Confrontation with stressors can affect the immune system and increase cancer occurrence in human (75). The immune system mediates the relationship between stressors and cancer occurrence (76). Exposure to stressors is found to be accompanied by pro-inflammatory responses in animal and human research (77), which may stimulate tumor growth. Furthermore, stress related molecules, such as cortisol or the catecholamines (nor)epinephrine, can regulate diverse signaling pathways through their specific receptors that enhance the proliferative and invasive abilities of cancer cells in relation with the tumor microenvironment (78, 79). Cortisol or glucocorticoids have a pivotal role in regulating stress reactivity of organ systems (80) through glucocorticoid-receptor-mediated modulation of target genes (34). Glucocorticoids can activate survival genes that protect cancer cells from the effects of chemotherapy (81), and were shown to influence virus activation including human papillomaviruses and other cancer-associated viruses (34). Instead, to counteract the stimulating effects of (nor)epinephrine on tumor growth, the administration of beta blockers that interfere with the physiological stress response was associated with a lower incidence of prostate cancer in a population based study (82).

The influence of stressors on cancer incidence is also hypothesized to be moderated by socioeconomic status. Disparities in socioeconomic status are associated to inequalities in behavioral factors such as physical inactivity, obesity, smoking, diet, alcohol and drug use, screening and treatment uptake (83). These health-impairing behaviors are thought to be stress-related behaviors and a lower socioeconomic status has been associated with higher levels of distress (84). Higher cancer occurrence is found in groups with lower socioeconomic status (83, 85). Even though confrontation with long lasting stressors may be one of the factors that plays a role in these processes, research should detangle the mechanisms with which stress modulates cancer occurrence and therapy response.

In humans, pooled analyses of epidemiological studies showed that more physical activity during leisure time was associated with a decreased risk of 10 different type of cancers, independent of body mass index (BMI) (86). Overall, physical activity is associated with a 7%-20% lower cancer risk in individuals, with the strongest impact on colorectal and breast cancer (74, 87). One of the mechanisms by which physical activity may reduce risk of cancer occurrence is a reduction in chronic, low-grade inflammation and improved immune surveillance and function (88, 89). Cumulative evidence of both animal and humans studies shows that exercise modulates local and systemic inflammatory processes by altering both the number and function of circulating cells of the innate immune system (neutrophils, monocytes and NK cells), and of the adaptive immune system (T and B cells) (89). Exercise may also reduce the visceral fat mass, which is accompanied by less adipokine secretion and less macrophage infiltration into the adipose tissue, thereby reducing inflammation (90). Exercise also activates the HPA axis, initiating cortisol release, which in the case of exercise can contribute to an effective anti-inflammatory systemic host environment by downregulating cytokines as tumor necrosis factor (TNF)-α. The aforementioned physiological events activated by exercise are only examples among many other processes that have been detailed in a number of reviews (89–91). For example, exercise is accompanied by a higher level of catecholamines, such as (nor)epinephrine, which were related to similar anti-inflammatory effects as cortisol (90).

Different from psychosocial stress or exercise, which can also be part of lifestyle, classical pharmacological conditioning is always an intervention and, therefore, there are no epidemiological studies investigating this behavioral mechanism in relation to cancer onset.

There are indications that pharmacological stress modulation can improve cancer progression. The physiological stress response seems to drive therapeutic resistance in murine tumor models (93, 94). The cellular and molecular microenvironment of cancer includes (peripheral) nerves that can modulate behavior or malignant cells, promoting tumor growth and illustrating the cross-talk between the neuroimmune system and cancer progression (35). Regulation of the tumor microenvironment by the sympathetic nervous system has been demonstrated in animal studies (95). Intratumoral neurotransmitters and neuropeptides have regulatory roles in the physiological and pathological functions of tissues, and emerging data suggest that cancer cells may take advantage of neurotransmitters-initiated signaling pathways to activate uncontrolled proliferation (96). For example, norepinephrine and epinephrine activate β-adrenoreceptors expressed on both cancer and immune cells thereby promoting growth of malignancies and inflammation. Moreover, these catecholamines can induce an endothelial cell metabolic switch mediated by β-adrenoreceptors resulting in increased tumor vascularization (96). The β-adrenergic pathway may be suppressed by beta blockers and as an example, it was shown that propranolol, a medication of the β-blocker class, was used to complement the treatment of several types of cancer, directly blocking cancer cell proliferation induced by epinephrine in vitro (97). This experimental evidence is supported by clinical studies that combine propranolol with other agents to stop metastasis (98), and epidemiological evidence showing that of 24,238 patients, the 12,119 propranolol users (for over six months) had lower risk of head and neck, esophagus, stomach, colon and prostate cancers (99). On the other hand, preclinical administration of dexamethasone, a synthetic glucocorticoid, induced chemotherapy and resistance in breast cancer, as well as in vitro tumor samples and cancer cell lines (100–102). This evidence underlines the importance of the HPA-pathway modulation in conventional cancer treatments.

While psychological interventions seem to influence immunity (103), evidence in relation with cancer remains very limited. Studies using psychological interventions in patients with cancer often did not assess treatment efficacy or health-related quality of life (104). The results of interventions on psychological wellbeing, such as cognitive behavioral therapy, in cancer treatment are variable finding mostly effects on outcomes such as anxiety, and fatigue (105). However, these interventions, e.g. cognitive behavioral therapy and mindfulness, cannot be tested in preclinical models.

There is both pre-clinical and clinical evidence for a relation between exercise and the immune system and effectiveness of exercise during chemotherapy or radiotherapy (16–26). In patients, exercise has been associated with reduced side effects of cancer and its treatment (27, 28). Thereby, exercise improved the physical and mental health and the overall self-reported quality of life of patients (27–29). Patient-reported outcome measures were complemented by immunological readouts, including the number of NK cells, expression of IL-6, or TNF-α production (16–26). In observational and randomized controlled exercise trials - both increases and decreases in defined immune markers were reported. For example, immune markers of NK cells differed: exercise had an inhibitory effect on the absolute number of NK cells in patients with breast cancer (106), an augmenting effect on NK cells percentages in patients with lung cancer (107), whereas no effect in another cohort of patients with breast cancer and other solid tumors was observed (17, 22). Comparably, IL-6 expression differed after exercise interventions, showing either a decrease in some studies (24–26), but no effects (17, 20, 23, 108), or an increase in other studies (16, 18, 109). Given the diverse effects that exercise has on the number of immune cells in patients with cancer, for instance in NK cells, it has been suggested that exercise may instead affect the cytotoxic activity of the immune cells, mirroring the effects exercise has on healthy individuals through inflammatory response pathways (110, 111). The large variation in the type of exercise interventions ranging from aerobic to resistance training may explain differences in exercise responses (26, 106, 107).

To date, most causal evidence of exercise on the anti-tumor efficacy of cancer treatment comes from animal studies (112, 113). A large advantage of these in vivo experiments is that there is little variation in interventions; i.e. most studies examined the effects of voluntary running. These experiments indicate that physical exercise modulates factors that are inherent to cancer treatment sensitivity, including the tumor microenvironment, e.g. hypoxia, tumor cell metabolism and tumor perfusion, next to having profound effects on immune cell populations (112–114). Illustrating these results, voluntary running tumor bearing mice experienced reduced tumor growth in diverse cancer models, e.g. lung cancer or myeloma, and displayed higher NK cell mobilization compared to sedentary control groups (113, 115, 116).

Conditioned effects in cancer patients during therapy were shown on outcomes such as nausea and immune modulation (117–120). Patients who were given a beverage prior to adjuvant chemotherapy experienced more nausea at later time points when they were confronted with the beverage alone compared to patients who did not receive a beverage before the therapy (118), and the other way around, e.g. a conditioning paradigm was applied to reduce nausea (120). Similarly, pediatric cancer patients undergoing chemotherapy showed increased natural killer cell activity and interferon-γ levels upon arrival at the hospital, when previously confronted with two cycles of chemotherapy in the hospital (117). These results may be seen as an indication that conditioned effects in cancer patients are possible on immune cell populations, however, the effects on cancer outcomes and therapy response remain to be investigated.

The mechanisms of how the immune system is conditioned are largely unknown. An association between the conditioned stimulus and the immune response needs to be established in the brain and conditioning thus relies on the interaction between the central nervous system and the immune system (15). Some murine studies demonstrated that lesioning of the insular cortex and central nucleus of the amygdala obstructs immunological conditioning, suggesting that these areas mediate conditioning (121, 122). Other murine studies hypothesize that continuous administration of substances, such as antigens, may be disruptive for the hosts’ homeostasis, and that conditioning may be a favorable alternative given a decreased substance administration (123). By linking the immune reaction with the central nervous system, it is assumed that the effect of the substance might be achieved without the disruptive effects of the substance (123). Furthermore, it is thought that different pharmacological conditioning paradigms rely on different mechanisms, given that the physiological reaction mimics diverse medication effects (124, 125). Despite little evidence on the pathways of this associative learning process, the conditioning paradigm has been successfully used in numerous rodent studies of nonconventional therapies, echoing the effects of cyclosporine A, opioids, lipopolysaccharide (LPS), lithium chloride, anti-lymphocyte serum, ovalbumin and bovine serum albumin using taste and odor as conditioned stimuli (126–132).

Immune suppressive mechanisms in cancer hamper effective immune responses (133). By therapeutically assisting anti-cancer immune responses, tumor growth and progression may be counteracted and cure can be promoted. Promising targets for immunological therapies are immune checkpoint proteins, which are used as a break in the immune system and consequently block over-activation of the immune system preventing autoimmunity (3). In addition, checkpoint signals are required for optimal T-cell recognition and generation of long-lasting T cell memory responses (3). One of the most well-known checkpoint proteins is Cytotoxic T lymphocyte antigen-4, or CTLA-4 (134), which is considered critical for maintenance of T cell homeostasis and tolerance (135). T cell activation requires engagement of the T cell antigen receptor-CD3 complex and ligation of costimulatory receptors, such as CD28, that bind to CD80 (B7.1) and CD86 (B7.2) on antigen-presenting cells. CTLA-4 is transported to the immunologic synapse when there is a potent or long-lasting stimulus (via the T cell receptor) (136), and outcompetes binding of CD28 to CD80 and CD86, hereby acting as negative regulator of proliferation and effector function of T cells (137). When tumor cells express ligands (e.g. CD80, CD86) for CTLA-4, T cell activity is inhibited after binding, hereby evading clearance by the immune system (138). Monoclonal antibodies, i.e. checkpoint inhibitors, can block CTLA-4, allowing activation of T cells and killing of tumor cells (7).

Programmed death (PD)-1 is another immune checkpoint molecule involved in regulating the balance between immune activation and tolerance, similar to CTLA-4 (134). Its ligands PD-L1 (B7-H1) and PD-L2 (B7-H2) are expressed on antigen presenting cells, but can be expressed on tumor cells as well, resulting in an immunosuppressive tumor microenvironment. Therefore, also anti-PD-1 or anti-PD-L1 monoclonal antibodies promote T cell-mediated tumor cell death (139, 140).

Immune checkpoint inhibitors (i.e. Ipilimumab (anti-CTLA-4), Nivolumab (anti-PD-1), Pembrolizumab (anti-PD-1), Atezolizumab (anti-PD-L1)) have been approved or are studied in clinical trials to treat multiple types of cancer of which melanoma and lung cancer respond best to therapy. Next to this, there are currently several antibodies and small molecules in development, targeting other immune checkpoints such as TIM3, CD39, B7H3, CD73, LAG3, and more (141–145).

Unfortunately, large groups of patients do not respond to or benefit from immunotherapy (8–10). It is not completely clear yet why some patients respond and others do not, which may have to do with tumor-intrinsic qualities. For instance, it was shown that high microsatellite instability (MSI) results into a high number of mutations and increased number of tumor-infiltrating lymphocytes (146, 147). As such, patients with an MSI-tumor are suitable candidates for immunotherapy. Nonetheless, even in tumors with high MSI observed response rates range between 30% and 50% (147), indicating that there are other factors that come into play besides the genetic and immunologic aspects of the tumor. Similarly, predictive biomarkers, like tumor-cell PD-L1 expression, are used to stratify the immunotherapy responders from the non-responders (148). However, PD-L1 testing alone is insufficient for patient selection in most malignancies and immune responses are not uniform across all malignancies. It is estimated that in the US 38% of patients with cancer are eligible for ICI therapy, given the molecular profile of their tumor, but only up to 11% respond to the ICI therapy (10). The remaining 27% of patients were eligible but did not respond, indicating the necessity for better predictive biomarkers, next to the need of increasing treatment sensitivity. Another drawback of checkpoint inhibition is that it is not cost effective in certain malignancies, with an economic benefit for choosing chemotherapy to treat i.e. recurrent or metastatic head and neck cancers and non-small cell lung cancers (12). Furthermore, both CTLA-4 and PD-1 blockade can have severe immune related autoimmune complications, for example side effects on the gastrointestinal tract, brain, thyroid, lungs and skin (3, 13, 14). In the light of the diverse drawbacks, efforts are needed to improve and support immunotherapy, enhance its anti-cancer effects and decrease the side effects.

Behavioral factors have been shown to influence both the immune system and cancer treatment and may therefore possibly offer opportunities to improve immunotherapy. Given that behavioral factors can influence the effectiveness of conventional chemo- and radiotherapy, and that the immune system is thought to modulate this effect, immunotherapy may offer an ideal opportunity for behavioral intervention (Figure 2).

Few preliminary studies investigated the impact of stress modulation with pharmacological interventions on checkpoint inhibitors efficacyalthough there are indications that stress influences immunotherapy. For instance, social disruption stress compromised a vaccine based immunotherapy (poly(d,l-lactide-co-glycolide) microsphere) in a murine melanoma model through impairing CD8⁺ T cell responses (149). Also, either behavioral stress or surgical stress, weakened the inhibition of metastasis by immunostimulating agents (CpG class C and glucopyranosyl lipid-A stable emulsion) (150, 151). Studies linking stress to cancer immunotherapy assume the involvement of the of HPA-axis, for example by glucocorticoid-induced expression of the immunosuppressive transcription factor TSC22D3 of dendritic cells (152).

With regard to ICI in a retrospective analysis of 109 medical records of non-small lung cancer patients, treated with either ICI therapy or ICI in combination with chemotherapy, the 28 patients who were concomitantly prescribed any beta blocker had a longer progression free survival with a hazard ratio of 0.58 (153). A possible association between beta blocker use and improved progression free survival in non-small-cell lung cancer patients treated with ICI, should however be confirmed in clinical randomized controlled trials. In humans, there are no randomized controlled trials using psychological interventions to complement immune checkpoint inhibition cancer therapy.

An indication of a relation between stress exposure and the efficacy of anti-PD-L1 was found in a mouse study (154). In this study researchers applied daily chronic unpredictable mild stressors, such as water or food deprivation, tail pinching, bodily restraint et cetera, for a total of 28 days. The effect of anti-PD-L1 therapy was attenuated in the stressed groups, coinciding with a decrease in CD8⁺ lymphocytes and increase of regulatory T cells at tumor sites. In line with this, chronic cold stress strongly reduced anti-PD-1 efficacy in breast cancer and melanoma mouse models (155). An experimental group that experienced cold-induced stress (by being housed in an environment of 22 degrees Celsius instead of the thermoneutral temperature of 30 degrees Celsius) had larger tumors but this effect was counterbalanced by β-blocker treatment. The enhancement of anti-PD-1 efficacy by co-treatment with propranolol treatment was likely to be CD8⁺ dependent, i.e. there was an increased frequency of effector CD8⁺ T cells in subcutaneous breast cancer tumors and melanoma. The effects of propranolol treatment did not persist in T cell-deficient mice, suggesting that the β-adrenergic system influences T cell activity (155, 156). Similar results were found in mouse models of fibrosarcoma and colon cancers, where reduced tumor growth as well as enhanced response to anti-CTLA-4 therapy was observed after blocking the β-adrenergic receptor with propranolol (156). Here, propranolol treatment resulted in a reduction of tumor angiogenesis, increased T-cell infiltration, but a decrease in myeloid derived suppressor cells, as well as modifications on tumor associated macrophages, together leading to a tumor-suppressive environment (156). Despite these promising preliminary results, we are unaware of any ongoing clinical trials in humans.

Several murine studies investigated the synergistic effects of exercise and immune checkpoint inhibitors on cancer treatment (157–160). Higher tumor necrosis and less apoptosis was found in a patient-derived xenograft model of non-small cell lung carcinoma when anti-PD-1 treatment was combined with exercise, indicating that excise may improve anti-PD-1 effectivity (157). Another study demonstrated that aerobic exercise, i.e. daily 30 minutes treadmill exposure, sensitized pancreatic tumors to anti-PD-1 therapy, which resulted in anti-tumor immunity though IL-15Rα⁺ CD8⁺ T cells and decreased tumor growth (160). Two other in vivo studies found no synergistic effects of the combination of exercise and immune checkpoint inhibition. However, one study observed an increase in CD8⁺ T cells in orthotopically implanted breast tumors when anti-PD-1 was combined with voluntary wheel running, as compared to the tumors of mice without voluntary running regime (158, 159). These preliminary preclinical results offer a stepping-stone to translate exercise interventions to the clinic.

To the best of our knowledge there are no published studies investigating the effects of a learned immune reaction on the efficacy of immune checkpoint inhibition. However, a recent study showed that conditioning of rapamycin-induced immunomodulation reduced tumor growth effectively in a murine glioblastoma model (161). In this study, the mTOR inhibitor Rapamycin was repeatedly paired with a novel gustatory stimulus. The experimental group, receiving only 10% of the initial drug dose together with the gustatory stimulus during the testing phase, showed similar tumor inhibition as the control group receiving 100% of the drug dose. the tumor growth inhibition was driven by a central and peripheral upregulation of pro-inflammatory markers and a decrease in anti-inflammatory cytokines such as IL-10. Similarly, older studies showed that conditioning of immunotherapy was more effective in delaying tumor growth in mice than immunotherapy alone (162, 163). For example, in a syngeneic in vivo study, the unconditioned stimulus was the injection of immunostimulating DBA/2 spleen cells, and the conditioned stimulus was camphor odor. When conditioned mice were re-exposed to the odor of camphor only, tumor growth was still delayed compared to non-conditioned mice (162). These experiments suggest that the immune system of the mice consistently mimicked the effect of the immune modulator when presented with the conditioned stimulus, which influenced health outcomes, demonstrating the feasibility of conditioning immune responses (123, 161–167). Therefore, conditioned effects of immunomodulatory inhibitors may be suitable also for immune checkpoint inhibition. Of note, both mechanistic animal studies and controlled human studies in healthy subjects and patients are necessary to understand whether learned immunity is a promising addition to immunotherapy. In patients, the conditioning paradigm could be applied using specific stimuli, for instance combining the use of checkpoint inhibitors with a distinctive stimulus: a taste, sound or smell. This setting may serve for reducing medication dosages, i.e. checkpoint inhibition could be given in reduced quantities or placebo medication could be administered intermittently. Potentially, the use of mechanisms that harness mimicking placebo effects could reduce healthcare costs associated to the high expenses of several medications, including immune checkpoint inhibition.

Several ongoing clinical trials in patients study how behavioral factors may affect the quality of life and other cancer-related outcomes during ICI (Table 1). To the best of our knowledge there are no ongoing clinical trials on stress modulation or classical pharmacological conditioning in checkpoint inhibition, but various studies use exercise as intervention. The results of these studies are yet unknown and most studies focus on feasibility of the intervention as primary outcome measure. Therefore, the mechanisms of behavior on cancer outcomes in patients undergoing immune checkpoint inhibition therapy remain unknown.