A systematic scoping review on natural killer cell function in colorectal cancer

The role of NK cells in immune surveillance

NK cells have been a recent focus in cancer research given their role in immune surveillance as most mutated cells either undergo apoptosis, or are lysed by innate (NK cells) or humoral (T cells) actions. As it is known that even ‘early stage’ cancer sufferers have viable circulating tumour cells, it is reasonable to assume that these cells are being destroyed by the immune system, otherwise they would establish metastases much more readily than is seen. In addition, NK cell cytotoxic function has long been known to be a prognostic factor in CRC recurrence [1].

NK cells express a glycoprotein implicated in cell-to-cell adhesion. Expression of CD56 can be used to subtype them into CD56 “bright” and CD56 “dim” cells. The “bright” cells are secretory, producing interferon-gamma (IFN-γ) and Tumour Necrosis Factor-alpha (TNF-a) and they constitute about 10% of the normal peripheral NK cell population.

The most important cytokine secreted by NK cells is IFN-γ which promotes T cells, up-regulates MHC-1 and has an anti-proliferative effect on malignant cells [2]. It seems likely that CD56 bright cells are immature versions of CD56 dim cells and occur more in the secondarily lymphoid tissues such as tonsils [2].

CD56 dim cells, the remaining 90%, are cytotoxic and can kill cancer cells. One mechanism is by Antibody-Dependent Cellular Cytotoxicity (ADCC). This requires antibodies to bind to antigens on the tumour cell surface. The NK cells recognise the cell-bound antibodies, cross-link to them, triggering degranulation and destruction of the tumour cell. It is this mechanism that has been suggested as the primary effector when treating lymphoma and breast cancer with rituximab and trastuzumab, respectively [2, 3]. NK cells also recognise loss (or change) in MHC1 expression which can trigger degranulation and cytolysis.

Another relevant mechanism of tumour cell lysis involves a ligand-receptor interaction with the NKG2D receptor, which is found on all NK cells. This can detect multiple versions of its ligand and, when activated, causes the NK cell to degranulate and lyse the ligand-bearing cell. Importantly, these ligands are only expressed by cells at times of genotoxic or cellular stress including malignant transformation. While there are many receptors capable of causing activation of the NK cell, NKG2D has been most widely studied in the literature pertaining to CRC.

The genetic influences on NK cell function

Cytotoxic function of NK cells is largely determined at a genetic level within the Natural Killer Complex found on chromosome 12. The generation of NKG2D receptors is particularly important; having high activity receptor haplotypes is correlated with a lower lifetime risk of CRC in a studied Japanese population [4]. Disease progression is associated with a decrease in NKG2D receptors, despite a high or increasing number of circulating NK cells. The expression of NKG2D is the lowest in metastatic disease [5]. It is likely that this effect is mediated by down-regulation of the receptor due to a high amount of circulating soluble ligands secreted by the tumour cells. Tumour cell metalloproteases (which are overexpressed in malignancy) cause shedding of soluble ligands to NKG2D and other NK activating receptors leading to down-regulation of these receptors on the NK cell and immune evasion [6]. As the tumour develops and continues to mutate it switches off NKG2D ligand expression. This pattern was seen in a retrospective CRC study which showed decreasing NKG2D ligand expression in tumours for increasing stage [7].

NK cell function and colorectal cancer risk

Imai et al. prospectively recruited a cohort of 154 people and measured the cytotoxic function of their circulating NK cells. They then followed these subjects for 11 years monitoring for development of cancer. Those with a higher cytotoxic function were found to have a significantly reduced risk of cancer of all types [8].

Similarly, a prospective study looking at NK cell (secretory) function in patients having screening colonoscopy identified a significant and dramatic decrease in function in patients found to have CRC [9]. Interestingly, however, there was no difference in activity of NK cells in patients with or without polyps. In a similar study in Korea, also using secretion of IFN-γ as a surrogate for NK cell activity, Jung et al. found a significant difference in NK cell function in those with advanced adenoma versus no adenoma [10]. Both these latter studies have methodological flaws and use a proprietary kit to stimulate cells and measure IFN-γ release, whilst this may have merit it neither directly measures NK cell function nor seeks to separate out NK cell function from other secreting immune cells.

Halama et al. found that there were far fewer NK cells in adenomas than adjacent colonic mucosa and that this decrease was maintained in both primary colonic cancers and colonic metastases. The surrounding ‘normal mucosa’ retained its number of NK cells [12]. In other studies, there is poor recruitment of NK cells to liver metastases whilst also demonstrating a different phenotype expression of NKG2D and KIR (Killer-cell Immunoglobulin-like Receptor) [13].

NK cell function appears to be a prognosticator in established CRC. Kondo et al. found that patients with lower pre-operative circulating NK cell cytocidal activity had a worse prognosis with a significantly higher rate of metastatic progression and worse 5-year survival [11]. In addition, they showed that decreased NK cell activity also correlates with risk of development of metachronous disease.

The effect of colorectal cancer on NK cells

Epsi et al. compared pre-operative NK cell function in CRC patients to those of healthy volunteers and found that function was depressed in an increasing fashion from stage 1 to stage 3 disease. After curative surgery, the NK cell function initially decreased rapidly but recovered by day 21. This may be a result of the presence and subsequent removal of tumour, although it is also possible that the effect is mediated by other confounders, including the psychological stress of impending surgery [14]. Similar results were shown in a later study with the additional finding of lower expression of TIM-3 (T cell immunoglobulin and mucin domain), suggesting that not only total number of cells but also phenotypic expression is affected in CRC [15].

Circulating tumour cells (CTCs) and NK cells

In their study into the interaction of CTCs and NK cell function in CRC patients, Santos et al. found that NK cell function is negatively correlated with number of CTCs. The implication may be that poorer NK cell function means less cytotoxic activity in the blood allowing CTCs to survive longer and in larger numbers and this eventually leads to a poorer prognosis, although causality cannot be assumed from this study [16].

The effect of surgery on NK cells

Surgery provides many positive benefits for cancer sufferers but there is also evidence to suggest that it can accelerate metastatic progression, and even lead to poorer survival in a subset of patients. The term ‘surgical stress response’ is used to describe the impact of a surgical intervention on patients. This involves components such as the magnitude of the incision, the duration of surgery, the types of anaesthesia and analgesia, the pain response and the impact of peri-and post-operative sepsis. CRC clinical studies have shown that anastomotic leaks lead to poorer oncological outcomes [17].

In their studies with murine models of surgical stress, Tai et al. have described the depletion in number and function of NK cells in the post-operative period and a correlated increase in post-operative metastasis. The study used the model of abdominal nephrectomy and injected melanoma cells. They exposed both iatrogenically NK cell depleted mice and genetically NK cell depleted mice to abdominal surgery and found that in these mice (which lack NK cells), there was no difference in the rate of pulmonary metastasis in the operative or non-operative group. Furthermore, they inoculated NK deplete mice with tumour cells then infused NK cells donated from either surgically stressed or non-stressed mice. They demonstrated an increased rate of pulmonary metastases in the stressed donor group. This suggests that it is the post-operative effect on NK cells that causes the relative increase in metastatic vulnerability in the post-operative period [18]. This period has become known as the post-operative “immunological window”.

The ability of NK cells to kill K652 myeloid cells immediately after surgery is reduced in the first 24 h [19]. What is yet to be fully understood is the mechanism by which this happens. One proposition is that this is a catecholamine mechanism. In a mouse model, Ben-Eliyahu et al. found a similar deleterious effect on NK cell function in ‘stressed’ mice as other authors, which was associated with higher numbers of metastases. This association was reversed by the addition of ß blockade, and by de-medullation. Similar effects can also be induced by the administration of ß agonists alone [20]. This effect is due to a functional decline in NK cell cytotoxicity, rather than a reduction in number of cells. Adrenaline infusion causes a transient increase in circulating NK cells but a paradoxical decrease in their cytotoxic activity, which may be explained by NK cell exhaustion.

Another explanation for the reduction in NK cell activity post-operatively is the down-regulation of NKG2D caused by a soluble exogenous ligand called sMICA. This protein is released by tumour cells and may be increased by tumour manipulation during surgery. sMICA inhibits NK cell cytotoxicity by binding NKG2D and, ultimately down-regulating it. In addition, NKG2D has been found to be down-regulated in surgically stressed mice [18]. NK cell recovery may depend on the subsequent change in genetic expression up-regulating the NKG2D receptor [21].

Wichmann et al. [22] found that a laparoscopic approach to CRC resection leads to a smaller post-operative reduction in the total number of circulating NK cells than open surgery. Interestingly this was not seen in the humoral immune system (T and B cells) which were equally affected in both groups. This does suggest one mechanism for the observed oncological benefit of laparoscopic surgery. Unfortunately, no functional analysis was performed on the cells.

Perioperative blood transfusion

It is well documented that perioperative blood transfusion leads to poorer oncological outcomes for patients. One suggested reason is the effect blood transfusion has on the recipients’ immune system. In this large, retrospective study by Qiu et al., CRC patients had their perioperative blood transfusion status recorded and were followed for 10 years. Both circulating T cell and NK cells were significantly altered after transfusion, and oncological outcomes at 10 years were significantly worse in the transfused group. The authors suggest that the negative effects of blood transfusion may be potentiated by its immunological effects in perioperative immunological window [23].

The effect of perioperative medications on NK cells

Volatile anaesthetics, ketamine and thiopental have been shown to decrease both the number and function of circulating NK cells [24] but propofol anaesthetic seems to preserve function [25]. In murine models, all opioids have been shown to suppress NK cell activity; it is likely that there is a varying degree of suppression between opioid groups [26]. It is likely that this effect is linked to an increase in circulating steroids and, therefore, may be more predominant during the “surgical stress response” [27].

Restoring NK cell function

IL-2 (interleukin-2) infusion may repair some of the post-operative immunological impairment but it is unable to restore NK cell function to that of a healthy volunteer and (unlike the ex vivo environment) does not cause NK cell expansion. IL-15 and IL-21 are not approved for clinical use but have been used to expand and up-regulate NKG2D in vitro [6]. There is an increasing focus on medications which may enhance the anti-tumoural effect of NK cells and whilst the mechanism is unclear, lenalidomide has been found to enhance the in vivo functions of NK cells [28]. Prostaglandin E2 is a potent down-regulator of NK cell cytotoxic function and is potentially modifiable by COX-2 inhibitors and antihistamine [6].

In this review by Chretien and colleagues [6], the authors explore the role of both inhibitory and activating receptors on NK cells. The most widely studied activating receptor is NKG2D which is previously discussed but there are others including DNAM-1 and Natural Cytotoxic receptors (NCRs). Tumours are able to induce down-regulation of the activating receptors and often overexpress inhibitory ligands to evade immunological surveillance. Cytokine infusion increases the expression of activating receptors; unfortunately, these effects are diminished and not sustained in the circulating NK cells of patients with haematological malignancy.

Psychological well-being and NK cells

It has been suggested that poorer psychological well-being may lead to immunocompromise. This was tested in a Danish study that provided psycho-social support in the form of home visits as provided to CRC patients during their follow-up. No difference was seen in the control vs. intervention arm NK cell cytotoxic function [29].

The future of NK cell manipulation

Given that NK cell cytotoxic function appears very important in many aspects of tumour control, and that physiological, surgical and possibly psychological stress impairs this, strategies to increase NK cell function at the times of stress might be beneficial. These strategies can be grouped into those aimed at increasing the number and/or function of the cells present, and those that block the factors that impair function and/or number.

One option is to use NK cell therapy. Transfusion of NK cells in the perioperative period can be achieved using autologous transplant or commercially available NK cell lines (e.g., NK-96). The cell group is then stimulated and expanded [30]. This has only been reported in early-stage safety trials.

Both NK cells and T cells express KIR which, if activated, send inhibitory signals and prevent secretion and degranulation. Cancer cells exploit this by expressing KIR ligands. Anti-KIR antibodies could theoretically be co-administered to prevent the inhibition of NK cell function from autologous cells. There have been no safety issues reported with infusion at this stage [31]. A different group co-administered IL-2 to improve cytotoxic function, the effect and requirement for this co-administration are unclear.[32].

Commercial cell line NK-92 is highly cytotoxic as it lacks KIR and it is unable to mediate antibody-dependent cellular cytotoxicity. A group used these in the adjuvant setting, alongside standard chemotherapy for stage 3 CRC. In this paper, the retrospective design is limiting and the follow-up is not sufficient to fully determine efficacy, but there was a statistically significant difference in disease-free survival [33].

There may be opportunities to modulate the NK cell function in the perioperative period by the off-label use of common drugs such as propranolol (a non-selective B blocker) and NSAIDS (non-steroidal anti-inflammatory drugs). These have been shown in animal models to protect against the post-operative vulnerability to metastases [34]. There is less evidence in human studies, but retrospective analyses have shown improved oncological survival in patients who are taking beta blockers when they come for surgery [35]. NSAIDs are a more complicated area in colorectal surgery due to the potential association with impaired anastomotic wound healing, although large series do not consistently demonstrate this [36].

Blocking the PGE2 (prostaglandin E2) mediated down-regulation of NKG2D with antihistamine has been demonstrated in laboratory studies. This pathway is mediated by the release of prostaglandins from mononuclear and polymorphonuclear phagocytes as part of the intensely complex regulation of the immune response [6].

Another promising intervention may be to pre-operatively boost the NK cell function by delivering oncolytic vaccinations, or even influenza vaccination to minimise the negative effects on NK cells by surgery [18, 37].

A Chinese study that sought to identify biomarkers for prognostication and detection of colorectal cancer found changes in the genetic expression of NK cells that occur early in the course of the disease [38]. They surmise that genetic manipulation of NK cells may be useful in the treatment of CRC.

IFN-γ is secreted by NK dim cells and causes T helper and cytotoxic T cells to recognize and destroy cancer cells. However, despite the theoretical benefit of administration of IFN-γ it is associated with a profound side effect profile. An Italian group has trialed the use of specifically prepared IFN-γ sequential kinetic activation (SKA- IFN-γ) which allows its use at a much lower dose to achieve similar anticancer effects. It may be that cytokine manipulation will develop as a role in cancer therapy alongside immunological manipulation [39].

Conflict of interest

None declared. This research did not receive grants from any funding agency in the public, commercial or not-for-profit sectors.