The guidelines for using cell lines in biomedical research, published recently in BJC (Geraghty et al, 2014), include a pioneering safety warning about human-to-human cancer transmission through cancer cells, a route that we will call extracorporeal metastasis (XM). Because XM has been tacitly dismissed as implausible in everyday laboratory practice, if considered at all, we emphasise the warning by reviewing the underlying evidence and preventive measures.

The possibility of XM was suggested 50 years ago (Bloom et al, 1951) and confirmed by finding tumours that are transplantable in outbred mice (Eiselein and Biggs, 1970; Fidler et al, 1981). In wild animals, XM has been documented for dogs (Murgia et al, 2006; Rebbeck et al, 2009; Belov, 2011), in which cancer is sexually transmitted (Rebbeck et al, 2009), and Tasmanian devils, in which it is passed by biting (Murchison et al, 2012). Although the devils are highly inbred, dogs are not, implying that transmitted cancer cells evade the immune response. Indeed, they do so through the loss of major histocompatibility complex proteins, thus preventing humoral immune response, and by increasing TGF-β expression, which protects from natural killer cells (Siddle and Kaufman, 2013). Because evading the immune surveillance is also common to human cancer (Cavallo et al, 2011), it should not come as a surprise that XM happens in humans, albeit under particular circumstances.

The possibility of XM in humans was first tested half a century ago in experiments that now appear medieval and that involved inoculating human cancer tissues or cancer cell lines into healthy individuals or into cancer patients (Moore et al, 1957; Langer, 1964; Brunschwig et al, 1965). Most inoculates failed to survive but some persisted, metastasising into local lymph nodes or recurring after the primary tumours caused by the inoculation had been excised (Moore et al, 1957; Langer, 1964). In another experiment, a slice of melanoma transplanted from a patient into her 80-year-old mother killed the recipient 451 days later by disseminated metastases, although the initial implant was resected 21 days after the implantation and the patient was treated with chemotherapy (Scanlon et al, 1965).

The implications of these now-unthinkable experiments became clear once organ transplantation became common and XM (especially by melanoma) through transplanted organs became a serious problem, as even organs free of overt cancerous tumours can still transmit cancer, apparently by harbouring disseminated or circulating cancer cells from the donor (Strauss and Thomas, 2010; Desai and Neuberger, 2014). This problem has been minimised by screening donors, but not yet eliminated (Desai and Neuberger, 2014). The risk of XM does not seem to apply to blood transfusion from donors who previously had cancer, at least to immunocompetent recipients (Yang et al, 2010), perhaps because cancer cells do not survive or adhere to the plastic containers during processing and storage of blood (Matsui et al, 1989; Simanovsky et al, 2008; Brennen et al, 2013).

Unfortunately, XM is not limited to immunocompromised individuals and does not require organ transplant to occur. In one reported case, a sarcoma was transmitted from a patient to the surgeon who pricked his hand during surgery (Gartner et al, 1996). The transmission was noticed and documented only because the pathologist who examined the patient’s tumour also happened to examine the surgeon’s tumour and noticed that their histopathology was remarkably similar, which prompted the investigation (Gartner et al, 1996). A similar accident occurred in a laboratory at the National Institutes of Health (USA), when a healthy young woman accidentally pricked her hand with a needle ‘that had been previously used to draw up a suspension of a human colonic adenocarcinoma cell line’ (Gugel and Sanders, 1986). The wound was superficial, but 2 weeks later it produced a nodule formed by the adenocarcinoma cell line. Remarkably, the nodule showed no signs of inflammation (Gugel and Sanders, 1986), highlighting the ability of cancer cells to avoid immune surveillance. Such accidents—pricking yourself with a needle or scalpel previously exposed to cancer cells—are not common, but by no means extraordinarily rare in the operating room or the laboratory, implying that these two reported cases of transmission may be exceptional only in that XM was noticed, documented and communicated to warn the broader biomedical community.

Besides pricking accidents, it is reasonable to assume that other routes used by infectious agents can also enable XM. These include the cracks on the skin, entering through the eyes, which might be particularly vulnerable because of the limited activity of the immune system (McKenna and Chen, 2010), and inhalation of aerosols, which are commonly formed while handling cells and have been documented as a route of cell line cross-contamination (Torsvik et al, 2010). Although inhaling cells while handling them in tissue culture hoods is highly unlikely, as the hoods are designed to prevent this possibility, cancer cells are routinely collected and processed outside the hoods, at which point they are considered as merely a reagent rather than an organism that can invade a human.

In our experience, the possibility of XM is generally unknown to laboratory researchers, as it is not reviewed during their safety training, or is dismissed as implausible. Yet, without awareness, the risk of accidental XM in the laboratory may increase in the future as more cancer cell lines are established, and the lines that are already in use continue to evolve. Thousands of human cell lines have been established over the last 50 years (Barretina et al, 2012) by explanting cancer tissues, which implies selection for new properties. In addition to natural selection, the diversity of cancer cell lines has been further increased by routinely modifying them genetically. At the same time, the ability of cells to evade the immune response is usually tested only in the studies that are concerned directly with this question.

Out of an abundance of caution, we propose two actions to minimise the risk of XM in the laboratory. First, the notion that cancer cells themselves are possible pathogens should be included into routine laboratory safety training. Second, cancer cell lines, perhaps starting with those provided commercially and by cell banks, should be tested in vitro for their ability to evade immune responses in humans. The lines that show potential for immune evasion should be labelled accordingly and used with all due care.