Leukaemia follows a blood-vessel track to enter the nervous system

Certain cancers are prone to invade the nervous system, which leads to poorer prognosis. A study of leukaemia in mice reveals an unexpectedly direct invasion route from the bone marrow to the central nervous system.
Frank Winkler is in the Department of Neurology, University Hospital Heidelberg, and is also at the German Cancer Research Center, 69120 Heidelberg, Germany.

Search for this author in:

If malignant cells from solid or blood cancers enter the central nervous system (CNS) and grow there, the treatment options and clinical outlook deteriorate rapidly. In a type of leukaemia called acute lymphoblastic leukaemia (ALL), invasion of the CNS commonly occurs. To try to limit this, people with the condition often receive radiation or chemotherapy that targets the CNS. If more-effective and less-toxic approaches became available to prevent disease spread to the CNS, this might benefit many people with ALL. Writing in Nature, Yao et al.1 report a hitherto unknown route that ALL cells use to enter the CNS, and suggest a possible therapeutic approach that is worth investigating.

When leukaemia spreads into the CNS, this process, termed metastasis, is mainly limited to the region known as the subarachnoid space, which contains cerebrospinal fluid that bathes the brain and spinal cord. The subarachnoid space is surrounded by membranes called the dura mater, the pia mater and the arachnoid, which are collectively called the meninges and are also colonized by cancer cells (Fig. 1). Yao and colleagues used mouse models of ALL to investigate how human leukaemia cells spread into the CNS. They focused on the enzyme PI3K, which is a key regulator of signalling pathways needed for growth, survival and invasion by cancer cells.

Figure 1 | A route for cancer entry into the central nervous system. Cancer migration to the central nervous system (CNS, which comprises the brain and spinal cord) is often associated with poor prognosis. Yao et al.1 report that human blood-cancer cells in mice reach the central nervous system by migrating along the external surface of blood vessels. This solves the mystery of how these cells leave their site of origin in the bone marrow and reach a region called the subarachnoid space, which contains cerebrospinal fluid. This is located in a region termed the meninges, which covers the CNS and contains membrane layers called the dura mater, arachnoid and pia mater. The authors report that this migration process depends on an enzyme called PI3Kδ (not shown), and requires a receptor protein called α6 integrin on cancer cells. This receptor can bind to the protein laminin, which coats the surface of blood vessels and is also found in the meninges. Laminin aids the migration of neuronal cells during development2, so perhaps these cancer cells have hijacked components of a natural migration process.

The authors studied three different mouse models of ALL, and gave animals a molecule called GS-649443 that inhibits a version of PI3K called PI3Kδ. Animals that received the drug had reduced signs of cancer invasion of the CNS compared with those that did not receive it. The authors studied bone-marrow sites that are commonly rich in leukaemic cells, and found no evidence that the drug was affecting the growth or motility of leukaemic cells or cancer progression at sites outside the CNS. This result suggests that PI3K inhibition specifically affects the ability of leukaemic cells to enter the CNS — which is surprising, because the blood–brain barrier usually poses a formidable obstacle that restricts the ability of molecules or cells to exit blood vessels and enter the brain or the cerebrospinal fluid. Cancer cells that reach the CNS are likely to be beyond the reach of drugs present in the body’s bloodstream.

Yao and colleagues analysed gene expression in cancer cells to identify genes whose expression decreased as a result of PI3K inhibition. This pointed them towards the gene that encodes a receptor protein called α6 integrin. This receptor can bind a protein called laminin, which is a key component of the extracellular material that surrounds large blood vessels. Laminin is also a component of the meninges2 and is present in a CNS structure called the choroid plexus, which is where immune cells and cells from solid tumours often enter the CNS3.

When solid brain tumours arise from a cancer that has spread from elsewhere in the body, a common theme is tumour-cell entry into the brain by a process that requires signalling by a type of integrin protein called β1 integrin, which binds to laminin-containing blood vessels deep in the brain4,5. However, when Yao and colleagues analysed brain cells using microscopy techniques, they found no evidence that the choroid plexus or these vessels in the brain are places where ALL cells enter the CNS, confirming the results of previous human and mouse studies6.

Laminins and laminin-binding integrin receptors act in neuronal path-finding processes during development2, and are also key mediators of migration processes for healthy and cancerous cells5,7. Using microscopy approaches in their mouse models, Yao and colleagues discovered that small blood vessels coated with laminin that connect the bone marrow of the skull and the nearby meninges are sites with high levels of ALL cells. Transit of ALL cells along the external surface of these vessels could provide a direct route for the cells to reach the CNS. Previous studies6,8 have led to speculation about whether ALL cells spread to the CNS by a direct route, given that the outer layer of the meninges — the dura mater — can have a high level of infiltration by ALL cells.

Conducting studies in vitro, the authors found that cerebrospinal fluid contains chemokine molecules that can provide a chemical attractant for ALL cells. Yao and colleagues’ work also indicates that ALL-cell migration towards laminin depends on the presence of α6 integrin and can be blocked by PI3Kδ inhibition. The authors observed that treating mice with GS-649443 consistently diminished the invasion of ALL cells along the vascular corridors between the bone marrow and CNS. Moreover, giving animals an antibody that blocks α6 integrin prevented ALL spread to the CNS without having strong effects on ALL progression outside the CNS. Finally, the authors carried out a limited investigation of clinical samples from 26 people with ALL and found that a higher level of expression of α6 integrin correlated with a higher probability that a person’s cancer had invaded the CNS.

Interestingly, the ability of cancer cells to hijack pathways required for neurodevelopment has been demonstrated for key steps leading to brain-tumour progression9, which suggests that this might be a key mechanism underlying how cancer cells successfully colonize the CNS.

Efforts to prevent cancers from reaching the CNS might be a particularly effective approach for improving the clinical outlook for many of these diseases. In a mouse model4 of tumour progression to brain sites, inhibition of the protein VEGF-A hinders a crucial early switch needed for the formation of tumour blood vessels. In ALL, VEGF-A expression levels are high, and a VEGF-A blocking antibody can prevent the disease invading the meninges of the CNS10. Yao and colleagues found that PI3Kδ inhibition results in substantially lower expression of VEGF-A (see Extended Data Fig. 4b of the paper1), raising the question of whether this protein is also involved in the colonization of the CNS in ALL. Because blood-vessel formation is not likely to be involved for ALL, given that it is not a solid cancer, perhaps VEGF-A alters the blood vessels that penetrate the CNS, to make them more likely to provide routes for ALL invasion.

Questions remain about the exact route of ALL entry to the CNS. Do the microvessels that directly bridge this route and penetrate tiny holes in the skull, as demonstrated by Yao and colleagues in their mouse studies, also exist in humans? If cancer cells follow the surfaces of known larger blood vessels to travel from the bone marrow to the subarachnoid space in humans, such a route would be of considerable anatomical complexity. Moreover, cancer cells from solid tumours regularly colonize the bone marrow and frequently invade bone, so it would be interesting to determine whether such cells can also use the route taken by ALL to reach the subarachnoid space.

The presence of laminin at all known sites of cancer-cell entry to the CNS, and the ability of laminin exposure to boost the survival of cancer cells3,4 clearly indicates that more-extensive studies to investigate laminin’s role in cancer-cell entry to the CNS are called for. Which versions of integrins and laminin are involved, and which exact routes are used by cancer cells? Finally, Yao and colleagues’ insights could be relevant to the clinic. Another PI3Kδ inhibitor, called idelalisib, is already in use to treat certain blood cancers. Could this drug finally offer a way to target the CNS entry of ALL or other types of cancer?

Nature 560, 35-36 (2018)

doi: 10.1038/d41586-018-05693-x
Nature Briefing

Sign up for the daily Nature Briefing email newsletter

Stay up to date with what matters in science and why, handpicked from Nature and other publications worldwide.

Sign Up


  1. 1.

    Yao, H. et al. Nature 560, 55–60 (2018).

  2. 2.

    Colognato, H., ffrench-Constant, C. & Feltri, M. L. Trends Neurosci. 28, 480–486 (2005).

  3. 3.

    Boire, A. et al. Cell 168, 1101–1113 (2017).

  4. 4.

    Kienast, Y. et al. Nature Med. 16, 116–122 (2010).

  5. 5.

    Carbonell, W. S., Ansorge, O., Sibson, N. & Muschel, R. PLoS ONE 4, e5857 (2009).

  6. 6.

    Thomas, L. B. Cancer Res. 25, 1555–1571 (1965).

  7. 7.

    Gritsenko, P. G., Ilina, O. & Friedl, P. J. Pathol. 226, 185–199 (2012).

  8. 8.

    Williams, M. T. S. et al. Blood 127, 1998–2006 (2016).

  9. 9.

    Osswald, M. et al. Nature 528, 93–98 (2015).

  10. 10.

    Münch, V. et al. Blood 130, 643–654 (2017).

Download references