The Effect of Pregnancy and Inflammatory Bowel Disease on the Pharmacokinetics of Drugs Related to Inflammatory Bowel Disease—A Systematic Literature Review

Due to ethical and practical reasons, a knowledge gap exists on the pharmacokinetics (PK) of inflammatory bowel disease (IBD)-related drugs in pregnant women with IBD. Before evidence-based dosing can be proposed, insight into the PK has to be gained to optimize drug therapy for both mother and fetus. This systematic review aimed to describe the effect of pregnancy and IBD on the PK of drugs used for IBD. One aminosalicylate study, two thiopurine studies and twelve studies with biologicals were included. Most drugs within these groups presented data over multiple moments before, during and after pregnancy, except for mesalazine, ustekinumab and golimumab. The studies for mesalazine, ustekinumab and golimumab did not provide enough data to demonstrate an effect of pregnancy on concentration and PK parameters. Therefore, no evidence-based dosing advice was given. The 6-thioguanine nucleotide levels decreased during pregnancy to 61% compared to pre-pregnancy levels. The potentially toxic metabolite 6-methylmercaptopurine (6-MMP) increased to maximal 209% of the pre-pregnancy levels. Although the PK of the thiopurines changed throughout pregnancy, no evidence-based dosing advice was provided. One study suggested that caution should be exercised when the thiopurine dose is adjusted, due to shunting 6-MMP levels. For the biologicals, infliximab levels increased, adalimumab stayed relatively stable and vedolizumab levels tended to decrease during pregnancy. Although the PK of the biologicals changed throughout pregnancy, no evidence-based dosing advice for biologicals was provided. Other drugs retrieved from the literature search were mesalazine, ustekinumab and golimumab. We conclude that limited studies have been performed on PK parameters during pregnancy for drugs used in IBD. Therefore, more extensive research to determine the values of PK parameters is warranted. After gathering the PK data, evidence-based dosing regimens can be developed.


Introduction
Inflammatory bowel disease (IBD) is an overarching term for chronic inflammation in the gastrointestinal tract [1]. IBD is characterized by exacerbations. Medications play a main role in maintaining the remission of IBD. Considering the main drug classes used as therapy, aminosalicylates, thiopurines, corticosteroids, immunosuppressants, biologicals and JAKinhibitors play a dominant role [2]. The two most common variants of IBD are ulcerative colitis (UC) and Crohn's disease (CD). A Dutch population-based cohort study found

Study Selection
First the title and abstract were screened for the three main topics (Table 1). When a study included all three topics, the full article was studied. In a later phase, a distinction between IBD patients and non-IBD patients was made. Studies including non-IBD participants were excluded. Studies not meeting the study aim and inclusion criteria were excluded. Two investigators (TW and PM), separately from each other, conducted the search strategy and the study selection. The obtained results were discussed, and in the case of disagreement, a third author (DT) was consulted.

Data Extraction
When the studies were included in this study, the data were extracted in a Microsoft Excel sheet. The data extraction was performed separately by two investigators (TW and PM) for all included studies. In the case of disagreement, a third author (DT) was consulted. The study characteristics of interest that were extracted were the study design, number of pregnant women included in the study, type of medication (with dosage and dosing interval), moment in time when participants were studied (before pregnancy (T0), trimester 1 (T1), 2 (T2), 3 (T3), at delivery (T4) and/or postpartum (T5)), age and bodyweight at inclusion, type of IBD and the analytical method used for drug concentration measurements. The timeframes for the trimesters were defined as 0 to 13 weeks for T1, 14 to 26 weeks for T2 and 27 to 40 weeks for T3. In addition, the PK parameters per study were extracted. Furthermore, it was investigated if, based on potentially changed PK during pregnancy, adapted dosages were advised by the studies. When the numerical values for the PK parameters were not available within the study, but a graph was available, the data were extracted by using R version 4.1.2 and R Foundation for Statistical Computing, Vienna, Austria, with the use of the package Digitize version 0.0.4.

Study Selection
A total of 430 studies were identified. After the removal of duplicates (n = 36) and removal of 341 studies based onnot meeting the criteria, full texts were obtained from 53 studies, of which 37 were excluded. The reasons for exclusion are provided in Figure 1. These include, among other reasons, ex vivo data, non-applicable outcomes for this study, data related only to the fetus or infant, or letters to the editor as a reaction on publications. Consequently, 15 studies were included. One study covered aminosalicylates, two studies thiopurine therapy and 12 studies biologicals. The PRISMA flow diagram is presented in Figure 1.

Aminosalicylates
One aminosalicylate study was found [9]. In this study, the outcomes of five participants were found to be suitable for this review. One woman used a suppository, three women used a tablet and one woman used a combination of both drugs. The age and weight of these women were not provided. The drug concentration was measured at delivery, with a timeframe from dosing to delivery ranging from 5 to 24 h. The lowest concentration was found in a patient using only the mesalazine tablet, with a concentration of 0.2 µmol/L. The highest concentration was found in another patient using only the tablet, with a concentration of 2.6 µmol/L.
In conclusion, based on the limited existing data, no conclusion can be drawn on possible changes in the PK of aminosalicylates throughout pregnancy. Furthermore, based on the limited available data, no evidence-based dosing regimen could be provided.

Aminosalicylates
One aminosalicylate study was found [9]. In this study, the outcomes of five participants were found to be suitable for this review. One woman used a suppository, three women used a tablet and one woman used a combination of both drugs. The age and weight of these women were not provided. The drug concentration was measured at delivery, with a timeframe from dosing to delivery ranging from 5 to 24 h. The lowest concentration was found in a patient using only the mesalazine tablet, with a concentration of 0.2 μmol/L. The highest concentration was found in another patient using only the tablet, with a concentration of 2.6 μmol/L.
In conclusion, based on the limited existing data, no conclusion can be drawn on possible changes in the PK of aminosalicylates throughout pregnancy. Furthermore, based on the limited available data, no evidence-based dosing regimen could be provided.

Thiopurines
Two studies focused on the pro-drug azathioprine (AZA) and mercaptopurine (MP)

Thiopurines
Two studies focused on the pro-drug azathioprine (AZA) and mercaptopurine (MP) [10,11]. AZA is converted mainly by glutathione S-transferases into MP. A big portion of MP is then metabolized by thiopurine-S-methyltransferase into the metabolite 6-methylmercaptopurine . Another portion of the MP will be metabolized via the purine salvage pathway into the three nucleotides, 6-thioguanine monophosphate, 6-thioguanine diphosphate and 6thioguanine triphosphate. The enveloping name of these three nucleotides is 6-thioguanine nucleotides (6-TGN). Since 6-MMP and 6-TGN are the metabolites of interest for the therapy, the studies reported their outcomes in the levels of these metabolites [12]. When those two studies were combined, the total amount of participants included was 72. The percentage of participants with UC was 25%, CD 71% and undetermined IBD 4%. The patient and study characteristics are elaborated in Appendix B Table A2. The participants using AZA (71%) were more prevalent than MP (29%). The studies showed similarities on multiple aspects in their analytical quantification methods. Quantification occurred by using a modified Dervieux method. Both measured the active metabolites of AZA and MP, namely 6-TGN and its potentially toxic variant 6-MMP in red blood cells (RBCs). The results are presented in Table 2. All measurements were performed from pre-pregnancy until after the delivery. Both studies show the same phenomenon when studying the changes of 6-TGN and 6-MMP levels throughout pregnancy. The 6-TGN levels are lower during the first, second and third trimesters compared to preconception. The most noticeable differences per trimester compared to pre-pregnancy levels were found in the study by Flanagan et al. and are as follows: T1 with 83%, T2 with 61% and T3 with 73% [10]. In contrast, the 6-MMP levels increased during pregnancy compared with the preconception state. The most extensive alteration per trimester was shown by Jharap et al., with 166% in T1 [11]; Flanagan et al., with 209% in T2 [10]; and Jharap et al. in T3, with 205% [11] compared to pre-pregnancy levels. After delivery, both 6-TGN and 6-MMP levels returned to the preconception baseline levels. Figures 2 and 3 show the differences in metabolite levels for both studies over time.
Although the PK of the thiopurines changed throughout pregnancy, no evidence-based dosing advice was provided. One study [10] suggested that caution should be exercised in case the doses are to be changed during pregnancy. This advice is based on their observation of shunting 6-MMP levels due to dosage change. An increase in thiopurine dose is sometimes inevitable, as a consequence of rising 6-MMP levels. However, if the 6-MMP levels stay below the toxic threshold and toxic side effects are absent, alterations in dosage seems to be possible.
In conclusion, two studies were available that covered the PK of thiopurines during pregnancy [10,11]. The therapeutic 6-TGN levels decreased during pregnancy, while the potential toxic 6-MMP levels rose. Although the PK of the thiopurines changed throughout pregnancy, no evidence-based dosing advice was provided. One study [10] advised to be cautious when the dosage is altered by monitoring for toxic side effects and keeping the 6-MMP levels below the toxic threshold.
Both measured the active metabolites of AZA and MP, namely 6-TGN and its potentially toxic variant 6-MMP in red blood cells (RBCs). The results are presented in Table 2. Al measurements were performed from pre-pregnancy until after the delivery.
Both studies show the same phenomenon when studying the changes of 6-TGN and 6-MMP levels throughout pregnancy. The 6-TGN levels are lower during the first, second and third trimesters compared to preconception. The most noticeable differences per trimester compared to pre-pregnancy levels were found in the study by Flanagan et al and are as follows: T1 with 83%, T2 with 61% and T3 with 73% [10]. In contrast, the 6-MMP levels increased during pregnancy compared with the preconception state. The most extensive alteration per trimester was shown by Jharap et al., with 166% in T1 [11] Flanagan et al., with 209% in T2 [10]; and Jharap et al. in T3, with 205% [11] compared to pre-pregnancy levels. After delivery, both 6-TGN and 6-MMP levels returned to the preconception baseline levels. Figures 2 and 3 show the differences in metabolite levels for both studies over time. Although the PK of the thiopurines changed throughou pregnancy, no evidence-based dosing advice was provided. One study [10] suggested tha caution should be exercised in case the doses are to be changed during pregnancy. This advice is based on their observation of shunting 6-MMP levels due to dosage change. An increase in thiopurine dose is sometimes inevitable, as a consequence of rising 6-MMP levels. However, if the 6-MMP levels stay below the toxic threshold and toxic side effects are absent, alterations in dosage seems to be possible.
In conclusion, two studies were available that covered the PK of thiopurines during pregnancy [10,11]. The therapeutic 6-TGN levels decreased during pregnancy, while the potential toxic 6-MMP levels rose. Although the PK of the thiopurines changed throughout pregnancy, no evidence-based dosing advice was provided. One study [10] advised to be cautious when the dosage is altered by monitoring for toxic side effects and keeping the 6-MMP levels below the toxic threshold. The concentration of 6-thioguanine nucleotides (6-TGN) during the different states of pregnancy. Concentrations of 6-TGN are expressed in pmol × 10 8 Red Blood Cell (RBC) count on the y-axis (median with corresponding 25th and 75th percentile). The different states of pregnancy per study are shown on the x-axis. The different states are expressed as pre-pregnancy (T0), trimesters one until three (T1, T2 and T3), delivery (T4) and postnatal (T5). F, the blue bar, represents the study of Flanagan et al. (2021); and J, the red bar, represents the study of Jharap et al. (2013) [10,11].  The concentration of 6-thioguanine nucleotides (6-TGN) during the different states o pregnancy. Concentrations of 6-TGN are expressed in pmol × 10 8 Red Blood Cell (RBC) count on the y-axis (median with corresponding 25th and 75th percentile). The different states of pregnancy pe study are shown on the x-axis. The different states are expressed as pre-pregnancy (T0), trimesters one until three (T1, T2 and T3), delivery (T4) and postnatal (T5). F, the blue bar, represents the study of Flanagan et al. (2021); and J, the red bar, represents the study of Jharap et al. (2013) [10,11].

TNF-α Inhibitors-Infliximab, Adalimumab and Golimumab
Within the group of the TNF-α inhibitors, nine studies presented suitable data for IFX, four studies for ADL and one study for GLM. The cumulative numbers of observed participants per drug were 83, 40 and 1 for IFX, ADL and GLM, respectively. When al participants within this group were combined (excluding the study of Borlik et al.), 75% were diagnosed with CD, 23% with UC and 2% with unspecified IBD. The range o median and mean ages was between 28.9 and 36 years within the studies (Appendix B  Table A2). In the case of IFX and ADL, respectively, four and two studies presented data over multiple timeframes. The study that covered GLM only presented data at delivery Five studies measured IFX data at one point, being three studies at delivery and two studies after pregnancy. In the case of ADL, two studies obtained their data at delivery

TNF-α Inhibitors-Infliximab, Adalimumab and Golimumab
Within the group of the TNF-α inhibitors, nine studies presented suitable data for IFX, four studies for ADL and one study for GLM. The cumulative numbers of observed participants per drug were 83, 40 and 1 for IFX, ADL and GLM, respectively. When all participants within this group were combined (excluding the study of Borlik et al.), 75% were diagnosed with CD, 23% with UC and 2% with unspecified IBD. The range of median and mean ages was between 28.9 and 36 years within the studies (Appendix B  Table A2). In the case of IFX and ADL, respectively, four and two studies presented data over multiple timeframes. The study that covered GLM only presented data at delivery. Five studies measured IFX data at one point, being three studies at delivery and two studies after pregnancy. In the case of ADL, two studies obtained their data at delivery. The included TNF-α inhibitor studies were predominantly found to be prospective cohort studies, four covering ADL and six IFX. One study was found to be a retrospective cohort study obtaining data from participants using IFX. Lastly, two IFX studies and one GLM study were determined as a case report (Appendix B Table A2).
When focusing on the PK parameters, all studies within the group of TNF-α inhibitors reported either the trough concentration (Ctrough), n = 4, or unspecified concentration (Cunspecified), n = 6 ( Table 2). One study used a population PK model to determine the clearance and volume of distribution for IFX [21]. They reported a clearance of 0.608 L/d and volume of distribution of 18.2 L. Four and two studies presented data on multiple time points throughout different stages of pregnancy for IFX and ADL, respectively. The data of these studies are shown in Figures 4 and 5, respectively.
For IFX (Figure 4), the authors, who measured IFX levels pre-pregnancy, during pregnancy and postpartum, determined an increase during pregnancy compared to prepregnancy levels [14,15,21]. The highest percentage of increase compared to pre-pregnancy levels was 123% [14] in the first trimester, 205% [21] in the second trimester and 305% [14] in the third trimester. The IFX levels after delivery compared to pre-pregnancy levels were higher in Seow et al. [14] and Flanagan et al. [15] (10.17 against 6.9 µg/mL and 10.3 against 7.9 µg/mL, respectively) and were lower in the study of Grišić et al. [21] (5.9 against 7.3 µg/mL). The IFX levels after pregnancy were all lower than during pregnancy. However, the degree in change was different among studies. Lastly, Figure 4 shows a large dispersion in data at the after-delivery moment. Two studies seem to have high concentrations compared to all other studies [13,18]. It has to be noted that the high variability ( Figure 4) is probably due to the fact that these studies were case reports in which outliers are more easily visible compared to a median or mean values represented in cohort studies. concentration (Cunspecified), n = 6 ( Table 2). One study used a population PK m determine the clearance and volume of distribution for IFX [21]. They reported a c of 0.608 L/d and volume of distribution of 18.2 L. Four and two studies presented multiple time points throughout different stages of pregnancy for IFX an respectively. The data of these studies are shown in Figures 4 and 5,respectively For IFX (Figure 4), the authors, who measured IFX levels pre-pregnancy pregnancy and postpartum, determined an increase during pregnancy compare pregnancy levels [14,15,21]. The highest percentage of increase compared pregnancy levels was 123% [14] in the first trimester, 205% [21] in the second t and 305% [14] in the third trimester. The IFX levels after delivery compared pregnancy levels were higher in Seow et al. [14] and Flanagan et al. [15] (10.17 ag μg/mL and 10.3 against 7.9 μg/mL, respectively) and were lower in the study of al. [21] (5.9 against 7.3 μg/mL). The IFX levels after pregnancy were all lower tha pregnancy. However, the degree in change was different among studies. Lastly, shows a large dispersion in data at the after-delivery moment. Two studies seem high concentrations compared to all other studies [13,18]. It has to be noted that variability ( Figure 4) is probably due to the fact that these studies were case re which outliers are more easily visible compared to a median or mean values rep in cohort studies.  [13][14][15][16][17][18][19][20][21] in μg/m on y-axis) during the different stages of pregnancy (shown on x-axis). The different expressed as pre-pregnancy (T0), trimesters one until three (T1, T2 and T3), delivery postnatal (T5).
In Figure 5, the ADL concentrations over time are provided. The authors [14,15] mentioned that the ADL concentration during pregnancy is relatively stable compared to the ADL concentration before and after pregnancy. It is, however, observed that pregnancy levels are lower than pre-pregnancy levels. Furthermore, the second trimester contains a discrepancy. No reason for this discrepancy was reported by the authors. the ADL concentration before and after pregnancy. It is, however, observ pregnancy levels are lower than pre-pregnancy levels. Furthermore, the second t contains a discrepancy. No reason for this discrepancy was reported by the autho Not all included studies provided dose advice, and when they did, it was advice [14,15,17,[19][20][21]. However, one study from Steenholdt et al. [17] provided a target advice. A concentration of 0.5 mg/mL was considered as a therapeutic th [17]. Looking at the other studies presenting dose advice for IFX, the followin were found. Four studies mentioned that dosing for IFX should be halted at the en second trimester or the beginning of the third trimester [14,[19][20][21]. The main re above-mentioned advice is to suppress, as much as possible, immune respon birth.One study suggested that the dose could be changed during pregnancy to t end of the therapeutic range [14], while another study did not recommend a ch dose [15]. Although the PK of biologicals changed throughout pregnancy, non studies indicated how dosing should be adapted during pregnancy. The applicable to ADL, where three studies [14,19,20] advised to stop dosing after th trimester. No specific dosing advice was provided for earlier trimesters based on PK data. The same study as with IFX saw no problem in changing the dose pregnancy to the lower end of the therapeutic range, while another study recommend a dose change [14,15]. For GLM, no dose advice was given by the au

Integrin Inhibitor-Vedolizumab
Two studies provided data for a total of 28 pregnant women with IBD. women, 42% were diagnosed with CD and 58% with UC. The median ag participants was 30.7 years in the study of Flanagan et al. and 31 years in the Mitrova et al. [15,24]. The study of Flanagan et al. [15] provided data over moments within the pregnancy-until-delivery timeframe, namely T1 19.1 (13-23)  [14,15,19,20] in µg/mL (on the y-axis) during the different stages of pregnancy (on the x-axis). The different states are expressed as pre-pregnancy (T0), trimesters one until three (T1, T2 and T3), delivery (T4) and postnatal (T5).
Not all included studies provided dose advice, and when they did, it was general advice [14,15,17,[19][20][21]. However, one study from Steenholdt et al. [17] provided a specific target advice. A concentration of 0.5 mg/mL was considered as a therapeutic threshold [17]. Looking at the other studies presenting dose advice for IFX, the following results were found. Four studies mentioned that dosing for IFX should be halted at the end of the second trimester or the beginning of the third trimester [14,[19][20][21]. The main reason for above-mentioned advice is to suppress, as much as possible, immune response after birth.One study suggested that the dose could be changed during pregnancy to the lower end of the therapeutic range [14], while another study did not recommend a change in dose [15]. Although the PK of biologicals changed throughout pregnancy, none of the studies indicated how dosing should be adapted during pregnancy. The same is applicable to ADL, where three studies [14,19,20] advised to stop dosing after the second trimester. No specific dosing advice was provided for earlier trimesters based on changed PK data. The same study as with IFX saw no problem in changing the dose during pregnancy to the lower end of the therapeutic range, while another study did not recommend a dose change [14,15]. For GLM, no dose advice was given by the authors.
Concerning the PK data, both studies presented their values as concen  [15,24] in μ the y-axis) during the different stages of pregnancy (on the x-axis). The different states are e as pre-pregnancy (T0), trimesters one until three (T1, T2 and T3) and delivery (T4).
Flanagan et al. mentioned that no dose change was recommended for VDZ

Interleukin 12/23 Inhibitor-Ustekinumab
Two studies focused on the use of UST in pregnant women with IBD. In participants were included, of which 94% were diagnosed with CD and 6% with study of Mitrova et al. [24] was a prospective cohort study in which the median 28 years. The study of Sako et al. was a case report in which the woman was 35 age [22]. Both studies presented their data as unspecified concentration, only at d Since the concentration was only available at delivery, it was not possible to behavior of the UST concentration during the pregnancy. As a consequence, due of data, the authors of these articles could not provide a dose advice.
In conclusion, 12 studies were found that presented drug concentrations ADL, VDZ, UST and GLM. Most studies [16,19,20,[22][23][24] only presented a conce at delivery. The studies that presented data during the whole pregnancy sho increase in concentration for IFX, a relative stable concentration for ADL and a de concentration for VDZ. Although the PK of the biologicals changed thr pregnancy, no evidence-based dosing advice was provided. One study presented advice, being that 0.5 mg/mL for IFX seemed to be a therapeutic concentration.  [15,24] in µg/mL (on the y-axis) during the different stages of pregnancy (on the x-axis). The different states are expressed as pre-pregnancy (T0), trimesters one until three (T1, T2 and T3) and delivery (T4).
Flanagan et al. mentioned that no dose change was recommended for VDZ [15].

Interleukin 12/23 Inhibitor-Ustekinumab
Two studies focused on the use of UST in pregnant women with IBD. In total, 16 participants were included, of which 94% were diagnosed with CD and 6% with UC. The study of Mitrova et al. [24] was a prospective cohort study in which the median age was 28 years. The study of Sako et al. was a case report in which the woman was 35 years of age [22]. Both studies presented their data as unspecified concentration, only at delivery. Since the concentration was only available at delivery, it was not possible to see the behavior of the UST concentration during the pregnancy. As a consequence, due to a lack of data, the authors of these articles could not provide a dose advice.
In conclusion, 12 studies were found that presented drug concentrations for IFX, ADL, VDZ, UST and GLM. Most studies [16,19,20,[22][23][24] only presented a concentration at delivery. The studies that presented data during the whole pregnancy showed an increase in concentration for IFX, a relative stable concentration for ADL and a decreasing concentration for VDZ. Although the PK of the biologicals changed throughout pregnancy, no evidence-based dosing advice was provided. One study presented a target advice, being that 0.5 mg/mL for IFX seemed to be a therapeutic concentration. Table 2. Summary of the pregnancy-induced changes in the pharmacokinetics of IBD-related drugs. The data are presented as mean (SD), median (IQR) or alternative method, indicated next to the corresponding value. Each row is dedicated to a medicine. When a column overlaps multiple rows, the data are shared over multiple rows.
The interval between the last intake of the drugs and the delivery was between 5 and 24 h. The data were extracted via a plot digitizer. The 6-TGN median levels in T2 were significantly lower than observed from T0 to T5 (p < 0.001). This was still the case when adjusted for patient weight during pregnancy.
When considering an increase in thiopurine dosing during pregnancy, extra attention should be paid to TDM, since this study observed an increase in 6-MMP levels with even the slightest change in dose elevation.
Data were included only if at least two observations between T0 and T5 were available; on stable dosing, for at least four weeks before testing.
Two patients were excluded due to a dose change between T0 and T5. The total amount of participants included in the study went from 42 to 40.    The authors found that an IFX concentration of 0.5 µg/mL and higher is associated with maintained response in both CD and UC. They suggest this as a valid cut-off level for clinically relevant IFX concentrations.
Infusions happened at 20 and 31 weeks of GA. After delivery, only two infusions are specified with the corresponding date.
It should be noted, that at infusion 12, which is 4 infusions later than the last measured concentration, had a high Ctrough of 2.1 µg/mL T5, concentration of 0.6 µg/mL is extracted via a plot digitizer.  A significant correlation for VDZ was found between maternal drug level and gestational week of the last administration (ρ = 0.751, p = 0.001). Another correlation for VDZ was found between maternal drug level and the interval between the last infusion and delivery (ρ = −0.917, p < 0.001).

Discussion
To our knowledge, this is the first time a systematic review was conducted on the available data of PK parameters related to IBD drugs in pregnant women with IBD. Our ultimate goal was to provide an in-depth overview of the available PK data. Before evidence-based dosing can be proposed for pregnant women, insight into the PK has to be gained to optimize drug therapy for both the mother and fetus. Limited PK studies on IBD drugs have been performed during pregnancy, and, in general, they have not resulted in obtaining PK parameters in the different pregnancy trimesters. Although the PK of the IBD-related drugs changed throughout pregnancy, no evidence-based dosing advice was provided.
When focusing on the present guidelines of the European Crohn's and Colitis Organization (ECCO) and the American Gastroenterological Association (AGA), both state that staying in remission is important to minimize adverse outcomes. Therefore, non-teratogenic medication should be used during pregnancy in order to reduce the chance of a flare during pregnancy [25,26]. Flares are a risk factor for adverse outcomes for both the fetus (e.g., increased chance of preterm birth and low birth weight) and for the mother (e.g., emergency caesarian delivery and thromboembolic events) [25,27]. The ECCO and AGA consider mesalazine, sulfasalazine, thiopurines, biologicals and corticosteroids (for a short period) to be safe when used for maintenance therapy during pregnancy. Tofacitinib is a relatively new small-molecule drug with limited human data in pregnancy. The producer and the AGA suggest that tofacitinib should not be advised to be used, especially not in the first trimester of pregnancy [26,27].
Our systematic review shows that concentrations of IBD drugs vary during the different trimesters of pregnancy. However, since information is too lacking to give dosing advice, there is a need to expand the study duration over multiple trimesters to obtain the PK of IBD drugs. In this discussion, the most important findings arising from this review and the still remaining PK-related knowledge gaps are discussed. Furthermore, a recommendation is made regarding information that still needs to be collected in order to develop evidence-based dosing for IBD-drugs in pregnant women with IBD, while also taking the fetus into account.
One study was found in which concentrations at delivery were presented for mesalazine [9]. No dose advice and different dosages and routes of administration, in combination with only five participants studied, made us question the usability of this study. We conclude that this commonly used drug in IBD is overlooked in the literature. Only two prospective studies presented concentrations of the thiopurine metabolites 6-TGN and 6-MMP. Both studies found that, during pregnancy, the therapeutic 6-TGN levels decreased, while the potential toxic 6-MMP levels increased. After delivery, both levels returned to pre-pregnancy levels. The authors hypothesize that enzymes are likely to be the cause of this shift, but further research needs to confirm this suggestion. Especially thiopurine S-methyltransferase and NUDT15 play a key role in this metabolism. Despite the increase of the 6-MMP levels, thiopurines are not considered teratogenic in humans [10,11]. Twelve studies covered the biologicals, in which data for five types of biologicals were presented. Except for one study, all studies reported only concentrations and no PK-parameter values. With the concentrations, the influence of pregnancy on the drug levels could be determined. The IFX levels increased, ADL levels stayed relatively stable and VDZ levels decreased during pregnancy. After pregnancy, the drug levels of the biologicals were lower compared to the pre-pregnancy levels. The IFX levels in Figure 4 showed discrepancies at T5 for two studies [13,18]. The discrepancies at T5 may possibly be related to the time of measurement after delivery.  [17,21]. For UST and GLM, the data were too scarce to observe a trend in concentration during the pregnancy. The reasons for these trends remain unclear. Some suggestions were made about the size of monoclonal antibodies. Due to their high molecular size and hydrophilic characteristics, the biologicals tend to have a small volume of distribution, limited to the plasma and extracellular fluids. One could argue that the increased plasma volume in a pregnant woman has an impact on the PK of the monoclonal antibodies, but since the volume of distribution for biologicals is small, consequentially the PK of monoclonal antibodies is not altered. [15]. For corticosteroids, no studies were found for pregnant women with IBD. However, from the literature search, five articles concerning betamethasone (BET) and two articles covering prednisone and prednisolone were found for other indications than IBD [28][29][30][31][32][33][34]. However, in the clinical setting, sometimes dosing advice needs to be determined for drugs (e.g., corticosteroids) that are not investigated in pregnant women with IBD. Investigation of the PK of a drug used in a population of pregnant women through an alternative route or for a different indication can be a helpful first step. The article characteristics are available in Appendix C Table A3, and the results are available in Appendix C Table A4. In contrast to the thiopurine studies and studies with biologicals, the corticosteroid studies for other diseases than IBD presented their data in PK parameters instead of concentrations. The same trend in dose-independent PK changes in corticosteroids during pregnancy was found in several studies [32]. All studies mentioned that the clearance of BET increased during pregnancy. This is likely to be originating from the CYP3A4 enzyme and 11β-hydroxysteroid dehydrogenase 2 (11β-HSD2) activities [32][33][34]. Prednisone and prednisolone are affected by these enzymes, too.
When specifically focusing on the limitations of the included studies in this review, the studies with thiopurines and biologicals only presented sparse drug concentrations over time. These specific data give insight into the behavior of these drugs during pregnancy, but a more complete view of the PK parameters would be desirable. One of the reasons for the lack of PK parameters generated from the obtained concentrations over time for thiopurines, as well as biologicals, is the fact that often only one blood sample per patient per trimester has been determined. PK parameters cannot be calculated based on just one sample per trimester in such a small study population. Population PK modeling can be a useful tool, not only to predict PK parameters, but also to develop more evidencebased dosing in special populations, such as pregnant women and fetuses [35]. The advantage of such a population PK model is that all individual concentrations of all patients will be analyzed together in a population setting, while, at the same time, data from individual patients are still distinguishable. Inter-and intra-patient variability can still be characterized. The advantage of this technique is that no complete PK profile of thiopurines or biologicals per patient, for example, is needed. The patient-related (i.e., age, trimester, weight, disease state and single vs. twin pregnancy) and treatment characteristics (i.e., route of administration) can thereby be used to (partly) understand and explain the inter-individual and intra-individual variability in these pharmacokinetics parameters in pregnant women. Therefore, those covariates can be used to determine if and how dosing can be individualized. After the development of such a pharmacokinetics model, the dosage needed to reach a specific target concentration can be developed. After the development of a PK model and model-based dosing, it would be of the utmost importance to prospectively validate the model-based dosing in a clinical study, not only to investigate whether the target concentration is reached, but also to investigate if the safety values are within the reference range. A first step could be to evaluate the already performed PK studies on quality and the amount of data, including clinical characteristics, drug concentrations in plasma, number of patients and time of sampling, retrieved from these studies in order to perform a pooled-PK analysis [35]. Such a pooled analysis has already been performed by Colin et al. for vancomycin in other special populations, with the aim to study all common covariates in adults in datasets on intravenous vancomycin [36]. In this way, a pooled analysis could be performed with all PK data of the pregnant population. After developing a PK model specific for pregnant women, a next step could be to design a new study with a specific focus on, for example, additional covariates that have not yet been studied in already published datasets and that could possibly explain the residual variability. In this way, we should use these already available datasets and published population PK models to put new datasets into these perspectives. This is an effective approach to explore additional covariates or specific subpopulations, but it should be preceded by a critical assessment of the published models [35].
Furthermore, in this review, the focus was on the PK of IBD drugs used by pregnant women with IBD. Therefore, the effect of the drugs on the fetus was not in the scope of this review. However, for IBD drugs transferring the placenta (e.g. thiopurines, ADL and IFX) fetal exposure as well as fetal outcomes (e.g., safety-related parameters) are important to monitor. Within the group of the thiopurines, Jharap et al. [11] reported that thiopurine exposure may cause neonatal anemia. This outcome, however, was not supported by Flanagan et al. [10]. Those authors reported that 80% of the infants at 6 weeks of age showed neonatal thrombocytosis and abnormal liver function [10]. When more data are gathered, a more conclusive statement can be made. In regard to the biologicals, these drugs are not linked to short-term severe adverse outcomes. On the other hand, these drugs are relatively new, and, therefore, the long-term outcomes are yet to be uncovered. Drugs such as IFX, which belong to the IgG1 subfamily, are actively transported over the placenta and, thus, expose the fetus to these drugs [23]. The corticosteroids, although mostly investigated in pregnant women with another indication than IBD, did not show any life-threatening adverse outcomes. The fetus seems to be protected by the more prevalent 11β-HSD2 enzyme, which turns the active prednisolone into inactive prednisone. Compared to the maternal body, in which the 11β-HSD1 enzyme is more present, the opposite drug ratio is observed [30,31]. The ECCO states that some risks, such as orofacial malformations, are found in the newborns, but with a small risk. Despite the low risk of serious adverse outcomes for both the newborn and mother, clinicians should be aware of the potential risks that corticosteroids could cause. Due to the potential risks, the use of corticosteroids is reserved only for case of flares.

Conclusions
In conclusion, we conducted a systematic review of the literature containing the available values of PK parameters related to IBD-drugs in pregnant women with IBD. We found relevant studies that presented the results for aminosalicylates, thiopurines and biologicals. In general, no PK values were found other than concentrations. Thiopurine metabolite concentrations tend to alter per consecutive trimester, while biologicals show that the concentrations are either rising, remain stable or are decreasing depending on the specific biological. Studies concerning corticosteroids presented values for a wide variety of PK parameters, but they did not include IBD pregnant women. We confirmed that there is a knowledge gap concerning the PK of IBD-related drugs in pregnant women with IBD. In the future, more PK studies on IBD drugs have to be performed in order to develop evidence-based dosing.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable. This systematic review was not registered in advance. Appendix B Table A2. Patient and study characteristics of the included studies in this systematic review. The data are presented as mean (SD), median (IQR) or alternative method, indicated next to the corresponding value. Each row is dedicated to one medicine. When a column overlaps multiple rows, the data are shared over multiple rows.     Table A3. Patient and study characteristics of the included studies in this systematic review. The data are presented as mean (SD), median (IQR) or alternative method, indicated next to the corresponding value. Each row is dedicated to one medicine. When a column overlaps multiple rows, the data are shared over multiple rows.     twin pregnancies compared to singleton pregnancies (p < 0.017).

Appendix C. Corticosteroids Study Characteristics and PK-Data throughout Pregnancy
- The clearance of BET seemed to be higher in pregnant women with twins than singleton pregnancies.
However, this was not proven to be significant. The serum albumin level was significantly higher in multichorionic twin pregnancies compared to dichorionic twin and singleton pregnancies (p = 0.0222).
The V d /F was not different among groups, but it did correlate with the BMI (ρ = 0.4530, p = 0.02).
After a single i.m. dose, AUC0-∞ was significantly higher in singleton pregnancies compared to dichorionic twin pregnancies (p = 0.0345).
The Cl/F was significantly lower in singleton pregnancies (p = 0.0324) compared to dichorionic twin pregnancies.
Pregnancy had a lower amount of unbound concentration compared to postpartum for prednisone (p = 0.003) and prednisolone (p < 0.001).
Prednisolone may be preferred over betamethasone or dexamethasone because prednisolone marginally reaches the fetus compared to the previous mentioned drugs. Early preterm pregnancy does not compromise oxidation of prednisolone.
Prior to the start of the study, all participants received two i.m. doses of 11.4 mg BET, with an interval of 24 h.