals by the AACB. The analytes addressed in this commentary will include sodium, potassium, chloride, bicarbonate, urea and creatinine.

The currently-proposed reference intervals in pregnancy (as discussed at the 2015 AACB Harmonisation Meeting) are outlined. However it should be noted that these are currently not ratified and may be subject to change. All laboratories are encouraged to assess their own pregnancy data against these proposed intervals and provide feedback to the AACB Harmonisation Committee. Data from a large Australian data set of pregnant women has been provided (Figure 1) and may provide a helpful reference point for individual laboratories as they conduct their own internal data reviews.

Figure (A to G).

Changes in electrolyte parameters observed in 30,321 women of varying gestational age. The gridlines represent the non-pregnant reference limits for young women and the slanted percentages given at the 50^th centiles represent the percentage of the pregnant median compared to the non-pregnant median.

Why are Pregnancy Reference Intervals Needed in Clinical Biochemistry?

Reference intervals are designed to identify the range of results observed in healthy individuals. If healthy physiology changes, such as in pregnancy, a different interval may be required to identify health in this setting. Having appropriate reference intervals for pregnancy allows clinicians to avoid interpreting normal results as pathological and assists them in identifying results which are truly abnormal.^1,2 Taking the further step of providing common reference intervals in pregnancy provides uniform information and may assist clinicians by reducing confusion and may optimise patient care.³

Not all biochemical parameters are altered by changing physiology in pregnancy but for those that are, it is necessary to consider whether the change is sufficiently significant to warrant a different reference interval.⁴ Whether a change is significant or not is ideally determined by clinical outcome,⁵ but if this is not possible, significance of a change may be determined by clinical opinion or statistical evidence.⁴

Partitioning Within Pregnancy is Arbitrary

When considering whether different references intervals are needed for discrete periods during pregnancy, similar principles for partitioning of any group apply. If, during different stages of pregnancy, further significant changes occur, further partitioning of pregnancy reference intervals may also be required.⁵ Changes are slowly and continuously occurring during pregnancy, and it may be very difficult and result in further complexity to have separate reference intervals for every week or day of pregnancy for every analyte that changes. There are, however, instances when this is appropriate. The approach of providing a ‘day-specific’ reference interval for some analytes in pregnancy is, in fact, in place. An example is the reporting of beta subunit human chorionic gonadotrophin (ßhCG) and pregnancy-associated plasma protein A (PAPP-A) used in risk assessment of Down syndrome in pregnancy.⁶ These levels are changing every day in pregnancy (one falling during the time of interest, and one rising), and these small differences are important in the risk assessment calculations. However to display the different reference intervals for every day of pregnancy for two different analytes would be complex and may be potentially confusing. Rather than providing different intervals with every set of results, the median for the gestation is used as a reference point and the individual values are reported as multiples of the expected median (MoM) for the day of gestation. This provides a simple mechanism by which the value of the changing analytes can be assessed against the expected reference median for each day of gestation.

Dating a Pregnancy and Defining Partitions

Human gestation (pregnancy) averages 40 weeks or 280 days in length from the time of the last menstrual period (LMP).⁷ Although the timing of ovulation may vary, this is generally around day 10–14 of cycle and two weeks after the LMP. As such, when the ovum unites with a sperm and fertilisation occurs at ovulation, dating from the LMP means that the woman is already 2 weeks pregnant. When next menstruation is due and the conceptus is two weeks old, the pregnancy is dated as being of four weeks of gestation.

Obstetricians have traditionally divided the nine months of human gestation into three trimesters.⁷ Although separate reference intervals may be used when significant changes occur, the dominant form of partitioning in pregnancy is via social consensus, with the traditional division of pregnancy into three 14 week periods – the trimesters.⁷ Thus the 1^st trimester would be 1 day of pregnancy to 13 weeks 6 days of gestation, the 2^nd trimester 14 weeks 0 days until 27 weeks 6 days and the final 3^rd trimester 28 weeks 0 days through until term.

Although arbitrary partitions are not ideal, these divisions are well recognised and offer the advantage of familiarity for most clinicians. However it is important to note that reference intervals have been published for various partitions of pregnancy. These have included trimesters of pregnancy or changes in 4–8 week periods.^1,2,8–10 As changes throughout pregnancy do not happen arbitrarily at the trimesters,⁷ there is a sound basis for considering the biochemical changes in pregnancy over shorter time periods. Over the months of pregnancy profound changes occur not only for the developing embryo (5–10 weeks) and foetus (11–40 weeks) but also in the mother.¹¹

Vascular and Renal Physiology Changes in Pregnancy

In early pregnancy (even by 6 weeks gestation) there is a change in systemic and renal vascular tone with vasodilation of peripheral and renal vasculature.^12–15 This primary vasodilation is important as it provides an explanation for many subsequent changes of pregnancy including lower measured sodium, volume expansion, reduced haematocrit and reduced serum albumin.¹³ Early vasodilation also results in lower blood pressure and increased cardiac output, increased renal blood flow and increased glomerular filtration rate (GFR) together with reduced serum creatinine.¹³

Sodium

Lower Serum Sodium, Increased Plasma Volume, Haemodilution

Sodium and water retention occur secondary to early vasodilation to increase filling pressures with a subsequent increase in plasma volume, extracellular volume and cardiac output.¹³ It is estimated that plasma volume may increase 30–50% during pregnancy,¹³ extracellular volume may increase by 6–7 L¹⁵ and an extra 500–900 mmol of sodium is retained.^9,15 The increased plasma volume contributes to the reduced haemoglobin and haematocrit of pregnancy¹⁶ and also contributes to the lowered albumin by haemodilution.¹⁵ Despite the overall increased total body retention of sodium during pregnancy, the increased plasma volume and alterations in osmolality set point result in mildly reduced measured serum sodium concentrations.

The effect on serum sodium is a decrease of approximately 2–5 mmol/L⁹ with reference intervals being decreased in pregnancy to ~97–98% of normal non-pregnant intervals.⁹ Partitioning for trimesters during pregnancy does not appear to offer further clinical advantage as the changes between trimesters are not significant. The reference interval proposed at the AACB Harmonisation Meeting for sodium was 132–142 mmol/L; however it is acknowledged that this may need revising once laboratories have assessed their own data against the proposed intervals. It is noted that the intervals on the data set in Figure 1A suggest 134 mmol/L may be a more appropriate lower limit.

Reduced Plasma Osmolality

The other important change in water handling during pregnancy is the reduced plasma osmolality associated with the mild reduction in plasma sodium.¹³ Plasma osmolality falls as early as 5 weeks gestation and reaches a nadir by 10 weeks of pregnancy.¹⁵ There is up to a 10 mmol/kg drop in osmolality with vasopressin secreted at lower osmolalities than in the pre-conception state.^9,17–19 This ‘re-setting’ of the osmostat results in thirst at lower osmolalities with increased water retention.^13,15 However countering the more ready release of vasopressin is the secretion of vasopressinase by the placenta.¹⁵ Transient diabetes insipidus of pregnancy may occasionally occur in pregnancy as the clearance of vasopressin increases fourfold during pregnancy.¹⁵

Primary Vasodilation and Reduced Osmolality

These two important changes in early pregnancy – early vasodilation and early altered plasma osmostat – underpin the early changes in physiology and biochemistry of pregnancy. However it is unknown exactly what factors mediate these critical changes. Heenan et al.¹⁴ postulated these changes were secondary to ‘gestational hormones’ and the role of oestrogens, progesterone, prostaglandins and other endothelium-derived factors has been explored.^13,15 In early work, Danielson et al.²⁰ showed that prostaglandins did not mediate vasodilation in pregnant rats, but that this was mediated by nitric oxide (NO). Deng et al.²¹ showed that NO deficiency in rats disrupted the normal vasodilatory haemodynamic changes of pregnancy. More recently βhCG has been linked to these changes.^13,22,23

Relaxin – Role in Vasodilation and Reduced Osmolality

The mechanisms for these early changes are becoming clearer, at least in rat models, and are thought to be caused by the hormone ‘Relaxin’. Relaxin is an insulin-like growth factor which is released from the ovary in early pregnancy when βhCG stimulates the corpus luteum.²² It accentuates the vasodilatory action of NO and impairs the vasoconstrictor actions of angiotensin II (AGII).²² An important study by Danielson et al.²² showed that Relaxin causes renal vasodilation, increased renal blood flow and increased GFR when given to rats. Conrad et al.²³ extended this work to demonstrate the Relaxin accentuates the endothelial receptor-NO pathway which results in lowered vascular resistance. It had been known that βhCG has a role to play in the reduced osmotic threshold for release of vasopressin²⁴ and the last piece of the puzzle seemed to fall into place when Danielson et al.²² showed that Relaxin (driven by βhCG) also induces a reduction in plasma osmolality. Could Relaxin be the crucial hormone that not only vasodilates renal circulation but also modifies osmoregulation in human pregnancy? This theory is currently under investigation.²³

Secondary Activation of the Renin Angiotensin System

The vasodilation that occurs in early pregnancy also results in activation of the renin-angiotensin-aldosterone axis with increased renin and aldosterone,²⁵ further salt retention, and volume expansion.¹² It is believed that the hyperaldosteronism of pregnancy is secondary, with the renin-aldosterone axis responding normally to stimuli but around a higher set point.¹⁵ Influences on aldosterone in pregnancy are complex with not just renin and prorenin influencing levels. Other factors such as potassium, ACTH, atrial natriuretic peptide (ANP) and dopamine are also postulated to have a role in this setting. (Changes of ANP are controversial in pregnancy with some authors reporting increased levels¹² and others reporting reduced levels.²⁶) Most literature agrees that there is less responsiveness to the vasoconstrictor effects of angiotensin II in pregnancy.^13,26

Potassium

It is likely that a number of complex interactions may influence potassium levels in pregnancy.^15,27–29 Most authors report a decrease in serum and plasma potassium commencing early in pregnancy.^1,2,9 Lockitch reports a decrease of ~95–98% from pre-pregnant levels in early pregnancy although in late gestation a rise in potassium may occur.⁹ Plasma levels of potassium are likely to be lower than serum values. Further partitioning of potassium between trimesters does not appear to be helpful as the changes are very minor within pregnancy. The proposed reference interval of 3.5–4.8 mmol/L (in serum only) appears to be appropriate when reviewed against the data provided in Figure 1B.

Chloride

The changes reported for chloride in pregnancy are minimal. Lockitch reports chloride in pregnancy that is 99–100% of non-pregnant levels.⁹ Larsson similarly notes minimal changes.¹ It was proposed at the AACB Harmonisation Meeting that no change was required for chloride intervals during pregnancy (Figure 1C).

Renal Blood Flow, GFR and Serum Creatinine and Urea

Vasodilation in the renal blood vessels and increased plasma volume result in up to 60% increased renal blood flow to the kidneys¹⁵ and up to 50% increase in GFR.^12,13,15,30 It has been found that the MDRD-based calculation of estimated GFR (eGFR) underestimates true GFR in pregnancy when compared with inulin clearance,³¹ and that eGFR equations based on a serum creatinine are not reliable for assessment of renal function in pre-eclamptic pregnancies – a setting in which accurate assessment of renal function is critical.³²

Creatinine

In normal pregnancy serum creatinine may be well below that of age-matched non-pregnant women.¹⁵ Creatinine levels decrease significantly early on in pregnancy and remain low throughout pregnancy.^1,2,8,9 Levels may be as low as 67% of non-pregnant intervals.⁹ As such a 15–20 µmol/L drop in creatinine may occur in pregnancy. No specific interval for pregnancy was proposed at the recent meeting, however laboratories were encouraged to test their data against intervals around 30–70 µmol/L. The data extract provided here suggests a slightly broader interval of 30–80 µmol/L may be required to encompass all gestations of pregnancy (Figure 1D).

Urea

Urea also undergoes increased clearance in pregnancy and levels are reduced.⁹ Significant decreases in urea to as low as 63% of non-pregnant levels are reported.^1,2,8,9 Low levels persist from 1^st trimester throughout pregnancy and further partitioning of urea intervals within pregnancy is not required. The proposed interval for urea was 1.5–5.5 mmol/L. However with further data review, these intervals may need to be adjusted. Figure 1E suggests intervals of 1.6–6 mmol/L may provide intervals that remain appropriate to encompass the minor changes across gestation.

Bicarbonate, Anion Gap and Acid-Base Status in Pregnancy

In pregnancy, the breathing rate increases and this is thought to be secondary to progesterone having effects on the central respiratory centres^9,33,34 as well as reduced lung capacity from pressure of the gravid uterus on the diaphragm in late pregnancy.¹⁵ This mild hyperventilation results in lowered pCO₂, a mild respiratory alkalosis and compensatory reduction in serum bicarbonate of at least ~2 mmol/L in pregnant compared to non-pregnant women.^9,19 Urinary acidification and distal H⁺ ion secretion remain normal in pregnancy but the urinary bicarbonate threshold may be lowered and it is thought that this might explain why bicarbonate is not fully corrected in pregnancy.³³ Lockitch suggests that bicarbonate drops to ~85% of non-pregnant levels early in pregnancy and remains consistently reduced throughout all trimesters. ⁹ The proposed interval for bicarbonate was 19–29 mmol/L in pregnancy. Review of Figure 1F suggests an upper limit of 28 mmol/L may also be considered.

Some reports suggest the anion gap is reduced in pregnancy by ∼2 mmol/L¹⁹ however others suggest that anion gap increases. In the data reviewed here (Figure 1G) a decrease in anion gap of approximately 4–5 mmol/L at the upper limit and 3–4 mmol/L at the lower limit was noted. Differences in calculation or measurement of anion gap, a general paucity of published data, together with unspecified sample type (serum, whole blood arterial or venous) result in a lack of clear data and mean that it is difficult to propose a single reference interval for pregnancy anion gap.

Proposed Pregnancy Reference Intervals and Gestational Terminology

The following is a summary of the discussion and consensus from the 2015 AACB Harmonisation Meeting. There was agreement that laboratories should use the same terminology to define gestation (age and weeks) and trimesters.

T1:	0–97 days	≥0w–<14 weeks
T2:	98–195 days	≥14w–<28 weeks
T3:	196–293 days	≥28w–<42 weeks

It was proposed that if a laboratory information system did not provide functionality for partitioning in pregnancy a comment should be added outlining intervals for each trimester.

It was agreed that further data gathering and ‘flag rate’ analysis was required to substantiate the proposed reference intervals (Figure 2). It is important to note that these proposed intervals are not formally ratified by the AACB at this time. It was suggested that each laboratory should assess their data against the proposed intervals and provide feedback to the AACB Harmonisation Committee.

Table.

Proposed pregnancy-specific reference intervals (not ratified)

Analyte	Sample type Serum(S) Plasma(P)	Reference interval	Duration of pregnancy (Whole/Part)
Sodium	S/P	132 – 142 mmol/L	Whole
Potassium	S only	3.5 – 4.8 mmol/L	Whole
Bicarbonate	S/P	19 – 29 mmol/L	Whole
Urea	S/P	1.5 – 5.5 mmol/L	Whole

Chloride: No specific interval required. It was suggested that laboratories should continue with the adult chloride reference interval for pregnancy (95–110 mmol/L).

Creatinine: It was recognised that a lower reference interval for pregnancy would be required and that enzymatic methods were preferred in pregnancy. No specific interval was proposed, although laboratories may wish to test their data against intervals around 30–80 µmol/L.