European Journal of Clinical Pharmacology

, Volume 61, Issue 7, pp 517–522 | Cite as

Pharmacokinetics and transplacental distribution of fentanyl in epidural anesthesia for normal pregnant women

  • Elaine Christine Dantas Moisés
  • Luciana de Barros Duarte
  • Ricardo de Carvalho Cavalli
  • Vera Lúcia Lanchote
  • Geraldo Duarte
  • Sérgio Pereira da CunhaEmail author
Pharmacokinetics and Disposition


BACKGROUND: Fentanyl is an opioid drug widely used as a co-adjuvant in abdominal delivery, a fact that justifies its pharmacokinetic study under these conditions. OBJECTIVE: Our objective was to investigate the pharmacokinetics and placental transfer of fentanyl in parturients whose pregnancies were resolved by cesarian section with epidural anesthesia. PATIENTS and METHODS: Ten clinically normal parturients who delivered at term received 5 ml of 2% lidocaine hydrochloride without a vasoconstrictor for skin and subcutaneous blockade, followed by epidural injection of 2 ml fentanyl citrate (0.05 mg/ml), 15 ml 0.5% bupivacaine hydrochloride with 1:200,000 epinephrine, and 10 ml 2% lidocaine hydrochloride without a vasoconstrictor. Maternal blood samples were collected at various times after injection (1–840 min), and the fentanyl plasma concentrations were determined by gas chromatography-mass spectrometry. Pharmacokinetic analysis was performed using the bi- or tricompartmental model. The fetal/maternal ratio of the plasma fentanyl was determined at birth. RESULTS: The values of the pharmacokinetic parameters were: t½α = 13.5 min, t1/2β = 192.5 min, t1/2γ = 620 min, AUC0-∞ = 137.404 ng.min per millilter, Cl/f = 464.984 ml/min, Vd/f = 299.974 l, Cl/f/kg = 6.875 ml/min per kilogram, and Vd/f/kg = 4.441 l/kg. The latency between drug administration and birth was 28.5 min, with a maternal and fetal plasma concentration of 0.310 and 0.245 ng/ml, respectively, at a median fetal/maternal ratio of 0.892. CONCLUSION: The study demonstrated a rapid passage of fentanyl from the epidural space to maternal blood and a significant transplacental transfer of maternal fentanyl of about 90%, which should serve as an alert to obstetricians.


Obstetrical analgesia Epidural route Fentanyl Pharmacokinetics Placental transfer 


A particular property of obstetrical anesthesia is the possibility of drug transfer to the fetus. Many drugs can freely cross the placental barrier, leading to adverse fetal reactions such as changes in the heart rate and in tonus and respiratory depression. Fetal exposure and the neonatal effects of drugs administered to the mother before delivery depend directly on the pharmacokinetics of the drug in the mother, in the fetus, and in the placenta [1].

Fentanyl is a drug extensively used in obstetrical anesthesia because of its short time to action peak (about 5 min), the rapid termination of its effect after a single dose, and the occurrence of relative cardiovascular stability [2]. However, at high doses or when continuously infused, this drug has longer lasting effects. When fentanyl is administered by the epidural route, the time to the beginning of its action is three- to fivefold slower and the duration of its effect is three- to sixfold longer than when it is administered intravenously [3].

A sound knowledge of fentanyl pharmacokinetics in pregnant women is necessary in order to contribute to the improvement of analgesia and anesthesia utilized in obstetrical practice. We report here the first complete evaluation of fentanyl pharmacokinetic parameters in parturients submitted to epidural anesthetic blockade for delivery by cesarean section. The fentanyl plasma concentrations were analyzed by gas chromatography-mass spectrometry (GC-MS). A second objective was an investigation of the placental transfer of fentanyl in parturients.

Materials and methods

Patients and methods

The study was approved by the Research Ethics Committee of the University Hospital, Faculty of Medicine of Ribeirão Preto, University of São Paulo, Brazil. All patients who participated in the study signed a term of informed consent. It should be pointed out that the study did not interfere with the clinical conduct applied to each case.

The group consisted of ten parturients with no base disease that had been admitted to the Obstetrical Center of the Airport Complex Maternity (MATER), a maternity unit considered to be of low risk linked to the Department of Gynecology and Obstetrics of the Faculty of Medicine of Ribeirão Preto, University of São Paulo. The parturients were submitted to resolution of pregnancy by the abdominal route (cesarean section) due to obstetrical indication. Patient selection was aleatory, with no randomization. The parturients were not paired but had similar characteristics.

The exclusion criteria were as follows: delivery under urgent/emergency obstetrical situations, presence of maternal disease, chronic use of medication (except multiple vitamin supplements), altered prenatal exams, and patients who did not agree to participate in the study or did not sign the informed consent term. The discontinuation criteria were: patients who no longer wished to participate in the study and patients who required drugs that might interfere with the pharmacokinetics of fentanyl during the anesthetic-surgical procedure.

Epidemiological data were collected using a standard protocol which included the following parameters: age, life habits, use of medications, anthropometric data, parity, gestational age, and ultrasonographic information about fetal and/or placental changes. The indications for the resolution of pregnancy by cesarean delivery were also recorded.

The parturients were submitted to locoregional peridural anesthesia with 2% lidocaine hydrochloride without a vasoconstrictor (Xylestesin®, Cristália, Itapira, SP, Brazil; lot 02030982) in the amount of 5 ml for skin and subcutaneous blockade and 10 ml in the peridural space, 2 ml of 0.05 mg/ml fentanyl citrate (Fentanest®, Cristália, Itapira, SP, Brazil; lot 02041393) in the peridural space, and 15 ml of 0.5% bupivacaine with 1:200,000 epinephrine (Neocaína®, Cristália, Itapira, SP, Brazil; lot 53200809) in the peridural space.

A maternal blood sample was collected before the administration of the anesthetics for hematologic and biochemical evaluation in order to detect possible unidentified base diseases. Serial maternal blood samples were then collected for the determination of fentanyl concentration at 1, 5, 15, 30, 45, 60, 120, 240, 360, 480, 600, 720, and 840 min after drug administration, and a blood sample was obtained from an umbilical cord vein at birth.

Determination of fentanyl in plasma

A stock solution of fentanyl citrate was prepared at the concentration of 0.5 mg fentanyl/ml methanol and then diluted to concentrations of 1–40 ng/ml methanol. The clobazam solution (internal standard) was prepared at the concentration of 20 ng/ml methanol. The sodium hydroxide solution (analytical grade, Mallinckrodt, St. Louis, Mo.) of 12 mol/l was washed with the extraction solvent cyclohexane. Methanol, cyclohexane, and toluene were obtained from Mallinckrodt as residue grade.

Chromatographic analysis of fentanyl in plasma

Twenty-five-microliter aliquots of 12 M NaOH solution were added to 2-ml plasma samples. After shaking for 1 h on a horizontal shaking table at 220 cycles/min, the tubes were centrifuged at 2000 rpm at 4°C for 10 min. The organic phases (3.5 ml) were transferred to clean tubes, 25 μl of the internal standard solution (clobazam) was added, and the mixture was evaporated to dryness with a Jouan evaporator (Winchester, Va.) model RC1022/RCT90 for 30 min. The dry extracts were reconstituted in 25 ml of a 12 mM triethylamine solution in toluene, and 2-μl samples were submitted to chromatographic separation and detection by mass spectrometry.

The chromatographic system consisted of a gas chromatograph coupled to a Shimadzu mass spectrometer (Kyoto, Japan) model CG-MS QP 5000 with an injector operating at 280°C in the splitless mode. Fentanyl and the internal standard were separated from the endogenous constituents of plasma through an HP ultra column (length: 12 m; inner diameter: 0.20 mm; film thickness: 0.33 μm; Hewlett-Packard, Wilmington, Del.). Ultrapure helium was used as the mobile phase at a pressure of 120 kPa. The column was programmed to operate under the following temperature conditions: an initial temperature of 150°C for 1 min, then increasing to 280°C at a rate of 40°C/min, remaining at the final temperature for 3.5 min. The detector operated in the single ion monitoring mode at 300°C, and ionization was obtained by electronic impact. For quantification, the ions m/z=300.0 (internal standard) and m/z=245.2 (fentanyl) were monitored.

For the construction of the calibration curve (0.025–1.0 ng/ml plasma), 50 μl of each standard fentanyl solution in methanol were added to 2-ml aliquots of blank plasma obtained from healthy volunteers who had received no medication during the preceding 10 days. The samples were then submitted to the extraction and chromatographic analysis procedures described herein. The calibration curve presented a linear regression equation of 3.047 × −0.03823, with a coefficient of determination of 0.9971.

The recovery of fentanyl from plasma was assessed by comparing the areas of the peaks obtained after plasma extraction to the areas of the peaks obtained after direct injection of the standard solution. Recovery was determined in triplicate for each fentanyl concentration, resulting in a coefficient of variation of 10% or less. The limit of quantitation (LOQ) was obtained by the analysis – in quintuplicate – of plasma samples spiked with fentanyl at concentrations as low as 0.01 ng/ml. The LOQ was defined as the lowest plasma concentration of fentanyl analyzed that had an error of 20% or lower. Under these conditions, LOQ was 0.025 ng/ml with an acceptable precision (18%) and accuracy (−12.5%). Intra- and interassay precision and accuracy were determined by analyzing plasma samples spiked with fentanyl. The samples were analyzed with one replicate using a single calibration curve for intra-assay evaluation and in duplicate over five consecutive days for interassay evaluation. The coefficients of variation obtained in the study of intra- and interassay precision for low and high fentanyl concentrations (0.05 and 1.0 ng/ml, respectively) were less than 15%.

Pharmacokinetic analysis of fentanyl

Plasma fentanyl concentrations versus time profiles were analyzed using both noncompartmental and compartmental pharmacokinetic methods (PK solutions, version 2.0, Monrose, USA). For noncompartmental analysis, Cmax and tmax were determined by visual inspection of the profiles. The terminal half-life (t½γ) was determined by linear regression of the terminal linear portion of the concentration versus the time curve. The distribution (t½α) and elimination (t½β) half-lives were determined by the method of the residuals. The rate constants (α, β, and γ) were calculated as 0.693/ t½. The area under the plasma concentration versus time curve (AUC0-840) from time 1 min to the last quantifiable concentration (Ct) was calculated by trapezoidal integration. The area with extrapolation to infinity (AUC0-∞) was calculated from AUC0-840 + Ct/γ. The apparent total clearance (Cl/f) and the volume of distribution (Vd/f) were calculated as dose/AUC0-∞ and dose/(AUC0-∞.λ), respectively.

The placental transfer of fentanyl was determined based on the ratios of fentanyl concentrations between the umbilical and maternal venous blood.

The position and dispersal measurements [expressed as median and 25th (P25) and 75th percentiles (P75)] of each variable analyzed were calculated using the Microsoft Excel® software.


The study was conducted on ten parturients considered to be normal in clinical and laboratory terms and who showed no untoward events during the prenatal period. The median and 25th and 75th percentiles for maternal age, gestational age, body weight, height, and body mass index are presented in Table 1. One patient was a primigesta and the remaining ones were multigestae; all delivered at term (more than 259 days of gestation).
Table 1

Maternal demographic characteristics


Parturients (n=10) [Median (P25–P75)]a

Age (years)

28.5 (23.8–31.8)

Gestational age (days)

271.5 (267.3–282.3)

Weight (kg)

69.3 (68.1–76.5)

Height (m)

1.6 (1.6–1.7)

Body mass index (kg/m2)

27.2 (24.8–29.1)

aP25, 25th percentile; P75, 75th percentile

No diseases were diagnosed in any infant at birth, with eight newborns being boys and two being girls. Median birth weight was 3170 g, and median length was 47 cm. All newborns presented an Apgar score equal to 10 at the 5th min and a favorable course during the postnatal period; they were subsequently discharged from the hospital together with their mothers. Median placental weight was 540 g, and there were no visible macroscopic changes to the placenta.

Maternal systolic and diastolic arterial pressure and heart rate were monitored during the collection of blood samples for the determination of fentanyl concentration in maternal plasma, demonstrating hemodynamic stability throughout the study.

The median and 25th and 75th percentiles of the plasma concentrations as a function of time of collection are presented in Fig. 1. Pharmacokinetic parameters of fentanyl in parturients are presented in Table 2. The data were calculated using a bicompartmental model (six patients) or tricompartmental model (four patients).
Fig. 1

Graphical presentation of the median and 25th and 75th percentiles of the plasma concentrations of fentanyl (in nanograms per milliliter) determined by GC-MS as a function of time (in minutes) in parturients (n=10)

Table 2

Pharmacokinetic parameters of fentanyl in parturients calculated from plasma fentanyl concentrations (in nanograms per milliliter) graphically plotted against time (in minutes)


Parturients (n=10) [Median (P25–P75)]b

t½α (min)

13.5 (9.0–24.5)

α (min−1)

0.05 (0.03–0.08)

t½β (min)

192.5 (119.3–242.5)

β (min−1)

0.004 (0.003–0.006)

t½γ (min)

620.0 (526.3–752.5)

γ (min−1)

0.001 (0.001–0.001)

AUC0-∞ (ng.min/ml)

137.4 (99.0–150.8)

Cl/f (ml/min)

465.0 (422.0–657.6)

Cl/f/kg (ml/min/kg)

6.9 (6.3–8.9)

Vd/f (l)

300.0 (149.8–400.2)

Vd/f/kg (l/kg)

4.4 (2.4–5.6)

at½α, Distribution half-life; α, distribution rate constant; t½β, elimination half-life; β, elimination rate constant; t½γ, terminal half-life; γ, terminal rate constant; AUC0-∞, area under the curve for plasma concentration versus time; Cl/ f, apparent clearance; Vd/ f, apparent distribution volume

bP25, 25th percentile; P75, 75th percentile

Placental transfer is shown in Table 3. The fact that patients 9 and 10 showed fetal plasma concentrations differing from those of the remaining patients was interpreted as a mistake in the collection procedure, and we decided to exclude these patients from the calculation of the fetal/maternal ratio and from the evaluation of placental transfer. The median latency between administration of the drug and birth was 28.5 min, and at delivery the median maternal plasma concentration was 0.31 ng/ml and the fetal median concentration was 0.25 ng/ml, with a median fetal/maternal ratio of 0.89 (Table 3).
Table 3

Fentanyl concentrations (in nanograms per milliliter) in maternal and fetal plasma determined by GC-MS, and fetal/maternal ratio


Parturients (n=8) [Median (P25–P75)]b

Latency (min)

28.50 (25.75–32.25)

Maternal concentration (ng/ml)

0.31 (0.28–0.32)

Fetal concentration (ng/ml)

0.25 (0.19–0.51)

Fetal/maternal ratio

0.89 (0.75–1.62)

aLatency, Time between drug administration and birth; fetal/maternal ratio, ratio between fetal and maternal fentanyl concentration in plasma at the time of birth

bP25, 25th percentile; P75, 75th percentile


The patients comprising the group included in the present study were homogeneous with respect to the demographic and anthropometric characteristics analyzed. This is an important and relevant factor given our knowledge that anthropometric data can influence the pharmacokinetics of drugs [4]. The median values of arterial pressure and heart rate during the periods of blood sample collection demonstrated the presence of maternal hemodynamic stability, ruling out the possibility of an influence of hemodynamic variations on the pharmacokinetics of the drug in the members of the study group.

The administration of fentanyl by the epidural route, an area rich in vascularization, led to the rapid absorption and subsequent rapid detection of the drug in plasma as early as by the first blood collection (1 min). Justins et al. [5] also reported the early detection of fentanyl in plasma following the epidural administration of this agent to pregnant women for labor analgesia. The plasma fentanyl concentration versus time curves do not demonstrate the systemic phase of absorption of the drug following epidural administration. This results is most likely due to the liposoluble characteristics of fentanyl and, consequently, its high ability to cross lipid membranes, which result in a shorter period of absorption and equilibrium between plasma and the epidural space [2].

Little [6], in a literature review of the pharmacokinetics of drugs during pregnancy, concluded that the AUC0-∞ tends to decrease in pregnancy. In the present study, the median AUC0-∞ was 137.4 ng/min per millilter (Table 2). In a survey of the literature we did not find any reports on AUC0-∞ parameters following the administration of fentanyl by the epidural route.

The distribution volume and total clearance are parameters that depend on the bioavailability of fentanyl when administered by the epidural route. This information is not available for fentanyl administered by the epidural route to pregnant women. We obtained a median Cl/f of 6.9 ml/min per kilogram and a median Vd/f of 4.4 l/kg (Table 2). Scholz et al. [7] reported a distribution volume of 4.0 l/kg and a total clearance of 0.78 l/h per kilogram in a population of non-pregnant women receiving a single intravenous dose of fentanyl, while Tegeder et al. [8] reported a distribution volume of 200–560 l and a clearance of 600–1000 ml/min for fentanyl administered intravenously in a single dose to healthy non-pregnant women. Sandler et al. [9] reported a clearance of 0.95 l/h per kilogram for fentanyl infused in more than one dose by the epidural route for the control of post-thoracotomy pain. The difference between the present data and those reported in the literature can be explained by differences in the route of administration, number of doses administered, and population studied.

Vd/f and Cl/f undergo changes as a function of gravidic modifications during pregnancy. Loebstein et al. [10] reported that the Vd/f of some drugs may increase by as much as 50% during pregnancy as a result of the expansion of plasma volume and the presence of a new compartment represented by the fetus and placenta. Evaluation of C l /f shows that pregnancy promotes enzymatic induction secondary to the action of progesterone. There is also an increase in the renal elimination of the drug as a function of the increase in renal plasma flow (25–50%) and of the increase in glomerular filtration rate (50%).

It should be emphasized that during pregnancy the elimination half-life of most drugs is increased due to the increased volume of distribution and total clearance [6]. The median t½β and t½γ of fentanyl in the study group were 192.5 and 620 min, respectively (Table 2). Leveque et al. [11] reported an elimination half-life of 460 min for fentanyl administered in multiple doses to pregnant women submitted to epidural labor analgesia.

In the present study, all patients delivered at term, placental weight was homogeneous, and no significant macroscopic changes that might alter drug transport were observed [12, 13]. The fetal hepatic metabolism, the effect of progressive dilutions in the fetal circulation, and the presence of pathologies in the fetus may influence the pharmacodynamic and pharmacokinetic processes of the drug [14]. However, in our study the influence of these variables was eliminated by the fact that the blood used for the analysis of placental fentanyl transfer was obtained from an umbilical vein. In addition, no pathologies were detected in the newborns during postnatal evolution.

The median latency time between drug administration and birth was 28.5 min, and at the time of delivery the median maternal plasma concentration was 0.31 ng/ml and the median fetal concentration was 0.25 ng/ml, with a median fetal/maternal ratio of 0.89 (Table 3). Desprats et al. [15] reported fentanyl concentrations of 0.13±0.04 ng/ml for the umbilical vein, large differences between the umbilical arterial and venous concentration of fentanyl, and a fetus extraction ratio of 53±19%. We obtained fetal plasma concentrations that were about 90% of the maternal fentanyl concentration at the time of delivery. This result should serve as an alert as this is a significant transfer of the drug through the placental barrier. At high doses the drug may act on the fetus to cause postnatal respiratory depression. This complication did not occur in the group studied, as demonstrated by the high Apgar scores for all of the newborn infants. However, it should be emphasized that high doses of this drug should be administered with caution to pregnant women.


The data obtained in the present study and analyzed by the methods described led us to conclude that fentanyl passed rapidly from the epidural space, a richly vascularized area, to the bloodstream, with only a short period of time for equilibration between these sectors. This conclusion is supported by the fact that the drug was detected in the plasma of the first blood collection (1 min after epidural administration) and by the impossibility of determining the absorption curve. The elimination half-life of fentanyl observed in the present study (192.5 min) is in agreed with data reported in the literature with respect to both a single intravenous infusion and administration by the epidural route to non-pregnant women. The pharmacokinetic parameters of clearance and volume of distribution corrected for body weight following epidural administration to parturients were 6.9 ml/min per kilogram and 4.4 l/kg, respectively.

Placental transfer of fentanyl occurs – with a median fetal/maternal ratio of 0.89 at maternal plasma concentrations of 0.31 ng/ml at the time of delivery – indicating the need for caution regarding the doses of the drug administered in order to prevent deleterious effects on the fetus and/or the newborn infant.


  1. 1.
    Reynolds FG, Taylor C (1970) Maternal and neonatal blood concentrations of bupivacaine. A comparison with lidocaine during continuous epidural analgesia. Anesthesia 25:14–23Google Scholar
  2. 2.
    Gutstein HB, Akil H (2001) Opioid analgesics. In: Gilman AG, Limbird LE, Hardman JG (eds) Goodman and Gilman’s The pharmacological basis of therapeutics, 10th edn. McGraw-Hill, New York, pp 569–619Google Scholar
  3. 3.
    Willens JS, Myslinski NR (1993) Pharmacodynamics, pharmacokinetics, and clinical uses of fentanyl, sufentanil, and alfentanil. Heart Lung 22:239–251Google Scholar
  4. 4.
    Hodgksisson R, Husain S (1980) Obesity and cephalad spread of analgesia following epidural administration of bupivacaine for cesarean section. Anesth Analg 59:89–92Google Scholar
  5. 5.
    Justins DM, Knott C, Luthman J, Reynolds F (1983) Epidural versus intramuscular fentanyl. Analgesia and pharmacokinetics in labour. Anaesthesia 38:937–942Google Scholar
  6. 6.
    Little BB (1999) Pharmacokinetics during pregnancy: evidence-based maternal dose formulation. Obstet Gynecol 93:858–868Google Scholar
  7. 7.
    Scholz J, Steinfath M, Schulz M (1996) Clinical pharmacokinetics of alfentanil, fentanyl and sufentanil. An update. Clin Pharmacokinet 31:275–292Google Scholar
  8. 8.
    Tegeder I, Lotsch J, Geisslinger G (1999) Pharmacokinetics of opioids in liver disease. Clin Pharmacokinet 37:17–40Google Scholar
  9. 9.
    Sandler AN, Stringer D, Panos L, Badner N, Friedlander M, Koren G, Katz J, Klein J (1992) A randomized, double-blind comparison of lumbar epidural and intravenous fentanyl infusions for postthoracotomy pain relief. Analgesic, pharmacokinetic, and respiratory effects. Anesthesiology 77:626–634Google Scholar
  10. 10.
    Loebstein R, Lalkin A, Koren G (1997) Pharmacokinetic changes during pregnancy and their clinical relevance. Clin Pharmacokinet 33:328–343Google Scholar
  11. 11.
    Leveque C, Garen C, Pathier D, Mazuir E, Maneglia R, Janse-Marec J, Cousin MT (1987) Fentanyl in peridural obstetrical analgesia Evaluation after 4 years’ use. J Gynecol Obstet Biol Reprod (Paris) 16:113–121Google Scholar
  12. 12.
    Santos AC, O’Gorman DA, Finster M (2001) Obstetric anesthesia. In: Barash PG, Cullen BF, Stoelting RK (eds) Clinical anesthesia, 4th edn. Lippincott Williams & Wilkins, PhiladelphiaGoogle Scholar
  13. 13.
    Pasanen M, Pelkonen O (1990) Human placental xenobiotic and steroid biotransformations catalyzed by cytochrome P450, epoxide hydrolase, and glutathione s-transferase activities and their relationships to maternal cigarette smoking. Drug Metab Rev 21:427–446Google Scholar
  14. 14.
    Mathias RS, Torres MLA (2000) Analgesia e anestesia: Técnicas em obstetrícia. In: Neme B (ed) Obstetrícia básica, 2nd edn. Sarvier, São Paulo, Brasil, pp 1035Google Scholar
  15. 15.
    Desprats R, Dumas JC, Giroux M, Campistron G, Faure F, Teixeira MG, Grandjean H, Houin G, Pontonnier G (1991) Maternal and umbilical cord concentrations of fentanyl after epidural analgesia for cesarean section. Eur J Obstet Gynecol Reprod Biol 42(2):89–94Google Scholar

Copyright information

© Springer-Verlag 2005

Authors and Affiliations

  • Elaine Christine Dantas Moisés
    • 1
  • Luciana de Barros Duarte
    • 1
  • Ricardo de Carvalho Cavalli
    • 1
  • Vera Lúcia Lanchote
    • 2
  • Geraldo Duarte
    • 1
  • Sérgio Pereira da Cunha
    • 1
    • 3
    Email author
  1. 1.Department of Gynecology and Obstetrics, Faculty of Medicine of Ribeirão Preto, University HospitalUniversity of São PauloRibeirão PretoBrazil
  2. 2.Department of Clinical Analyses, Toxicology and Food Sciences, Faculty of Pharmaceutical Sciences of Ribeirão PretoUniversity of São PauloRibeirão PretoBrazil
  3. 3.Departamento de Ginecologia e ObstetríciaHospital das Clínicas-8 andarRibeirão PretoBrasil

Personalised recommendations