Prostaglandins in biofluids in pregnancy and labour: A systematic review

Prostaglandins are thought to be important mediators in the initiation of human labour, however the evidence supporting this is not entirely clear. Determining how, and which, prostaglandins change during pregnancy and labour may provide insight into mechanisms governing labour initiation and the potential to predict timing of labour onset. The current study systematically searched the existing scientific literature to determine how biofluid levels of prostaglandins change throughout pregnancy before and during labour, and whether prostaglandins and/or their metabolites may be useful for prediction of labour. The databases EMBASE and MEDLINE were searched for English-language articles on prostaglandins measured in plasma, serum, amniotic fluid, or urine during pregnancy and/or spontaneous labour. Studies were assessed for quality and risk of bias and a qualitative summary of included studies was generated. Our review identified 83 studies published between 1968–2021 that met the inclusion criteria. As measured in amniotic fluid, levels of PGE2, along with PGF2α and its metabolite 13,14-dihydro-15-keto-PGF2α were reported higher in labour compared to non-labour. In blood, only 13,14-dihydro-15-keto-PGF2α was reported higher in labour. Additionally, PGF2α, PGF1α, and PGE2 were reported to increase in amniotic fluid as pregnancy progressed, though this pattern was not consistent in plasma. Overall, the evidence supporting changes in prostaglandin levels in these biofluids remains unclear. An important limitation is the lack of data on the complexity of the prostaglandin pathway outside of the PGE and PGF families. Future studies using new methodologies capable of co-assessing multiple prostaglandins and metabolites, in large, well-defined populations, will help provide more insight as to the identification of exactly which prostaglandins and/or metabolites consistently change with labour. Revisiting and revising our understanding of the prostaglandins may provide better targets for clinical monitoring of pregnancies. This study was supported by the Canadian Institutes of Health Research.


Introduction
It is widely believed that prostaglandins are important in the initiation of human labour [1]. Multiple studies have documented increased expression of cyclooxygenases, key enzymes in

Inclusion/Exclusion criteria
Primary study journal articles examining endogenous prostaglandins in blood, amniotic fluid, and/or urine during pregnancy and spontaneous labour were included in this review. Studies were excluded if the study was on animals, the study was examining exogenous prostaglandins for induction of labour or if participants experienced spontaneous abortion (prior to 20 weeks). As well, studies which only had samples collected following delivery were excluded. Publications with incomplete information (i.e., conference abstracts) were excluded. Only studies written in English or with an available English translation were included. The search did not include a time restriction, however, the databases MEDLINE and EMBASE include literature published since 1946 and 1947 respectively. The search was initially conducted on May 19, 2020 and was repeated on August 20, 2021.

Selection process/Data extraction
The following information was extracted by one reviewer (EW) from each of the final selected studies: population examined, sample number, type of biofluid collected, method of testing and measurement, metabolites/prostaglandins measured, time of sample collection, country of study origin, available measures of central tendency and variance, and major findings of the study.

Quality assessment
Studies were assessed for quality and risk of bias using a quality assessment tool (Table 1) adapted from Hadley et al. [9] for assessment of basic science research. Full details of the rubric can be found in Table 1. Studies were scored between 0-9. All studies were scored independently by two investigators (EW and KH) and disagreements in scores were resolved by discussion.

Data synthesis
A qualitative summary was generated, and tables were created with the main results from each study. No pooled analysis was performed.

Studies identified
The electronic search returned 2257 unique records after removal of duplicates from 2688 records. 2101 records were removed at the title/abstract level, leaving 156 records for assessment at the full-text level. Hand search of reference lists yielded an additional 35 records for review, resulting in a total of 191 full text records. Of the records assessed at the full text level, 108 were excluded, leaving n = 83 studies for inclusion in this review (Fig 1).

Main characteristics of studies
Summaries of the main characteristics and relevant findings of the included studies can be found in Table 2 (presented in chronological order). Of the 83 studies, most assessed only one

Amniotic fluid
Labour and non-labour. In total, 25 studies compared amniotic fluid prostaglandins in labour vs non-labour. PGF 2α increased in labouring participants compared to non-labouring participants in most studies, (   Predicting preterm labour. Six amniotic fluid studies investigated prostaglandins as predictors of preterm labour. Some studies suggest that PGF 2α may be predictive of preterm delivery in those with threatened preterm labour [88] and PPROM [86] however, results are mixed [71,73]. PGFM and bicyclo-PGEM were found in higher levels in participants with preterm labour leading to preterm delivery compared to those who eventually delivered at term [71]. Increased PGE 2 levels may be predictive of delivery before term [68,73] and before 34 weeks [76].
Predicting preterm labour. Two studies investigated prostaglandins in plasma as predictors of preterm labour. One study found that PGFM levels were higher in participants in Published data presented as ng/ml, nanomolars, or picomolars were converted to pg/ml and data presented as ng/g creatinine or ng/mmol creatinine were converted to pg/g creatinine. Data published in μg/hr were not converted and are presented as in the original article.

PLOS ONE
preterm labour who delivered preterm compared to those who went on to deliver at term [61] though the other study reported no significant difference [44].

Urine
Labour and non-labour. Five studies compared labouring vs non-labouring groups. The metabolite t-PGFM was reported to increase with labour compared to non-labour [20,31,41].
Prior to labour onset. Four studies measured changes in prostaglandins in urine throughout pregnancy. PGF 2α , PGFM, and t-PGFM were reported to increase around term, though this was only reported in one study for each prostaglandin/metabolite [20,83]. The metabolite 2,3-dinor-6-keto-PGF 1α did not change throughout pregnancy [81].
Predicting preterm labour. Two urine studies investigated prostaglandins as predictors of preterm labour. One found that PGF 2α levels in urine samples collected at median 32.1 weeks were not significantly different between participants who delivered at term and those that delivered preterm [89]. In contrast, averaged levels of PGF 2α in urine were also found to be associated with increased odds of preterm birth (OR = 1.98) [90].

Serial prostaglandin measurement during spontaneous labour
Although not a primary study question of this review, we noted that n = 40 studies obtained serial samples during labour. In general, prostaglandins measured in amniotic fluid increased throughout labour. Results were mixed for those that measured plasma and serum.

Quality assessment
Scores from the quality assessment were distributed as follows: 17% scored between 0-3, 41% scored between 4-6, and 42% scored between 7-9. The areas with the lowest scores were researcher blinding and sufficiency of sample number for internal validity. Scores for each study can be found in S3 Table.

Discussion
We demonstrate, through a systematic review of the literature investigating prostaglandins and metabolites in peripheral biofluids during pregnancy and labour, that prostaglandins of the PGE and PGF families do exhibit changes through pregnancy and labour, though results are inconsistent and inconclusive. Changes in PGE 2 , PGF 2α , and PGFM levels with labour are most prominent in amniotic fluid, and to a lesser extent in blood. Similarly, our synthesis suggests that PGE 2 , PGF 2α and PGF 1α increase in amniotic fluid as pregnancy progresses and peak around term, though in plasma, a consistent pattern is unclear. Patterns in urine prostaglandin levels were inconclusive due to a relatively small number of studies investigating this biofluid. An important limitation is a general lack of data on prostaglandins and metabolites outside the PGE and PGF families, and as such we are unable to comment on their potential role in pregnancy and labour. Further, few studies examined prostaglandins as biomarkers for preterm labour and more research is needed to provide conclusive evidence for which prostaglandins or metabolites examined could offer the best options for prediction.

Measurement techniques for prostaglandins
Inconsistent study designs and methods greatly limited our ability to compare findings across studies. Up to the late 1990's, researchers most commonly used radioimmunoassay techniques, which can be highly sensitive, but are often limited by the specificity of the antibody used and the potential of antibody cross-reactivity with similar molecules [87]. One study included in this review developed and reported on a radioimmunoassay for PGF 2α with a cross-reactivity with PGF 1α of 12.2% [17], which may have obscured patterns in PGF 2α and made it difficult to ascertain fine-tuning of the prostaglandin pathway among similar molecules. Furthermore, multiple other studies using radioimmunoassay techniques were unable to differentiate between PGF 2α and PGF 1α , and PGE 2 and PGE 1 and therefore could only report on levels of PGF and PGE, respectively, making it difficult to compare the results of these studies with others. Lack of specificity and accuracy in these radioimmunoassay techniques may have contributed to the discrepancies across results and highlights the importance of re-visiting dogma in light of novel evidence and technologies. In contrast, the high specificity and sensitivity of mass spectrometry for lipid identification suggests that this method may be more suitable and accurate for measurement of prostaglandins [87]. Additionally, the capability of mass spectrometry to co-assess multiple prostaglandins and metabolites can provide a quantitative profile of prostaglandins before and during labour, as well as identify prostaglandins and/or metabolites not previously measured that may play a role in pregnancy and/or labour [93].

Considerations among unique biofluids
Among the studies included in this review, the most assayed biofluid was maternal plasma. Although an appealing fluid due to its ability to be sampled relatively easily, results from measurements in plasma were often conflicting, especially among studies that measured primary prostaglandins. Accurate measurement of changes in primary prostaglandin levels in blood is complicated by their rapid metabolism and correspondingly short half-life [94,95]. This difficulty is further compounded by the production of prostaglandins by platelets that occurs during isolation of plasma and storage of samples [36,96,97]. Measurement of plasma PGFM appears to be a good alternative for PGF 2α , as there is no evidence that this metabolite is formed during sample collection or isolation and therefore may more accurately reflect endogenous prostaglandin production [19]. The primary metabolite of PGE 2 , however, is chemically unstable [98], which necessitates the measurement of its degradation product, bicyclo-PGEM, for an accurate index of PGE 2 production [99]. Therefore, results from early studies measuring primary prostaglandins and/or PGEM in plasma and/or serum should be interpreted with these considerations in mind and future studies in blood should aim to measure PGFM or bicyclo-PGEM as indices of PGF 2α or PGE 2 production, respectively.
Amniotic fluid lacks prostaglandin metabolizing enzymes [100, 101], which suggests that measurement of the primary prostaglandins in this fluid may be more accurate than in serum or plasma. However, sampling amniotic fluid is more difficult and may introduce infections harmful to the developing fetus, making this fluid impractical as a predictive resource. Additionally, prostaglandin levels vary based on method of collection and region of the amniotic sac [102, 103] which complicates any interpretation of results from studies and limits the clinical utility of an amniotic fluid test for prediction of preterm labour.
Measurement of the main urinary metabolite of PGF 2α may be preferable to measuring PGFM in plasma or serum in some cases, as a significant portion of circulating PGF 2α is eventually excreted into the urine [104]. In the present investigation, we identified only nine studies that assayed urine, and we suggest that the presence of urinary metabolites of prostaglandins during pregnancy and labour merits further study.

Demographic and clinical information
Among the articles included in this review, we noted that very few provided complete demographic and clinical information on their participants. Factors including age, race/ethnicity, membrane status, and gravidity/parity may impact prostaglandins levels and a lack of consideration for these variables may obscure patterns of prostaglandin levels throughout pregnancy and labour. Complete descriptions of gestational age groups and clearly defined outcomes for both term and preterm labour would additionally make studies more easily comparable. As well, preterm labour is generally defined as labour occurring before 37 weeks gestation, however the pathophysiological processes involved in extreme preterm birth (<28 weeks) may vary dramatically from those near term [105]. Therefore, stratification of outcome groups based on gestational age at delivery may be more informative, though would require larger sample sizes to maintain statistical power.

Role for other prostaglandins
While prostaglandins of the E and F series are most clinically targeted for labour management, there is evidence to suggest that other members of the prostaglandin family may play a role in pregnancy and labour. For example, PGD 2 has been shown to increase uterine contractility and blood flow in various mammals [106-108] and is associated with cervical dilation in humans [63]. Two metabolites of PGD 2 , 9α,11β-PGF 2 and PGJ 2 , were each identified only once among the articles included in this review and were both reported to increase with term labour [84,87]. These metabolites may be of interest to future researchers, as the development of new methodologies such as mass spectrometry have allowed for more accurate and sensitive measurements of select members of the prostaglandin pathway.

Limitations
The main limitation of this review is that only studies in English or with an available English translation were included, which may have resulted in missing some relevant articles. However, current resources limited the feasibility of including non-English studies.

Conclusion
We have identified evidence to suggest that prostaglandin levels, particularly within the PGE and PGF families, do increase in some biofluids during pregnancy and labour. However, changing prostaglandin levels throughout pregnancy and labour are likely highly complex and warrant further investigation, including serial measurements with more precise methodologies in higher-powered studies. Two important limitations identified in this review are the lack of data on the complexity of the prostaglandin pathway outside of the PGE and PGF families and the inherent difficulty in measuring primary prostaglandins in blood, due to their short halflives in this biofluid. With the advent of i) new methodologies that can assess multiple prostaglandins and metabolites together, ii) a more developed understanding of the range of prostaglandins and iii) a better understanding of the heterogeneous nature of term and preterm labour, future studies that take each of these parameters into account in their study design will help provide further insight into the changing levels of prostaglandins in pregnancy and labour.
Supporting information S1