Allopregnanolone in the peripartum: Correlates, concentrations, and challenges – A systematic review

Background: Allopregnanolone (ALLO) is a metabolite of progesterone and a neuroactive steroid hormone. As a positive allosteric modulator of gamma-aminobutyric acid (GABA) receptors, ALLO seems to have antidepressant and anxiolytic effects, and was therefore approved as a specific medication for the treatment of postpartum depression in 2019. Despite the growing number of publications investigating ALLO levels, results on the biological and psychological correlates in the peripartum period remain inconsistent, possibly due to methodological challenges regarding measurement. To date, however, there is no systematic review examining the correlates, concentrations, and challenges in measuring ALLO in peripartum women. Method: A systematic literature search of PubMed and PsycINFO was conducted in August 2023. Original research articles that measured ALLO concentrations in peripartum women were included. Reports were excluded if they were not original research, included non-human subjects, did not include peripartum women, did not include ALLO measurement as an outcome, included (pharmacological) interventions, constituted method validations, or used the same cohort as another study. Results: The literature search yielded 234 articles, and two articles were identified from other sources. After full-text screening, 19 articles (N = 1401) met the inclusion criteria, of which seven focused on biological correlates of ALLO and 12 on mood correlates. Of the latter, six found no association between ALLO and mood, four found a negative association, and two found a positive association. Overall, the results show an increase in ALLO levels during pregnancy and a decrease after birth, with levels then remaining low until six months postpartum. ALLO was most commonly measured in blood plasma and by gas chromatography-mass spectrometry (GC-MS). A significant matrix effect was found for blood serum and a significant method effect for radioimmunoassays (RIAs). A significant effect of time of measurement was found. Conclusion: ALLO measurement shows method and matrix effects. ALLO levels are higher when measured in serum compared to in plasma, and when measured using RIA compared to other methods. Time of measurement, study design, and standardization of measurement also influence the reliability of measurement and the interpretation of results.


Introduction
Allopregnanolone (ALLO) is a neuroactive steroid hormone, and a positive allosteric modulator of the gamma-aminobutyric acid (GABA) receptors in the brain (Paul et al., 2020;Paul and Purdy, 1992).ALLO was first discovered in the adrenal glands in 1938 (Beall and Reichstein, 1938) and was eventually found in the human brain, along with other neurosteroids, in 1986 (Majewska et al., 1986), sparking a major research interest in ALLO and leading to the first experiments with ALLO as a therapeutic agent.Finally, in 2019, an intravenous formulation of the ALLO analogue, brexanolone, was approved by the US Food and Drug Administration as the first specific treatment for postpartum depression (Meltzer-Brody and Kanes, 2020;Scott, 2019).ALLO acts by potentiating the GABA receptor function, and thus has anxiolytic, antidepressant, and stress-buffering effects, which have been found to alleviate symptoms of postpartum depression (Pinna, 2019;Reddy, 2010;Scott, 2019).At the same time, however, ALLO has been associated with irritation, aggression, and sedation (Pinna et al., 2022;Reddy, 2010;Scott, 2019).Its ability to rapidly alter neuronal excitability led to its description as a "neuroactive steroid" (Paul and Purdy, 1992).
As a metabolite of progesterone, one of the main female steroid hormones, ALLO levels increase during pregnancy in the maternal brain and body (Tsutsui and Haraguchi, 2021).During pregnancy, ALLO is synthesized not only by the gonadal and adrenal glands, and by the brain itself, but additionally by the placenta (Frye et al., 2014;Luisi et al., 2000;Paul and Purdy, 1992).The rising ALLO levels have manifold effects: They play an important role in the psychological well-being of the expectant mother and her physiological adaptation to pregnancy, as well as in the development of the fetus (Meltzer-Brody and Kanes, 2020;Schumacher et al., 2020).
Animal studies have demonstrated a downregulation of the hypothalamic-pituitary-adrenal (HPA) axis in late pregnancy, and this finding is frequently translated to humans (Almeida et al., 2021;Brunton and Russell, 2008).The HPA axis plays an important role in regulating the body's response to stress by releasing cortisol.During pregnancy, the rising levels of ALLO act as a natural inhibitor of the HPA axis, which is thought to have stress-buffering effects on both the mother and the fetus, helping to protect them from the potentially harmful effects of chronic stress (Brunton et al., 2014).However, experimental studies in pregnant women have shown that in contrast to rats, the HPA axis is not downregulated in humans (Nierop et al., 2006).So far, only one exploratory pilot study has investigated the relationship between ALLO, cortisol, and anxiety symptoms during a speech and arithmetic stress task in pregnant women in the second trimester, with the results revealing an association between lower ALLO levels and higher anxiety symptoms in response to stress (Crowley et al., 2016).Nevertheless, to enhance our understanding of the intricate interplay between the HPA axis and ALLO in pregnancy, further experimental studies investigating the impact of stress on ALLO levels in pregnant women are needed.
During pregnancy, the fetal brain is already fully capable of synthesizing ALLO by itself (Hirst et al., 2006).Therefore, the fetus also benefits directly from the rising ALLO levels, which have been shown to support fetal brain cell growth, nerve formation, and myelination (Schumacher et al., 2020).ALLO also has neuroprotective effects: During pregnancy complications such as intrauterine growth restriction or preterm birth, ALLO levels rise and effectuate a rapid cell proliferation to reduce brain damage and to protect immature fetal brain cells (Pluchino et al., 2016;Shaw et al., 2019).Furthermore, ALLO may prevent preterm birth by interacting with the oxytocin system (Bealer et al., 2010).In late pregnancy, ALLO inhibits oxytocin by stimulating the opioid system and by regulating oxytocin gene expression in the mother (Turkmen et al., 2021).This suggests that the rising ALLO levels during pregnancy have an adaptive function for fetal development.
After parturition, there is a sharp decline in ALLO levels, which is also thought to play a crucial role in the etiology of peripartum depression and anxiety disorders (McEvoy and Osborne, 2019;Meltzer-Brody and Kanes, 2020).Despite a growing body of literature investigating the course of ALLO in the peripartum and its association with mood, the findings remain inconclusive, and comparable and reproducible results are lacking (Pinna et al., 2022).Nevertheless, ALLO is nowadays used to treat postpartum depression and is being investigated as a therapeutic agent for a range of central nervous system (CNS)-related conditions such as major depression, anxiety disorders, sleep disorders, epilepsy, Parkinson's disease and Alzheimer's disease (Cerne et al., 2022).For the effective treatment of these disorders, a precise and reliable measurement of ALLO is crucial.Investigating the factors that influence ALLO measurement in the peripartum period may facilitate the establishment of ALLO as a predictive biomarker for the effective treatment of peripartum mood disorders (Pinna, 2020).
There are several challenges to the reliable measurement of neurosteroids such as ALLO, which may contribute to the problem of reduced comparability and reproducibility of results and must be overcome in order to fully understand their psychobiological effects.These challenges include different sample matrices such as blood plasma, serum, or saliva.A further difficulty lies in the different measurement methods used to determine ALLO concentrations, including different immunoassays (e.g.radioimmunoassays (RIAs) or enzyme-linked immunoassays (ELISAs)) and liquid or gas chromatography-mass spectrometry (LC-MS or GC-MS).It is known that neurosteroid concentrations can vary depending on the method of measurement and the matrices used (Farré et al., 2007), but these differences have not been systematically reviewed with regard to ALLO.
In the following, as a first step, we provide an overview of published peripartum ALLO concentrations.Furthermore, we summarize the relationships between the biological and psychological factors described so far, and explain the methodological challenges in measuring ALLO in the peripartum period.

Methods
We searched for original studies that reported ALLO concentrations in any matrix in peripartum women.This systematic review was registered in the International Prospective Register of Systematic Reviews (PROSPERO; registration no.CRD42023448885).The systematic literature search and review were performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA; Moher et al., 2009) guidelines.

Search strategy
The search strategy comprised a systematic literature search of the databases PubMed and PsycINFO conducted in August 2023.The following search terms were used: allopregnanolone, pregnancy, postpartum, and peripartum in title and abstract.The literature search was conducted by two independent reviewers.

Study selection
Articles were included if they were original research that measured ALLO concentrations in peripartum women at least once during pregnancy or postpartum.Reports were excluded if they were not original research, included non-human subjects, did not include peripartum women, did not include an ALLO measurement as an outcome, involved (pharmacological) interventions, were method validations, or used the same cohort as another study.The selection procedure can be seen in Fig. 1.

Data extraction
Each included article was screened for the relevant information such as author and year of publication, population and sample size, operationalization of outcome, time points in the peripartum, ALLO concentration, matrix and measurement methods, and a summary of the relevant findings.ALLO concentrations were converted into ng/mL for the purpose of comparison.The extracted information is reported in Table 1 and Table 2. Two reviewers assessed the quality of all included studies on six different dimensions (as recommended by Eichenauer andEhlert, 2023, Gao et al., 2023), with quality ratings lying between 44% and 93% (mean = 68%).The quality assessment is presented in Supplement 1.

Data analysis
Statistical analyses were conducted with the extracted ALLO concentrations in order to investigate the influence of measurement method and matrix.For this purpose, a two-way analysis of variance (ANOVA) was conducted, with the factors 'Method' and 'Matrix'.The factor 'Method' has three levels: 'GC-MS', 'LC-MS', and 'RIA'.'ELISA' was excluded because only one study used ELISA as a measurement method.The factor 'Matrix' has two levels: 'Serum' and 'Plasma'.'Saliva' was excluded because only one study included saliva as a matrix.Dunnett's T3 post-hoc tests were used to analyze whether the levels differed significantly from each other (Dunnett, 1980).This method is especially useful for unequal sample sizes with unequal variances.
Additionally, we analyzed whether the measurement method and matrix, or the time point, influenced the main outcome of the studies (as summarized in Table 1).Accordingly, we performed ANOVAs with either the measurement method and matrix together as factors, or time points or trimesters, with main outcome as the dependent variable.The time points were taken from the respective studies: once as factor variables (because they usually included ranges) and once rescaled as numerical variables, where each measurement time point was assigned a value relative to the others.The trimesters where calculated from the measurement time points as a numerical variable (1 = first trimester, 2 = second trimester, 3 = third trimester, 4 = postpartum).As criteria, the following gestational weeks were considered: first trimester = 1-12 weeks, second trimester = 13-24 weeks, third trimester = 25-40+ weeks.The main outcome was a categorical variable with three levels (negative association, positive association, no association).All significance levels were set at α = 0.05.Statistical analyses were performed using RStudio version 2023.06.0-421.

Search results
The database search yielded 234 articles on PubMed and 95 on PsycINFO.Two additional titles meeting the inclusion criteria were identified through hand search.First, 86 duplicates were removed.The remaining 245 articles were screened by title and abstract if they were original articles with human subjects, leading to the exclusion of 188 articles.The eligibility criteria were applied to the full texts of the remaining 57 articles and a further 38 articles were excluded.Finally, a total of n = 19 studies encompassing N = 1'401 participants were included in the review.Table 1 shows the characteristics of the study populations sorted by main outcome of the studies.Table 2 summarizes the measurement time points, measurement method, matrix, and measured concentration of the included studies, which are also depicted in Fig. 2.

Concentrations of ALLO in the peripartum
Fig. 2 provides an overview of the mean reported ALLO concentrations in the peripartum, starting from gestational week (GW) 4 up to six months after birth.Half of the measured concentrations were assessed in the third trimester, less than one third in the second trimester, and one fifth in the postpartum period, whereas only 4% were assessed in the first trimester.As can be seen, ALLO increases over the course of pregnancy, with a mean concentration of 6.25 ng/mL in the first trimester, 10.01 ng/mL in the second, and 14.80 ng/mL in the third, averaged across all included studies.Mean reported concentrations in the postpartum are much lower, at 0.94 ng/mL, and remain low for six months   after parturition.In Fig. 2 (A), an overall trendline over the measured time points is depicted to illustrate the overall course of ALLO in the peripartum.
The absolute concentrations as well as the relevant extracted methodological information are presented in Table 2.A total of eleven studies used blood plasma as the sample fluid, whereas seven used blood serum and only one study analyzed ALLO in saliva.GC-MS was the most frequently used method of analysis, in eight studies; RIA was used in six studies, LC-MS in four studies, and ELISA in only one study.The most frequent method combination was plasma and GC-MS, which was used in six studies, while the combination of serum and RIA was used in four studies.The combinations of serum and GC-MS, plasma and LC-MS, and plasma and RIA were each used twice, and the combinations of plasma and ELISA, saliva and LC-MS, and serum and LC-MS were each used once.Accordingly, there were eight different matrix-method combinations in total.

Challenges of ALLO measurement in the peripartum
Furthermore, we statistically tested whether the measurement method and the matrix have an influence on the measured ALLO concentration.For this purpose, a two-way analysis of variance (ANOVA) was conducted, which revealed a medium-sized significant effect of the factor 'Method' (F(2, 72) = 4.11, p = 0.02, η 2 = 0.09) and of the factor 'Matrix' (F(1, 72) = 10.92,p = 0.001, η 2 = 0.11).However, the interaction between 'Method' and 'Matrix' (F(2, 72) = 1.94, p = 0.15, η 2 = 0.04) did not show a significant effect.The Dunnett's T3 post hoc test revealed statistically significant differences in mean concentrations between 'Serum' and 'Plasma' (p adj = 0.024), with higher concentrations in the former than the latter, as depicted by the yellow and red trendlines in Fig. 2 (B).Based on the Dunnett's T3 post hoc test, the mean concentration of the measurement method factor 'RIA' was significantly higher than that of the factors 'LC-MS' (p = 0.023), but only marginally not as the factor 'GC-MS' (p = 0.051).The 'Matrix' factors 'LC-MS' and 'GC-MS' did not differ from each other (p = 0.67).

Discussion
The aim of this review was to provide an overview of published peripartum ALLO concentrations and to summarize the biological and psychological correlates described to date.The results show the expected increase in ALLO over the course of pregnancy as well as the expected drop immediately after birth.With eight different matrixmethod combinations, which refers to a specific pairing of biological sample type (e.g.blood serum) and method to analyze the sample (e.g.GC-MS), a wide variety of analytical methods was used in the 19 included studies.So far, ALLO research has mainly focused on blood as a matrix, and only one study has examined ALLO in saliva.Approximately one third of the studies investigated biological associations of ALLO in the peripartum, while two thirds focused on mood outcomes.Of these latter studies, six found no association between ALLO and mood, four found a negative association, and only two studies reported a positive association.In the following, we will take a closer look at ALLO concentrations during the peripartum in relation to the influence of the matrix and measurement method, in order to highlight the methodological challenges of ALLO measurement.In a next step, the quality assessment of the included studies will be used to further explore methodological challenges in depth.This may help to better classify the conflicting findings regarding ALLO correlates and to highlight specific current challenges to the reliable measurement of ALLO.
The measured ALLO levels in the studies show an overall increase during pregnancy and a drop after birth, with levels remaining low up to six months postpartum.As a metabolite of progesterone, the changes in ALLO levels can be primarily attributed to the rising progesterone levels during pregnancy (Dukic and Ehlert, 2023;Schumacher et al., 2020).Fig. 2 (A) visually summarizes the assessed concentrations from all included studies and provides a comprehensive overview of the course of ALLO as a general trendline in the peripartum between four weeks of gestation and six months postpartum.The trendline reveals a steeper incline in the third trimester, indicating a higher increase in ALLO levels at the end of pregnancy.This may be due to a higher metabolic rate of enzymes involved in the metabolism of ALLO, and has already been suggested by some authors, who also found that the metabolism of ALLO influences the timing of parturition (Hill et al., 2007;Parízek et al.,    2005).This indicates that ALLO expectant mother to adapt physiologically to the pregnancy and prepare for the birth.As data for the postpartum still largely lacking, it is not yet known when ALLO levels fully return baseline.
The present findings support a recent review by our work group, which suggested that peripartum sex steroids vary depending on methodological factors such as matrix and measurement method (Dukic and Ehlert, 2023).Specifically, in the present study, we found that mean ALLO concentrations in serum were significantly higher than in plasma, as illustrated by the trendlines in Fig. 2 (B), and a comparable matrix effect was reported for progesterone and estrogen in the aforementioned review (Dukic and Ehlert, 2023).This is attributable to the fact that blood serum is more sensitive than plasma, meaning that the hormone levels measured are generally higher (Yu et al., 2011).While we could not statistically analyze the difference between blood samples and saliva, given that there was only one study measuring ALLO in saliva, it can be visually seen that levels are much lower in saliva.This is due to the fact that steroid hormones are present in blood at much higher concentrations due to hormone binding proteins (Granger et al., 2004).This should be re-analyzed when more studies with salivary ALLO exist.Although we found an overall medium-sized effect of method, the post hoc test only revealed higher mean concentrations in studies measured using RIA compared to GC-MS.This might be due to cross-reactivity with other steroids or metabolites in the sample matrix, as says generally yield higher levels due to a lower specificity and sensitivity (Kaleta et al., 2021;Karashima and Osaka, 2022;Stanczyk et al., 2007).The significant effect might also be explained by the visible outliers of one study (Luisi et al., 2000), as shown in Fig. (B).Upon closer inspection, it is apparent that the description of hormone measurement in the study is unclear, and it might be possible that ALLO was measured outside the specified measurement range.This may have contributed to the increased levels compared to the other studies and to the overall significant effect of RIA compared to the other methods.Moreover, RIA shows the highest variance compared to the other methods, suggesting inferior sensitivity in the measurement of ALLO.Even though ELISAs also generally show higher levels in the measurement of steroid hormones (Kushnir et al., 2009), this was not found descriptively for ALLO.Given that only one study measured ALLO with ELISA, no analysis was conducted comparing ELISA with other methods.Generally, ELISAs are easy to perform, as they require less expensive equipment compared to mass spectrometry methods.Especially together with saliva, they represent a promising alternative for the measurement of steroid hormones (Gröschl, 2008;Gröschl et al., 2001).However, it is too early to draw any conclusions about the sensitivity of ELISA in the measurement of ALLO.As chromatography methods followed by mass spectrometry, both GC-MS and LC-MS show similar mean levels.However, GC-MS has a higher variance, with twice as many studies included.Comparison studies show that GC-MS and LC-MS have similar sensitivity in the measurement of steroid hormones (Marcos and Pozo, 2015).
To date, there are few data on sex or age differences with regard to ALLO.Levels are generally higher at fertile age, decrease with age in men but not in women (Fadalti et al., 1999;Genazzani et al., 1998), and are highest during pregnancy (Luisi et al., 2000).It is also known that ALLO levels vary over the course of the menstrual cycle (Genazzani et al., 1998), and one study showed that ALLO is subject to a diurnal rhythm as well, with a peak around midday (Möller and Bäckström, 2016).Several method validations have investigated the stability of ALLO in blood and saliva samples over repeated freeze-thaw cycles and with a short storage time (Cai et al., 2019;Dury et al., 2015;Gaffey and Wirth, 2014;Son et al., 2020).However, more evidence is needed to establish general guidelines for the storage and handling of ALLO to enable a standardized and reliable measurement.Such guidelines could help to minimize the influence of measurement factors on the results, which is crucial if ALLO is to be established as a diagnostic marker for peripartum mood disorders.
Having demonstrated the method and matrix effects of ALLO measurement, we now use the quality assessment of the included studies to highlight specific methodological challenges in more detail, which should further serve to advance the reliability of ALLO measurement.One such challenge, which has proven to be a significant predictor of ALLO outcomes, is the timing of the ALLO measurement in the peripartum, as hormone-related research is highly sensitive to changes in research design, and measurement time points need to be carefully planned based on relevant secretion patterns (Gildner, 2021).As can be seen in Fig. 2 and Table 2, ALLO concentrations can change drastically in only a few weeks.It is therefore also important that measurement intervals only encompass a short and specified time period (e.g., a specific week of gestation or week postpartum) to ensure that measured concentrations are more specific to the peripartum time point and comparable between studies.Additionally, the use of longitudinal designs and appropriate statistical methods may provide a more robust approach to measuring ALLO levels in relation to health or mood outcomes (Gildner, 2021;Granger et al., 2004).All reviewed studies measured ALLO levels and mood simultaneously, and often with only a small number of assessment time points.However, the change in ALLO and the development of symptoms do not necessarily have to occur concurrently.This was shown by Eisenlohr-Moul et al. (2020) in their study of patients with premenstrual dysphoric disorder.More longitudinal studies with frequent time points are therefore needed, which will also help to identify possible delayed mood onset.Based on the recommendations of Schmalenberger et al. (2021) in relation to menstrual cycle research, between-person differences in sensitivity need to be taken into account, while weeks of gestation should be included as a within-person time variable centered around birth.Between-person sensitivity was not fully been considered in the reviewed studies.Some studies (Deligiannidis et al., 2019;Epperson et al., 2006;Wenzel et al., 2021) did look at differences in ALLO levels between healthy women and women with PPD, with the hypothesis that they differ in their sensitivity to changes in hormone levels.However, there has to be a stronger focus on possibly different within-person effects of ALLO on mood, which has already been shown for estradiol (Gordon et al., 2021).Taken together, more longitudinal studies, with frequent measurement time points and a statistical focus on between-person sensitivity to mood, are necessary to better understand the relationship between ALLO and mood.A second challenge, which can be derived from the method and matrix effect of ALLO, is the standardization of hormone analysis, which needs to be reported together with the appropriate details such as sensitivity, specificity, or range (Andreasson et al., 2015;Casals et al., 2023;Kaleta et al., 2021;Karashima and Osaka, 2022) to allow for a clear interpretation and the possibility to replicate results.The use and development of standardized hormone measurement protocols and assays as well as the adherence to internationally recognized standards will help to improve the measurement of ALLO concentrations.A reliable measurement across methods and matrices will facilitate the generation of reference ranges in the peripartum and encourage cross-study comparisons.It should not be neglected that the interaction between method and matrix had an unspecific effect on the results regarding mood outcomes in the reviewed studies.This, together with the overlooked between-person sensitivity and the paucity of longitudinal studies, might explain the inconclusive findings to date.Overall, the significant interaction effect of matrix and method on the study outcome emphasizes the need for matrix-specific standardized methods for the measurement of ALLO.Only this will advance ALLO as a diagnostic marker of peripartum depression.The full quality assessment, including the criteria described above and the additional criteria, can be found in Supplement 1.Taken together, the results suggest a method and matrix effect in ALLO measurement in the summarized data.As ALLO levels vary depending on the measurement method and matrix, we recommend careful consideration of measurement method and matrix in future research.Currently, we recommend that priority should be given to sensitive methods such as LC-MS and GC-MS, while ELISAs should be further explored and validated.in combination with serum, cannot be recommended based on the current state of knowledge, as ALLO levels are generally overestimated.Serum and plasma differ moderately, but is currently possible to derive any recommendation with regard to serum and plasma.To ensure reproducibility, validation and standardization of protocols, as well as reporting guidelines are for the specific matrix.Further studies should also explore ALLO levels in saliva with regard to specific matrix effects.As the design factors have just as much impact on the outcome as the methodological factors, a greater focus needs to be placed on design factors in future studies.Moreover, longitudinal designs with frequent measurements are needed to better analyze the relationship between ALLO and mood outcomes, with a stronger focus on between-person sensitivity.
Our systematic review examined peripartum ALLO concentrations and their correlation with biological and psychological outcomes, considering methodological challenges.It highlights the impact that the matrix and method of analysis can have on measured ALLO concentrations and study outcomes.ALLO levels measured in serum are higher than in plasma, and levels measured with RIA are higher than those measured with other methods.As more studies describing salivary ALLO concentrations become available, matrix effects should be re-evaluated to include possible effects of saliva.Factors such as measurement time points, study design, and standardization of methods also play a role in reliable measurement.Ultimately, the aim of this systematic review is to enhance the understanding of ALLO as a biomarker for peripartum depression, by improving the reliability of measurement of ALLO in the peripartum.

Fig. 1 .
Fig. 1.Flow diagram of study selection according to PRISMA guidelines.

Fig. 2 .
Fig. 2. ALLO concentrations in the peripartum in relation to (A) measurement method and (B) matrix.Note.Allopregnanolone concentrations from gestational week four up to six months postpartum.Dots represent allopregnanolone concentrations measured in the included studies, ranging from 0.05 to 57 ng/mL.One study can be represented with several dots.Figure (A) is sorted by measurement method and Figure (B) by matrix, while the same studies and measurement time points are represented.In Figure (A), the linear trend lines visualise the difference in mean concentration by method between ELISA, GC-MS, LC-MS, and RIA.An overall trendline over the course of allopregnanolone in the peripartum is indicated with a locally weighted scatterplot smoother function.In Figure (B) the linear trendlines visualise the difference in mean concentration by matrix between Plasma, Saliva, and Serum.Significance levels = * < 0.05, ** < 0.01, ***< 0.001.

Table 1
Summary of characteristics and main results of the reviewed studies.

Table 2
ALLO concentrations and relevant methodological factors of the reviewed studies.