This study aimed to estimate the extent and distribution of entry of lithium into the brain during development from E18 to P17 compared to the adult, following administration of clinically revevant doses of lithium either to the pregant and lactating mothers or directly into the postnatal animals. This approach allows determination of the degree of protection provided by the placenta and breast tissue against maternally administered lithium as well as the protection provided by blood brain barriers during rat development. Acute (single dose i.p.) experiments were used to study directly the entry of lithium across the blood-brain and blood-CSF barriers in adult and developing animals. For technical reasons only i.p. injections were use in the E19 experiments. In longer-term treatment experiments injections (i.p.) were only made into the pregnant and lactating mothers. This mimicked the clinical situation of continuous treatment and allowed estimation of the role of placental and breast tissue protection. Samples of blood plasma, CSF and brain were collected from fetuses, postnatal animals and dams for analysis of lithium distribution between the dam, fetuses and pups and between blood and tissues at different stages of development. The period of treatment (E15-E19) is about 25% of the gestational period of the rat (21 days). E19 is approximately equivalent to 22–24 weeks gestation in humans (43. Clancy et al., 2001; 44. Workman et al 2013), corresponding to the earliest age of viability (45. Fischer et al., 2009; 46. Stoll et al., 2010).
Transfer of maternally administered lithium across the placenta
The extent of protection provided by placenta against lithium entry into the fetus was assessed from lithium concentration ratios between fetal and maternal plasma. In E18 pups from an acutely injected dam, lithium concentrations in the fetal plasma at 100-250 min were lower than the dam’s plasma, with fetal/maternal ratios between 15-30%. In E18 pups from a pregnant dam receiving daily injections of lithium from E15-E18, the fetal/maternal concentration ratios were between 20-35% within same timeframe (Figure 8). Thus rat placenta appears to significantly restrict lithium entry into fetal blood late in gestation by about 70%.
There have been few studies of placental transfer of lithium in humans. In ten patients with bipolar disorder treated with lithium throughout pregnancy, on delivery in late gestation fetal/maternal concentration ratios were around 100% (47. Newport et al. 2005). In 10 patients in a region where drinking water contained high levels of lithium (NW Argentina) in infants born at 39-weeks’ gestation the cord/maternal plasma ratios averaged about 150% (48. Harari et al 2015). In another study (49. Krachler et al. 1999) a large number of naturally occurring elements, the cord/maternal ratios for lithium showed wide variation with a mean of about 80% (n=29). These findings perhaps suggest that in the case of lithium therapy or ingestion of water with high naturally occurring levels of lithium there may be unrestricted transfer between maternal and fetal blood in humans. This contrasts with the much lower ratios observed in the present study. Factors that might contribute to this difference could include differences in where blood was sampled (cord blood in humans and heart right ventricle in fetal rats), different lengths of exposure (4 days in the rats compared to continuing lithium treatment during pregnancy in humans) and species-specific differences in placental structure. The rat and human placentas are both classed as hemochorial (50. Blood et al., 2007; 51. Dawe et al., 2007) but there are differences in morphology, in particular that the rat placenta has more morphological layers between the fetal and maternal circulations. Also, there may be age-dependent differences in placental function. This is suggested by a single case example of a fetus delivered from a mother with lithium toxicity by caesarean section at 28-weeks’ gestation where the cord/maternal blood ratio was 86% (52. Zamani et al 2017).
Transfer of maternally administered lithium across breast tissue and milk
Transfer of drugs through breast milk can be affected by a number of factors such as characteristics of the drug itself: molecular size, degree of ionization, lipid solubility and for many drugs but not lithium, extent of plasma protein binding (53. Newton & Hale, 2015) and maternal factors such as maternal plasma concentrations and pharmacogenomics (54. Hotham and Hotham, 2015). Lithium concentrations in human colostrum and breast milk have been found to be around 50% of the maternal blood concentration (49. Krachler et al. 1999; 15, Viguera et al. 2007; 53. Newton & Hale, 2015) although large variations were observed (30-70%, 15. Viguera et al. 2007; 49. Krachler et al. 1999). These studies also reported a wide range of lithium concentrations in infant serum. These variations perhaps relate to the time of breast feeding compared to the time of ingestion of lithium by the mother, whether as medication or in drinking water.
A key finding in the present study was detection of lithium in the plasma of pups from lactating dams administered long-term lithium therapy. Compared to pups directly injected with lithium (3.2 mg lithium/Kg i.p.) and sampled at steady-state (90-120 min), plasma and CSF lithium concentrations in the breastfed pups were much lower (~0.4 mM vs ~0.1 mM in plasma and ~0.15 mM vs ~0.05 mM in CSF respectively, Figures 5 and 6). Concentrations of lithium in breast milk were not measured in this study, nor how much milk was consumed by each pup, so it was not possible to quantitate the extent of protection provided by breast tissue.
Mechanism of lithium entry into the CSF
Due to its very small size (hydrodynamic radius 0.079 nm, 55. Mähler and Persson, 2011) lithium would be predicted to enter the CSF from blood either by passive diffusion or by transfer mechanisms via ion channels or exchange transporters where Li+ substitutes for Na+, see Introduction.
Previous studies of blood to CSF transfer have suggested that entry of molecules is dependent on molecular size, lipid solubility and stage of development (age) with smaller molecular radius and younger age correlating with higher apparent rates of entry. This relationship has been demonstrated in a number of different species (42. Saunders, 1992) including the rat (36. Habgood et al., 1993). However, the level of lipid insoluble molecules is heavily dependent on the turnover of CSF, which is much less in the developing brain. Thus, the much higher levels of these molecules in brain and CSF early in development, should better be referred to as an index of “apparent” permeability (24. Saunders et al. 2018). The smallest molecular marker previously investigated in rat was L-glucose (molecular size 180Da, molecular radius 0.43nm) which is much larger than the 0.079nm hydrodynamic radius of lithium (55. Mähler and Persson, 2011). Nevertheless, if lithium predominately enters CSF by passive diffusion, it would be expected that its CSF/plasma concentration ratio would fall on the correlation lines predicted by other passively transferred markers (36. Habgood et al., 1993).
In the acute exposure experiments, lithium CSF/ Plasma ratios fell below the levels predicted for passive markers across all ages studied (Figure 10A). In contrast, CSF/ Plasma ratios from the steady-state (long-term) lithium exposure experiments did fall on the lines predicted for passive permeability (Figure 10B) at each stage of development investigated (R2 values of 0.97, 0.99 for P2 and P16-20 respectively).
The much lower apparent entry of lithium into CSF in acute experiments, compared to steady-state and compared to other markers, is inconsistent with transfer by simple passive diffusion. The CSF/ Plasma ratios for the passive markers used to compare with the results of present study were all measured at steady-state (i.e. approaching the equilibrium between the rates of entry into and out of CSF). Times to reach steady-state for L-glucose, sucrose and inulin were between 4-5 h. Based on the very small molecular size of lithium it would be expected to approach steady-state within 1.5- 2 h, but the ratios reached were well below those predicted (Figure 10A).
There are numerous ion channels exchangers in choroid plexus epithelial cells (33. Damkier et al., 2013). Many of these are expressed in immature rat choroid plexus (34, 35. Liddelow et al., 2013; 2016). However, their permeability to lithium appears not o have been investigated.
The much lower CSF/ Plasma ratios for lithium in the acute experiments could perhaps be explained by a restriction on passive entry at the blood/CSF barrier interface. It seems unlikely that Na+/K+ ATPase activity (56. Naylor et al., 2016) was involved as inhibition with a large dose of digoxin had only a marginal effect on the entry of lithium into CSF (Figure 9). Consistent with this is the finding that Na+/K+ ATPase activity is low in fetal and newborn rats (57. Johansson et al 2008a)
In the long-term treated animals, lithium reached the predicted steady state ratios for the age groups studied (Figure 10B) presumably reflecting the much longer period of exposure.
Steady-state CSF/ Plasma ratios in developing rats for sodium were reported to be just below 100% in the age range of our study (58. Amtorp and Sørensen, 1974).
In the long-term treated animals there was an age-dependent decrease in entry of lithium into CSF; steady-state CSF lithium concentrations decreased from 0.2 mM at E18 to 0.07 mM at P2 and 0.009 mM at P16 (Figure 5B). Consistent with this were age-dependent decreases in CSF/plasma ratios falling from 54-92% at P2-4 down to 27-56% at P12-16. These age-related decreases are consistent with developmental increases in the rate of CSF turnover (CSF sink effect) with increasing age (59. Johansson et al., 2008b).
Entry and distribution of lithium in developing brain
The concentration of lithium in brains of treated dams, both acute and long-term, were similar (0.45 and 0.40 mM respectively, Table 2). These are consistent with previous studies (60. Smith, 1976; 61. Wraae, 1978), and also indicate there is no accumulation of lithium in brain with long-term exposure. Pups in the acute treatment group had more variable concentrations in brain homogenates, but were on average similar to the dams, mirroring concentrations in plasma. For pups receiving lithium via breast milk, concentrations of lithium in brain homogenates were much lower than in the dam, but these animals also had lower plasma concentrations (Table 2, Figure 4). This is consistent with the suggestion that the concentration of lithium in brain is a function of the concentration in plasma (Figure 5).
The Brain Homogenate/ Plasma ratios (Figure 6) in pups from acute experiments and long-term treatments had ratios of around 100% and 150% respectively which is much higher than CSF/ Plasma ratios in the same animals (around 35%, Figure 4C and <90%, Figure 5C respectively). This suggests a faster rate of lithium excretion from CSF. Previous reports have described greater rates of lithium loss from CSF compared to brain tissue (61. Wraae 1978).
The distribution of lithium within the brain has been investigated following lithium exposure using various imaging techniques. Sandner et al (22. 1994) used 6Li(n.⍺)3H nuclear reaction in the presence of a dielectric a particle track detector to image lithium in adult rat brain, but the resolution was poor. Lithium‐7 nuclear magnetic resonance (7Li‐NMR, 23. Stout et al., 2017) and 3D 7Li magnetic resonance (62. Smith et al 2018) are non-invasive in vivo methods that allow low resolution brain mapping of lithium and regional quantitation of lithium levels in patients.
In the present study, LA-ICP-MS was used as it provides a higher resolution of anatomical features and patterns of distribution compared to MRI. Brain sections from this study revealed even distribution of lithium within the brain, but with notable accumulation in the olfactory lobes in all treated animals (Table 2). Preferential accumulation of other metal ions in the olfactory lobe has been reported in human and rats for manganese, aluminium, nickel, zinc and cobalt (63. Bonilla et al., 1982; 64. Henriksson et al. 1999; 65. Fechter et al., 2002; 66. Chalansonnet et al. 2018; 67. Calderón-Garcidueñas et al. 2013; 68. 69. Perrson et al. 2003a, 2003b). High resolution ion imaging in the rat have provided images showing accumulation of lithium in the frontal lobe of the brain (70. Zanni, 2017). Accumulation of lithium in the olfactory lobe of rats detected in the present study could be of clinical relevance as loss of smell (hyposmia) and altered taste sensation (dysgeusia) in patients on lithium treatment (71. de Coo and Haan, 2016; 72. Terao et al., 2011) have been reported.
Additional observations that are outside the scope of this study of lithium entry into the developing brain were that there was a prominent increase in the number of neutrophils in the blood of animals exposed to lithium (see Additional file 4) which is consistent with leucocytosis following lithium therapy described in adults and children (73. Schou et al., 1970; 74. Chan et al., 1981; 75. Ishii et al., 1983). Another known side effect in patients on lithium therapy is weight gain (76. Kerry et al., 1970). This was evident in pups chronically exposed to lithium via breast milk (Additional file 4).