The Blood-Brain Barrier

by Stephen Western

One of the major factors making tumours of the brain so challenging to treat with systemic chemotherapy via the bloodstream is the existence of the blood-brain barrier, which is designed to protect the brain from harm by foreign elements. Unlike most capillaries in the body, which are relatively permeable, the capillaries serving the brain tightly restrict the passage of many solutes from the blood into the brain tissue. This blood-brain barrier consists of very tight junctions between the endothelial cells making up the capillary wall, and footlike “processes” extending from astrocytes (structural brain cells) which attach to the capillaries and form a second physical barrier. Thirdly, a family of active drug efflux pumps residing in the capillary endothelial cells efficiently eject back into the bloodstream many drugs that manage to enter these cells.

Tumour disruption of the blood-brain barrier

Tumour angiogenesis, or the growth of new blood vessels to feed the tumour, is associated with leaky capillaries, and a disrupted blood-brain barrier. MRI scans basically show the degree to which the tumour has disrupted the blood-brain barrier. The most intense areas of contrast enhancement on MRI scans correlate with the areas of greatest blood-brain barrier disruption. Higher grade gliomas, especially glioblastomas, have the greatest degree of angiogenesis and the greatest disruption of the blood-brain barrier. This provides an advantage when it comes to systemic chemotherapy via the bloodstream, as well as immunotherapy - immune effector cells and chemotherapeutic agents can pass through the disrupted blood-brain barrier far more easily. This may explain why dendritic cell immunotherapy is more effective in the more highly angiogenic mesenchymal subtype of glioblastoma than the less angiogenic proneural subtype. However, in any tumour there are invasive cells to be found outside the main tumour area, where the blood-brain barrier is still intact. Targeting of these invasive cells may require a strategy of blood-brain barrier disabling, such as by the inhibition of drug efflux pumps, called ABC transporters.

ATP-Binding Cassette (ABC) Transporters

ATP-binding cassette (ABC) transporters are proteins which use the energy from ATP hydrolysis to actively pump various solutes and drugs across cell membranes. ABC transporters may be found in cells of the intestines, liver, kidney as well as brain capillary endothelial cells where they form part of the blood-brain barrier. The three major ABC transporters which contribute to drug resistance are P-glycoprotein (ABCB1 or multidrug resistance protein 1/MDR1), multidrug resistance-associated protein 1 (MRP1 or ABCC1) and breast cancer resistance protein (BCRP or ABCG2).

It has been demonstrated that BCRP/ABCG2 is expressed not only in capillary endothelial cells at the blood-brain barrier, but also in stem-like cells in the tumour itself, making these cells particularly resistant to many chemotherapeutic agents (1).

Drug substrates of P-glycoprotein and Breast cancer resistance protein

The following drugs are known to be substrates for efflux by P-glycoprotein. Bioavailability and brain penetration of these drugs might therefore be increased by pharmacological P-glycoprotein inhibition.


One study (13) noted a 4-fold increase in the brain:plasma ratio (at 2 hours) of orally administered dasatinib in P-gp (Mdr1a/1b) knockout mice relative to controls. Additionally, there was a 9-fold increase in the brain:plasma ratio of dasatinib in Mdr1a/1b/BCRP1 triple knockout mice relative to controls. The brain penetration of dasatinib would likely be much improved by a combined targeting of P-gp and BCRP.


Gefitinib is an FDA-approved EGFR inhbitor which has been tested in clinical trials for malignant glioma. A similar study (14) by the same group as study (13) found that gefitinib is also a substrate of P-gp and BCRP at the blood-brain barrier in mice. At 90 minutes after oral administration of gefitinib, the brain:plasma ratio of gefitinib was increased 4-fold in P-gp (Mdr1a/b) knockout mice relative to controls, and the same ratio was increased 18-fold in Mdr1a/1b/BCRP triple knockout mice. Notably, plasma concentrations of gefitinib were not significantly different in triple knockout mice compared to controls, showing that gefitinib efflux by P-gp and BCRP mainly occurs at the blood-brain barrier, rather than in the intestines or liver.


Irinotecan is a significant substrate for P-glycoprotein efflux. In one study (10), the brain to plasma AUC (area-under-the-curve, 0-8 hours) ratio of SN-38 (the active metabolite of irinotecan) was increased 2.4-fold in mdr1a knockout mice compared to mdr1a normal mice. P-glycoprotein inhibition could therefore significantly increase the brain penetration of intravenous irinotecan.


Paclitaxel is a major substrate of P-gp, with brain concentrations of the drug increased 9-fold at four hours, and 22-fold at 24 hours, in P-gp knockout mice compared to normal mice (11).


Wild-type, Abc1a/1b knockout, Abcg2 knockout, and Abc1a/1b/Abcg2 triple knockout mice were treated with oral sunitinib to determine plasma pharmacokinetics and brain penetration of the drug in each group (15). The plasma AUC (area under the curve, 0-6 hours) was not significantly different between the groups. In contrast, the brain concentration of sunitinib at 6 hours was significantly increased in Abc1a/1b knockout mice (2.3-fold) and Abcb1a/1b/Abcg2 triple knockout mice (23.4-fold). Similarly, co-administration of oral elacridar (a potent ABCB1 and BCRP inhibitor) followed by oral sunitinib to wild-type mice did not change the plasma concentration of sunitinib at one hour, but very significantly increased brain concentrations of the drug, to a similar extent as seen with the triple-knockout mice. These experiments show that orally administered sunitinib is a substrate for both P-gp (ABCB1) and BCRP at the blood brain barrier, and that both of these drug pumps must be inhibited for maximum brain penetration of sunitinib. Inhibiting P-gp and BCRP does not change drug plasma levels in mice.

Tamoxifen and metabolites

Tamoxifen and especially its active metabolite endoxifen are sustrates for P-glycoprotein efflux. In another experiment (12) with P-glycoprotein (Abcb1a/1b) knockout mice, the brain:serum concentration ratio of tamoxifen was increased by 60% at four hours in the knockout mice versus controls. Endoxifen, the main active metabolite of tamoxifen (in terms of estrogen receptor inhibition) is an even more sensitive substrate of P-gp efflux. The brain:serum concentration ratio of endoxifen was increased 6.4-fold at 4 hours in the knockout mice versus control mice.


As shown recently in a study (10) with P-glycoprotein (mdr1a) knockout mice, temozolomide is a minor substrate of P-glycoprotein. The brain to plasma area-under-the-curve (AUC, 0-4 hours) ratio was increased by only 10% in the mdr1a knockout mice compared to the mdr1a normal mice. P-glycoprotein inhibition is therefore not likely to make a major difference to TMZ bioavailability and brain penetration. Additionally the absolute bioavailability of TMZ is reported as 96-100%, showing that TMZ is not a significant substrate of efflux transporters in the intestine or liver.

Increasing drug transport beyond the Blood-brain barrier

Improving brain/tumour penetration of drugs by P-glycoprotein and BCRP inhibition

P-glycoprotein (P-gp) and other ABC transporters are not only active at the blood-brain barrier, but also in the enterocytes of the intestines and the hepatocytes of the liver. Disabling P-gp to increase bioavailability of oral drugs could lead to increased systemic exposure and potentially toxic side effects due to overdosing. However, experiments with Abcb1 and Abcg2 knockout mice (14, 15) have shown that plasma levels of some drugs (such as gefitinib and sunitinib) are unaltered, while brain concentrations are dramatically increased in the knockout mice, indicating that these ABC transporters are primarily active at the blood-brain barrier, rather than at the intestine and liver. Disabling P-gp and/or BCRP to increase the brain exposure of intravenously administered drugs might avoid the risk of increased systemic drug exposure and side-effects, as this mode of administration bypasses first and second pass metabolism in the intestines and liver, and iv drug dosing reflects this. Depending on the drug, disabling drug efflux pumps at the blood-brain barrier could lead to increased drug penetration to the healthy brain and central nervous system and cause unwanted neurotoxicity. A cautious approach should therefore be taken when implementing a strategy of inhibiting ABC transporters, as the risk of unwanted side-effects may be increased.

Therapeutic drugs may be ejected by one or more of these ABC transporters. In some cases, P-glycoprotein and BCRP co-operate to remove a drug from the cell, pumping it back into the bloodstream. If one of these transporters is pharmacologically inhibited, the other will pick up the slack. Therefore, the most efficient drug penetration across the blood-brain barrier will often require a dual inhibition of more than one ABC transporter (1).

Fluoxetine (Prozac) and sertraline (Zoloft) as efflux pump inhibitors

A preclinical mouse study (8) published in Cancer Research in 2004 by researchers at the Department of Biochemistry, Tel Aviv University, shows that the common antidepressant drug Prozac (fluoxetine) may be a potent chemosensitizer. The mechanism of action appears to be inhibition of drug efflux pumps in chemoresistant cell lines. Notably, the addition of fluoxetine to various chemotherapeutic agents had little effect in chemosensitive cancer cell lines. In contrast, when fluoxetine was added to the same agents, a remarkable sensitization was apparent in chemoresistant cell lines, with an up to 70-fold increase in drug sensitivity.

In vitro, equimolar concentrations of verapamil (the historical benchmark drug for inhibiting drug efflux pumps) and fluoxetine added to paclitaxel in drug-resistant breast cancer cell cultures demonstrate that fluoxetine was far more potent than verapamil as a chemosensitizer in this particular model. Fluoxetine was effective at concentrations well below plasma levels found in humans taking Prozac as an anti-depressant, while concentrations equal to those plasma levels were even more effective. Doxorubicin chemotherapy efflux from chemoresistant cancer cells was also monitored over a 12 hour time period. The cells were also treated with either verapamil, cyclosporin A (both benchmark chemosensitizers), fluoxetine, or no chemosensitizer. 5 micromolar was the concentration used for each chemosensitizer. Verapamil and cyclosporin A were mildly effective in slowing doxorubicin efflux from the cancer cells, while the effect of fluoxetine was dramatic. The two former chemosensitizers had lost efficacy after 3-5 hours, while the fluoxetine delayed doxorubicin efflux for nearly 12 hours.

The efficacy of fluoxetine as a chemosensitizer was then tested in vivo. Mice bearing chemoresistant lung tumour grafts were challenged with the chemotherapy drug doxorubicin alone, or fluoxetine alone (applied to the drinking water), or both drugs combined. While either drug alone had virtually no effect on overall survival, the two drugs combined dramatically increased the survival time of the mice, with two of the mice surviving to at least 70 days. In contrast, all the mice treated with either drug alone, or without any treatment, were all dead by day 20-25. The same effect was observed in another model using chemoresistant P388-ADR ascites tumours. Fluoxetine had no effect when added to doxorubicin in P388-WT cells which had not been bred for chemoresistance. These tests show that the primary effect of fluoxetine at this dose was to reverse cancer cell resistance to chemotherapy with doxorubicin. Most remarkable was the miniscule dose of fluoxetine required to achieve this effect in vivo. When drugs are tested in tumour-bearing mice, doses of 10-50 mg of drug per kilogram body weight are common. In this study, the dose of 0.04 mg fluoxetine/kg body weight was applied to the daily ration of drinking water. This would amount to 2.4 mg of drug per 60 kg before conversion to human equivalent dosing. After converting to human equivalent dosing, the dose would be around 0.2-0.25 mg of drug for a 60 kg human, in other words at least 100 less than the standard dose of 20-80 mg per day. Also remarkable was the ability of fluoxetine to selectively increase drug uptake into otherwise chemoresistant tumours, without increasing drug concentrations in spleen, liver, or kidney. A later review (9) identified ABCB1 (P-glycoprotein) and ABCC1 as targets for inhibition by fluoxetine. The effect of fluoxetine on ABCG2 (breast cancer resistance protein) was not yet determined.

Another study (16) by the same group tested either low-dose fluoxetine, doxorubicin, or a combination of the two drugs in a colon cancer xenograft mouse model. While fluoxetine alone had no activity, and chemotherapy with doxorubicin was only slightly effective, the combination of the two drugs significantly inhibited tumour volume. The oral dosage used in this study would equate to about 5 or 6 mg for an adult human, well below the clinical antidepressant dose of 20-80mg per day. The fluoxetine dose in this study was delivered as a single daily oral gavage, whereas the drug was added to the drinking water in the previous study.

A third study (17), again by the same group from Tel Aviv University, was published in a November 2014 edition of Cancer Letters. In this study, the investigators chose to look at sertraline, which is like fluoxetine a selective serotonin reuptake inhibitor (SSRI) antidepressant. Sertraline was compared with fluoxetine and verapamil as an addition to doxorubicin applied to chemoresistant ovarian cancer cells. A 15 micromolar concentration of either sertraline, fluoxetine or verapamil had no effect on the cancer cell line. A 10 micromolar concentration of doxorubicin inhibited cell survival only slightly. When added to doxorubicin, the three drugs sertraline, fluoxetine and verapamil each proved to be effective chemosensitizers, with sertraline being the most potent, and fluoxetine being the least potent of the three. The three drugs were shown to improve chemosensitivity to doxorubicin by slowing doxorubicin efflux out of the tumour cells.

Sertraline was then tested in a xenograft mouse model of chemoresistant ovarian cancer. Sertraline as a single agent had no activity, while doxorubicin alone had modest effect. When the combination of sertraline and doxorubicin was applied to the mice, tumour volume was significantly reduced and mouse survival was prolonged compared with doxorubicin alone. Further studies were then conducted adding sertraline to Doxil® (pegylated liposomal doxorubicin). The oral dose of sertraline used in these studies equates to a dose of around 10 mg per day for an adult human, far less than the standard antidepressant dose of 50-150 mg per day.

The reversal of chemoresistance by fluoxetine and sertraline in vitro and in vivo at doses far less than those used by humans taking the drug as an antidepressant seems quite remarkable. I propose that low-dose Zoloft (sertraline) and Prozac (fluoxetine) be studied as a means of increasing the entry of specified chemotherapeutic agents into glioma tumours. This is an especially relevant strategy for chemotherapy agents which are known to be sensitive substrates of ABC drug transporters.

Fluoxetine: additional effects on barrier permeability
A preclinical study (18) published in 2014 demonstrated that administration of fluoxetine in the drinking water to adult nude mice for 3 weeks increased the number of brain metastases following injection of human breast cancer cells (MDA-MB-231BR, a subclone which selectively metastasizes to brain). This increase in the number of brain metastases was attributed to an increase in blood-brain permeability caused by fluoxetine. Increased barrier permeability was confirmed by tail vein injection of Evans Blue dye, which resulted in a 54% increase in dye absorbance compared to the untreated control group. Evans blue dye is mostly bound to serum albumin and does not normally cross the blood-brain barrier. This study suggests that fluoxetine may open the blood-brain barrier by a mechanism independent of ABC transporters.


Verapamil is a calcium channel blocker which dilates blood vessels and is therefore used to treat conditions such as hypertension and angina pectoris. It is also a non-specific ABC inhibitor, with a more potent inhibition of ABCB1 (P-glycoprotein) and ABCC1. It was the first drug used to experimentally combat multidrug resistance in cancer therapy (2).


Morphine is frequently prescribed as a pain-killer to cancer patients. A letter to the editor published in the Journal of Oncology describes an additional effect of the drug (4). Rats given doses of the chemotherapy drug doxorubicin with morphine had an accumulation of doxorubicin in the brain about 3 times higher than in rats given doxorubicin without morphine. There was no increase in doxorubicin accumulation in heart and kidney tissues with the addition of morphine. This demonstrates that in rats, morphine opens the blood-brain barrier to a drug such as doxorubicin, which normally has very limited blood-brain barrier penetration. This letter does not indicate the mechanism behind these effects of morphine, and these effects may be due to a mechanism other than inhibition of ABC transporters.

Increasing permeability of the blood-tumour barrier by PDE5 inhibition

Viagra (Sildenafil) and Levitra (vardenafil)

The search for drugs which might reverse multi-drug resistance or allow the penetration of chemotherapy drugs through the blood-brain barrier has gone on for decades. First generation agents such as verapamil were found to have serious adverse effects at doses required for efficacy. More recently, third generation agents such as elacridar have similarly lacked efficacy in later phase trials (5). Thus, the search for effective and well-tolerated inhibitors of multidrug resistance proteins such as P-glycoprotein and other ABC transporters continues.

In 2008, Keith Black and colleagues at Cedars-Sinai Medical Center published a study (6) showing that the clinically approved drugs sildenafil (Viagra) and vardenafil (Levitra), inhibitors of phosphodiesterase type 5 - PDE5- and approved for use in male erectile dysfunction, may have an important new role in increasing the transport of chemotherapy drugs through the blood-tumour barrier.

The target molecule of sildenafil and vardenafil is phosphodiesterase type 5 (PDE5), an enzyme which acts on and degrades cyclic guanosine monophosphate (cGMP). cGMP regulates vascular tone and permeability. By binding to and inhibiting PDE5, drugs such as sildenafil increase intracellular cGMP levels, thereby increasing the permeability of blood vessels, potentially including the capillaries feeding brain tumours. In the Cedars-Sinai study, high levels of PDE5 were found in the 9L, GL26, U87 and RG2 glioma cell lines, as well as in human brain tumour samples and a human endothelial cell line.

Two sets of experiments were performed on rats bearing intracranial 9L (gliosarcoma) tumours. First, the blood-tumour barrier permeability was determined following oral treatment with sildenafil or vardenafil. Optimal doses of the drugs were determined and both drugs were found to significantly increase blood-tumour barrier permeability with maximum effect at 60-75 minutes after oral drug administration.

Secondly, survival studies determined the ability of vardenafil (Levitra) to increase survival of tumour-bearing rats when administered together with adriamycin, a chemotherapeutic agent. Starting at day 4, Fischer rats bearing intracranial 9L tumours were given vardenafil orally, followed by tail vein injection with adriamycin. Treatments were given for 3 days. The mean survival of the rats on adriamycin alone was 42 days, and the addition of vardenafil increased the mean to 53 days (26% increase). Tumour size was also smaller when vardenafil was given with adriamycin, compared to adriamycin alone. There were no detectable side-effects attributed to vardenafil. Mean arterial blood-pressure was reduced by 10% by sildenafil and vardenafil. It must be noted that the optimal doses found in this study were about 5 times higher than the maximum recommended human dose, after converting to the human equivalent dose from the actual dose applied to the rats. The maximum recommended daily dose is 100mg (sildenafil) and 20mg (vardenafil).

A later in vitro study (7) was published by a different group in 2011 showing that a 2 micromolar concentration of sildenafil effectively reversed chemoresistance to the ABCB1 (P-glycoprotein) substrates colchicine, vinblastine and paclitaxel, demonstrating effective P-glycoprotein-inhibiting ability. This 2 micromolar concentration was achieved in human plasma after a single 200 mg dose of the drug, or roughly twice the maximum recommended dose. Sildenafil was less effective as an ABCG2 inhibitor, requiring at least 10 micromolar which is very unlikely to be clinically achievable.

A phase II trial is recruiting recurrent high grade glioma patients at Virginia Commonwealth University (Richmond VA) to test a combination of Viagra (sildenafil), sorafenib and valproic acid. The addition of sildenafil is expected to increase the permeability of the BBB to sorafenib. This trial is expected to be complete in 2017.


  1. Breast cancer resistance protein and P-glycoprotein in brain cancer: two gatekeepers team up. Agarwal et al. 2011.

  2. Use of verapamil to enhance the antiproliferative activity of BCNU in human glioma cells: an in vitro and in vivo study. Bowles et al. 1990.
    READ SOURCE DOCUMENT (abstract only)

  3. The molecular basis of the action of disulfiram as a modulator of the multidrug resistance-linked ATP binding cassette transporters MDRI (ABCB1) and MRP1 (ABCC1). Sauna et al. 2004.

  4. Morphine facilitates doxorubicin penetration in the central nervous system: a new prospect for therapy of brain tumors. Sardi. 2011.

  5. Roles of Sildenafil in Enhancing Drug Sensitivity in Cancer. Shi et al. 2011.

  6. PDE5 Inhibitors Enhance Tumor Permeability and Efficacy of Chemotherapy in a Rat Brain Tumor Model. Black et al. 2008.

  7. Sildenafil reverses ABCB1- and ABCG2-mediated chemotherapeutic drug resistance. Shi et al. 2011.

  8. Fluoxetine inhibits multidrug resistance extrusion pumps and enhances responses to chemotherapy in syngeneic and in human xenograft mouse tumor models. Peer et al. 2004.

  9. Fluoxetine and reversal of multidrug resistance. Peer et al. 2006.
    READ ABSTRACT Email me for a PDF copy

  10. Irinotecan and temozolomide brain distribution: a focus on ABCB1. Goldwirt et al. 2014.
    READ ABSTRACT Email me for a PDF copy

  11. Modulation of the blood–brain barrier in oncology: therapeutic opportunities for the treatment of brain tumours? Kemper et al. 2004.
    READ ABSTRACT Email me for a PDF copy

  12. P-Glycoprotein (ABCB1) Transports the Primary Active Tamoxifen Metabolites Endoxifen and 4-Hydroxytamoxifen and Restricts Their Brain Penetration. Iusuf et al. 2011.

  13. P-glycoprotein and Breast Cancer Resistance Protein Influence Brain Distribution of Dasatinib. Chen et al. 2009.

  14. Distribution of Gefitinib to the Brain Is Limited by P-glycoprotein (ABCB1) and Breast Cancer Resistance Protein (ABCG2)-Mediated Active Efflux. Agarwal et al. 2010.

  15. Brain accumulation of sunitinib is restricted by P-glycoprotein (ABCB1) and breast cancer resistance protein (ABCG2) and can be enhanced by oral elacridar and sunitinib coadministration. Tang et al. 2012.
    READ ABSTRACT Email me for a PDF copy

  16. Treatment of resistant human colon cancer xenografts by a fluoxetine-doxorubicin combination enhances therapeutic responses comparable to an aggressive bevacizumab regimen. Argov et al. 2009.
    READ ABSTRACT Email me for a PDF copy

  17. Modulating cancer multidrug resistance by sertraline in combination with a nanomedicine. Drinberg et al. 2014.
    READ ABSTRACT Email me for a PDF copy

  18. Fluoxetine modulates breast cancer metastasis to the brain in a murine model. Shapovalov et al. 2014.
    READ SOURCE DOCUMENT Email me for a PDF copy

This page was created on 12/28/2013 and last updated on 04/21/2019

Our privacy / cookie policy has changed.
Click HERE to read it!