Personalization of Cancer Treatment: Exploring the Role of Chronotherapy in Immune Checkpoint Inhibitor Efficacy

Measured outcomes differed from study to study. Almost all studies investigated overall survival (OS) (27 out of 29, 93%), except for two conference abstracts [43,52]. Barrios et al. focused on time to treatment discontinuation [43], while Sun et al. examined only progression-free survival (PFS) [52]. Importantly, Sun et al. are the only group so far to report more favorable outcomes in patients who received ICI later in the day (one or more infusions administered after 16:00) [52]. The majority of studies also reported PFS (23, 79%). Just under half reported either objective (ORR) or complete response rate (CRR) (12, 41%). Other measured outcomes included disease control rate (DCR) (two studies, 7%) and some variation in time and treatment interaction (time to treatment failure or discontinuation, time on treatment, or time to next treatment; three studies, 10%).

In the included studies, patients were stratified into early and late administration groups based on time of day. One of the most important differences in these studies was the decision to assign patients to these groups based on either the percentage of total infusions or initial infusions received before a time-of-day cutoff. Several early studies demonstrated the importance of the timing of initial infusions for patient outcomes. In a 2022 study by Karaboué et al., landmark analyses revealed that after two months of treatment, the late administration group already had significantly worse OS and PFS, suggesting that the timing of initial ICI treatments is important for efficacy [33]. Later in 2022, Cortellini et al. found that the total number of infusions received was the strongest determinant of patient outcomes and demonstrated that adjusting for this factor reduced OS and PFS from 47.1 and 19.7 months to 29.0 and 11.8 months, respectively, in patients receiving <20% of infusions in the evening [28]. Since then, multiple studies specifically considered outcomes by initial infusions (10 of 29, 31%). Of these, three reported no difference between earlier and later administration, with initial infusions defined as the first dose only [29,32,53]. Seven studies found more favorable results when initial infusions were administered earlier in the day. Of these, two also considered only the first infusion and found more favorable outcomes in the early administration group [35,51]. Other studies defined the early administration of initial infusions as receiving at least one of the first four [41,54,55] or one of the first two [50] doses before the time-of-day cutoff or receiving two to four of the first four doses before the cutoff [46]. For the studies stratifying patients into early or late administration groups based on the proportion of total infusions received before the time-of-day cutoff, various percentages were considered (Table 1).

The cutoff time to determine early or late in the reviewed studies varied widely (Figure 2). As previously described [34], these may have fallen into two broad categories: those chosen for biological reasons and those chosen for practical reasons, such as statistical or clinical constraints. Of the 29 studies, 19 (66%) reported the reasoning behind the choice of time-of-day cutoff. Two (of 19, 11%) studies reported consulting the body of the literature on circadian immunity to choose their cutoffs. As a rationale for this choice, Qian et al. cited pre-clinical and clinical studies of vaccination timing and circadian lymphocyte variation [37], while Janopaul-Naylor et al. cited studies of CD8 T cell diurnal concentrations [32]. Most studies (11 of 19, 58%) reported using the previous literature (i.e., the study by Qian et al. [37]) to choose their cutoff. Two of these (of 11, 18%) adjusted the cutoff to accommodate the clinic’s closing hours and/or statistical considerations, in both instances shifting the cutoff to 15:00 (versus 16:30) [31,39]. Seven of these 19 studies (37%) reported choosing time-of-day cutoffs based on other practical considerations. The most common method involved calculating the median time of all patient infusion times (5 of 7, 71%). Most of these studies, therefore, had seemingly arbitrary cutoff times (11:41 [40], 11:45 [46], 12:54 [33], 15:06 and 15:11 [53]), with the exception of Patel et al., who rounded to the nearest hour (12:00) [36]. One study [27] used a predictiveness curve method to objectively determine the optimal cutoff for separating patients into early and late groups based on overall survival, resulting in a cutoff time of 11:37. Nomura et al. chose as the cutoff the midpoint of clinic opening hours (13:00) [35].

Ten studies (of 29 reviewed, 34%) did not report the reasoning behind the choice of time-of-day cutoff. We note that eight of these ten were conference abstracts, which are subject to space constraints. Of these ten studies, nine (90%) investigated only one time-of-day cutoff. The other [48] had an unusual grouping strategy: patients were grouped into two-hour windows from 8:00 to 20:00 (i.e., 8:00–10:00, 10:00–12:00, 12:00–14:00, etc.) and one overnight cohort from 20:00 to 8:00. The authors reported that patients with head and neck cancer had shorter OS when receiving infusions between 8:00 and 10:00 compared to 12:00–16:00. Patients with lung, renal, and breast cancers who received overnight infusions had significantly shorter OS compared to patients receiving infusions from 10:00–16:00; patients with melanoma had shorter OS when receiving overnight infusions but also when infusions were administered between 8:00 and 10:00 or 18:00 and 20:00 (compared to 10:00 and 16:00). We note that the shorter OS observed for overnight treatment windows may be confounded by a higher likelihood of overnight infusions for hospitalized (and potentially sicker) patients compared to outpatients.

The majority of studies (24, 83%) focused on a single cancer type (Table 1). Of these 24, 9 (38%) were in non-small cell lung cancer (NSCLC), 5 in renal cell carcinoma (21%), 3 in melanoma (13%), and 2 each (8%) in head and neck squamous cell carcinoma (HNSCC) and gastric cancer. There was one instance (4%) each of esophageal cancer, urothelial cancer, and hepatocellular carcinoma. The majority of these studies (22 of 24, 92%) found improved outcomes with early ICI administration. One study of patients with HNSCC found no difference among early, mid, and late administrations of ICI on OS, PFS, or objective response rate [32]. Similarly, Cortellini et al. found no difference between early and late administrations on OS or PFS in patients with NSCLC [28]. It is interesting to note that, altogether, there were 13 studies that included patients with lung cancer (4 that examined multiple cancer types, 9 that studied only NSCLC). Of these, three (23%) found that early administration did not improve patient outcomes [28,48,52].

The remaining five studies analyzed multiple cancer types. Interestingly, three of these studies (all conference abstracts) reported results that do not corroborate the findings of Qian et al. Nelson et al. reported mixed results in patients with solid tumors (including melanoma, breast, colon, head and neck, liver, lung, and renal cancers) [48]. Many of these patients (31%) had non-small cell lung cancer. Van Rensburg et al. reported no difference between early and late ICI administrations in patients with advanced solid tumors (including melanoma, ovarian, head and neck squamous cell, and triple negative breast cancers) [53]. Sun et al. reported that later administration of ICI is more favorable for patients with advanced solid tumors (including esophageal and lung cancer) [52]. The other two studies that examined multiple cancer types were a meta-analysis [34] and a single center retrospective study [27]. The patients included in the meta-analysis by Landré et al. had esophageal carcinoma, melanoma, non-small-cell lung cancer, renal cell carcinoma, or urothelial cancer [34], while those in the study by Catozzi et al. had melanoma or breast, colorectal, head and neck, pancreatic, urinary, or non-small cell lung cancers [27]. Both reported more favorable outcomes with earlier ICI administration.

Of the 29 studies that we reviewed, 22 (76%) specified the ICI that patients received, while the remaining 7 reported only the target molecule (Table 2). Based on the varied results of time-of-day ICI administration reported thus far, we anticipate that the nuances of specific ICIs may become an important consideration and we encourage authors to include this information in future studies.

Some ICI studies suggest that efficacy may differ by patient sex, but overall results are mixed [56,57,58,59]. To synthesize the evidence on potential sex differences in ICI efficacy outcomes based on time-of-day administration, we examined each study for the proportion of females in the sample and whether studies reported on sex differences in efficacy outcomes. Most studies (76%) reported the sex of the sample, and, of those, female patients comprised less than 50% of the sample in all but one of the studies reviewed (Figure 3).

Reporting of sex-specific differences in efficacy outcomes was rare. Only five studies (of 22, 23%) analyzed outcomes by patient sex [27,30,35,36,37]. Of these five studies, three reported better OS [27,30,37] for females in the early administration group and no difference in OS by time-of-day administration for males. Two studies reported no sex differences in OS [35,36]. Interestingly, Catozzi et al. applied a novel sinusoidal Cox regression model to assess the impact of time-of-day administration on OS and predict the best and worst time of day for ICI infusion [27]. Notably, they found that the amplitude of the timing effect was higher for females than for males. For PFS, one study reported better PFS for males in the early administration group (based on first dose administration) and no difference for females [35], and one study reported no sex differences in PFS [30]. Finally, one study reported better ORR for females in the early administration group and no difference in ORR by time-of-day administration for males [27].

Overall, the under-representation of female patients in these studies and limited reporting on sex-specific differences in efficacy are major gaps in the existing literature on ICI time-of-day administration and hinder evaluation of whether chronotherapy may differentially benefit male or female patients. Unfortunately, this trend is reflective of a landscape of female under-representation in clinical trials [60] and deficiencies in sex-based outcomes reporting in oncology [61] despite NIH policy to consider sex as a biological variable [62] and development of the SAGER guidelines [63] to improve sex-based reporting in manuscripts.

Although each of these studies has unavoidable biases due to their retrospective natures, the accumulating evidence suggests that the time of day at which ICI is administered influences patient outcomes in many scenarios. In contrast to studies of chemotherapy and radiotherapy, which report significant differences in time-of-day-dependent toxicity but not treatment outcomes, current studies of patients receiving ICI suggest that earlier administration results in better treatment efficacy. We note that there is a lack of consistency in these results; therefore, prospective, randomized, and rigorous studies are essential to understand the conditions under which timing ICI administration will be reproducibly beneficial.

The sleep–wake cycle, a fundamental aspect of circadian rhythm, is crucial for regulating immune system function and response to external stimuli, influencing both the efficacy and toxicity of ICIs through these mechanisms. During normal sleep–wake cycles, T-helper 1 (Th1) cell activity, driven by IFN-γ-producing cells, is promoted at the start of the nighttime rest period, while T-helper 2 (Th2) activity, driven by IL-4-producing cells, dominates just before waking or during the later part of the rest period [69]. Sleep deprivation leads to an imbalance between Th1 and Th2 cell-derived cytokines, resulting in their excessive production and subsequent immune disturbances, which can lead to chronic inflammation and tissue damage [69]. Disruption in the sleep–wake cycle also causes the loss of the rhythmic temporal variations in IL-12 and IL-10 [69]. IL-12 is involved in naive T cell differentiation into Th1 cells, promotes IFN-y expression, and regulates T cell and natural killer cell responses [70]. IL-10 is an anti-inflammatory cytokine that downregulates Th1 cytokines, enhances B cell survival, and can help promote Th2 differentiation [71]. During sleep, monocytes producing IL-12 are increased and monocytes producing IL-10 are decreased; thus, nocturnal sleep represents the shift between IL-10 (Th2) activity and IL-12 (Th1) activity [69,72]. This circadian rhythm is disrupted during sleep deprivation, and the temporal variations between IL-12 and IL-10 monocytes are lost, causing immune system disfunction. Other noted effects of a sleep–wake cycle imbalance include decreased numbers and the weak lytic activity of NK cells, a decrease in CD8+ T cells, and an amplification of abnormal inflammatory responses including the activation of soluble adhesions molecule (sICAM)-modulated NF-kB inflammatory signaling and the cellular expression of inflammatory markers IL-6, TNF-a, and C-reactive protein (CRP) [3]. Therefore, sleep deprivation may lead to immune system changes that increase the risk of cancer.

Various cytokines such as TNF-a, TGF-b, IL-10, IL-1b, and IL-6 are involved in the crosstalk between cancer initiation/progression and sleep–wake cycles. IL-1b is pro-tumoral, having been shown to be upregulated in many solid tumors, promoting cancer progression [73]. This cytokine was also identified in the brain as a key mediator that affects rhythmical behavior, and sleep deprivation was found to stimulate the expression of IL-1b in the brain. IL-6 is upregulated and abundant in almost all types of tumors, which promotes tumorigenesis and facilitates the repair and induction of countersignaling pathways to protect cancer cells [74]. IL-6 is expressed at low levels during the day and peaks during sleep. Sleep deprivation results in the elevation of plasma levels of circulating IL-6. Both IL-6 and IL-1b illustrate how the sleep–wake cycle is connected to cytokines that play a role in cancer initiation/progression and how the disruption of the cycle is potentially a risk factor for tumorigenesis.

Endocrine factors like growth hormone, prolactin, thyroid hormone, cortisol, and gonadal steroids are governed by circadian rhythms and fluctuate throughout the day. Shift workers are at a higher risk of obesity and weight gain (increased body mass index) compared to dayworkers, potentially due to a mismatch in circadian rhythms from sleep deprivation [75]. Sleep deprivation triggers the decrease in levels of the appetite-restraining adipokine leptin and an increase in the levels of the appetite-stimulating peptide ghrelin, resulting in artificially inflated feelings of hunger and appetite. The hormonal imbalances associated with sleep deprivation lead to increased hunger, reduced satiety, higher cortisol levels, and impaired insulin sensitivity, which contribute to weight gain and can induce hyperplasia, a risk factor for cancer [75].

Disruption of the sleep–wake cycle can increase cancer risk by impairing immune system functionality and affecting pathways critical to ICI efficacy. Insufficient sleep and irregular sleep patterns, such as those experienced by shift workers, are known to influence cancer initiation, progression, and treatment [76]. For instance, sleep dysregulation leads to elevated levels of pro-inflammatory cytokines, such as TNF-α. Notably, TNF inhibition was shown to enhance ICI anti-tumor efficacy while reducing irAEs [77]. Sleep deprivation, a significant disruptor of the sleep–wake cycle, is a known cancer risk factor and can also influence ICI efficacy and the occurrence of immune-related adverse events (irAEs).

ICI drugs function by blocking checkpoint molecule interactions to maintain the functional role of immune cells, especially T cells. It is important to highlight that PD-1 suppresses the functionality of both cytotoxic and regulatory T cells [78]. Patients with high levels of PD-1-expressing T cells have better clinical response on ICIs than those who have low levels of PD-1-expressing T cells [78]. Additionally, anti-PD-1 increases PD-1 CD8+ TCR clonality after treatment, which is correlated with a positive clinical outcome in patients [79,80]. Regulatory T cells (Tregs) are a subset of CD4+ T helper cells that exhibit an immunosuppressive function and are capable of attenuating TCR signaling in effector T cells [81]. Notably, ICIs induce both a specific effector memory T cell subset that is associated with positive outcome and a separate Treg subset that is associated with toxicity [80]. In a sleep study with healthy volunteers, the number of peripheral blood Tregs peaked and remained steady during sleeping hours (19:00–7:00) and decreased shortly after waking up, with a nadir around 11:00. Congruently, Treg suppressive activity on CD4+ T cells was high during sleep, with peak suppression at 2:00 and lowest at 7:00, upon waking [82]. Early morning infusion of ICIs would thus be during a time when Tregs (and their suppressive function) within the peripheral blood are low, enabling a greater T cell-mediated immune response and TCR production. This concept also introduces a sleep-dependent mechanism (versus time-of-day) as a potential explanation for ICI response and toxicity.

Macrophages are a type of innate immune cell that mature from monocytes upon migration to tissue [83]. Upon infiltrating a tumor, macrophages play an important role in eliminating cancer cells by producing chitotriosidase/chitinases, proteases, nitric oxide, and hydrogen peroxide [84]. However, cancer cells can foster immune escape from macrophages by expressing the anti-phagocytosis regulator CD47 [85] or by polarizing macrophages into an anti-inflammatory/immunosuppressive M2 phenotype, which are commonly known as tumor-associated macrophages (TAMs) [86,87]. TAMs are associated with tumor progression and poor prognosis in many cancer treatment modalities [88,89,90]. In the context of ICIs, TAMs can express PD-1 and, upon PD-1 inhibition, anti-tumor immunity is improved in a macrophage-dependent manner [91]. In mice, the number of PD-1-expressing TAMs oscillates diurnally [92]. Administration of an anti-PD-1/PD-L1 agent at ZT18 (during active state), when PD-1-expressing TAMs peak, results in significantly increased phagocytic activity of TAMs and tumor growth suppression compared to ZT6 (during resting state at the TAM PD-1-expression nadir) [92]. Whether this circadian behavior and treatment response by TAMs translate to humans needs further investigation. We can infer that the TAMs’ peak during the activity phase at ZT18 in mice can correspond to the middle of the activity phase of humans at late morning/noon (Figure 4). Furthermore, low levels of monocytes in human peripheral blood during the early morning [93] could suggest an increase in the migration and maturation of macrophage in tissue, increasing the overall number of PD-1-expressing macrophages and thus the effect of anti-PD-1 therapy during the early morning.

Myeloid-derived suppressor cells (MDSCs) are a subset of myeloid cells, often neutrophil- and monocyte-derived, that exhibit potent immunosuppressive activity [94]. MDSCs are a strong contributor to the poor clinical outcomes seen in cancer [95,96]. PD-L1-expressing MDSCs peak in tissues at ZT16 (active phase) in WT mice compared to ZT4 (resting phase). Disruption in the circadian clock gene Bmal1 in intestinal epithelial cells in a mouse model of colorectal cancer abrogated this difference between ZT4 and ZT16 in the intestinal microenvironment, demonstrating that the fluctuation of MDSCs (and their PD-L1 expression) is circadian rhythm-dependent [97]. In the same mouse model, ICI administration at ZT16, when PD-L1+ MDSCs are at their highest, significantly reduces disease burden compared to ICI administration at ZT4 [97]. This response corroborates the time-dependent increase in ICI efficacy during the active phase in other mouse studies. In many cancer types where tumors overexpress PD-L1 and PD-L2, anti-PD-1 therapy significantly improves clinical outcomes compared to patients whose tumors have low PD-L1 and PD-L2 expression [98,99,100]. Altogether, administration timing of ICIs can influence their efficacy based on the oscillatory expression of immune checkpoint molecules on immune cells and tumors, which often peak early during the active phase.

Likewise, the macrophage migration inhibitory factor (MIF) is a diurnally regulated cytokine that peaks during early to late morning [101,102] and upregulates PD-L1 expression in melanoma cells [103]. MIF also plays a role in MDSC differentiation and T cell immunosuppression [104]. In our previous review, we highlighted the role of MIF and its cognate receptor CD74 as potential biomarkers for ICI therapy response and irAE development [105]. ICI administration in the morning may counteract the immunosuppressive functionality of MIF at its peak to reinvigorate the anti-tumor response and improve the clinical response of ICI [106,107].

Circulating tumor cells (CTCs) can express PD-L1, which inhibits the anti-tumor response in circulation and contributes to poor clinical outcomes in multiple cancers [108,109,110,111]. Accordingly, patients with high PD-L1+ CTCs had a higher response rate from ICIs than those who did not [112,113]. This established a correlative mechanism by which the CTCs can affect the efficacy of ICIs. An early study attempted to identify a diurnal oscillation of CTCs and found no significant difference in the level of CTCs at 8:00 versus 20:00 in patients with metastatic breast cancer [114]. A more recent and extensive study evaluated the CTC level at 4:00 (resting phase) and 10:00 (activity phase) in patients with breast cancer and found that the CTC level during the resting phase was higher than during the activity phase [115]. However, the peak level of CTCs was not determined and was postulated to be within the early morning (6:00 to 9:00). Interestingly, the same publication found CTCs exhibited greater metastatic potential during the resting phase in mice. This could be attributed to the observed higher cell proliferation and cell division occurring in the body (and, reasonably, tumors) during sleep [116]. Unfortunately, it is difficult to implement treatment with ICIs during natural rest/sleep times to combat this phenomenon. However, this process was found to be malleable in mice: the proliferative and metastatic potentials of tumors and CTCs were instead greater during the active phase upon insulin administration (upregulating cell growth and division) [115]. In another mouse study, CTC counts reached a peak at ZT13 (activity phase) and nadir at ZT1 (resting phase) [117]. ZT13 in mice (one hour into the activity phase) corresponded closely to early morning in humans (Figure 4). Thus, ICI infusion in the morning could allow the immune system to eliminate CTCs and reduce the risk of metastasis, providing a potential mechanism for improving OS and PFS.

Lymph nodes (LN) are important structures for immune activation and surveillance [118,119]. Murine studies demonstrated that lymphocyte egress and homing to and from the lymph nodes are under diurnal circadian control [120,121]. In mice, B cells and T cells peak in the blood at ZT5 (the middle of the resting phase), then gradually decrease to a trough at ZT13 (the start of the activity phase), and remain low throughout ZT13–20 (during the activity phase) [120]. In the mouse model of experimental autoimmune encephalomyelitis (EAE), disease severity (as a function of immune surveillance and activation) worsened when inoculation happened at ZT8 compared to ZT20, and Th17 cells, the T cell subset responsible for driving the EAE autoimmune response [122,123], were found to be significantly higher in the LN at ZT8 during LN homing compared to ZT20 during LN egress [121]. Thus, lymphocytes within the LN played a greater role than within the peripheral blood in the immune response and induction of EAE. Likewise, B and T cells in the LN were significantly elevated and remained high throughout the activity phase [120]. Additionally, in a mouse melanoma model, tumor size was significantly larger when tumor cells were inoculated at ZT21 (near the end of the activity phase) compared to ZT9–ZT13 (near the star8t of the activity phase) [124]. The number of leukocytes, including naive and activated CD4 and CD8 T cells, significantly increased 24 h later in the draining LN in the cohort inoculated at ZT9, demonstrating the time-of-day influence on leukocyte proliferation as a mechanism for anti-tumor immunity. A higher infiltrate of dendritic cells (DCs) was also found in the lymphatic vessels (LVs) at ZT7 than ZT19 [125] and exhibited greater costimulatory factors in T cell activation at ZT9 [124], when they traveled from LVs to LNs.

In addition to lymphocytes peaking in the LN at ZT13, leukocyte infiltration of tumors also peaked at ZT13 in mice [126]. The adoptive transfer of T cells at ZT13 resulted in a higher infiltration in tumor and better tumor control compared to ZT1, and anti-PD1 administration at ZT13 also enhanced tumor control [126]. Interestingly, the cell number of leukocyte infiltration was approximately equal before (ZT9) and after (ZT18), peaking at ZT13 [126]. This raises the question of the functional role of individual tumor-infiltrating lymphocytes (TILs) during homing versus egress and their functional association to ICIs. The administration of ICIs at ZT9 in this study may have acted to further militarize leukocytes homing to tumors, leading to better tumor control at ZT13, but this remains to be elucidated. Together, these data indicate the relevance of circadian timing in relation to the location, proliferation, and activation of leukocytes during disease inoculation and ICI administration.

Interleukin 10 (IL-10) is a pleiotropic cytokine and is produced by many immune cells, including T cells, B cells, DCs, and NK cells [127,128]. In its pro-inflammatory role, IL-10 is secreted by both Th1 and Th2 cells, while, when it is anti-inflammatory, it is secreted by regulatory T cells (Tregs) [129]. IL-10 is diurnally influenced by the clock gene REV-ERB, a key modulator of the circadian rhythm in mammals [130,131]. In human peripheral blood cytokine studies, high baseline IL-10 levels before ICIs were found to be prognostic of disease progression but also of irAEs [132,133,134] and lower serum and a reduction in IL-10 levels during ICIs were associated with better ICI efficacy [135,136]. IL-10 is known to exhibit an autoregulatory downregulation of MHCII expression in monocytes and limit the production of pro-inflammatory cytokines [130,137]. In addition, IL-10-producing B cells are capable of suppressing T cell proliferation and inducing Treg cells [138]. In fact, pro-inflammatory cytokines peak during nighttime, whereas IL-10 peaks during the day [139]. Thus, the administration of ICIs during the daytime (activity phase) versus nighttime (resting phase) may also play a secondary role in overriding both the anti-inflammatory and pro-inflammatory functionality of IL-10 to improve ICI efficacy while possibly decreasing irAE development. However, studies also highlighted the pleiotropic and contradictory role of IL-10 in the context of irAEs, with an increase in IL-10 during ICI being associated with a reduced risk of irAEs [140], or reduction in IL-10 being associated with an increased risk of irAEs [136], varying over multiple cycles of treatment.

Relatedly, a study on the irAE development based on the calendar season found the highest incidence of irAEs in patients who started ICI during the winter season [141]. The increase in specific irAEs throughout the season was attributed to the extrinsic factors unrelated to ICI treatment, such as the patient lifestyle, environmental condition, and infectious disease that can prime the immune system and influence irAE development [141,142]. Chronotherapy is therefore not limited to time of day and may need to be expanded to include seasonal and environmental fluctuations as well.

Th17 cells are a subset of CD4+ T helper (Th) cells characterized by their production of the cytokine interleukin-17 (IL-17) [143]. In studies of immune-related (ir) toxicities of ICIs, tissues with irColitis and irRash showed an upregulation of CD4+ T cells with IL17A expression, and the subsequent administration of secukinumab (anti-IL-17A) in these patients abrogated their respective irAEs [144,145,146]. Likewise, a higher starting baseline of circulating IL-17 levels significantly increased the risk of developing severe (grade 3+) irColitis/diarrhea [132]. Studies showed that Th17 proliferation and IL-17 secretion are regulated by the circadian clock, in particular, the orphan nuclear receptor RORgt [147]. Th17 cells were found to be higher at ZT16 (activity phase) than at ZT4 (resting phase) in the small intestine in mice [148]. The trafficking of Th17 cells into the small intestine was facilitated by the CCR6/CCL20 axis [149], which was also controlled by the circadian clock [150]. CCL20 mRNA levels in tissues were found to be higher at ZT8 than at ZT20 [148], suggesting the accumulation of CCL20 expression in the small intestine at ZT8 later influenced the tissue homing of Th17 cells at ZT16. The single-cell analysis of immune-related tissues in ICI patients revealed greater infiltration of Th17 cells compared to healthy tissues [144,151,152]. A subset of CD8 T cells in irColitis also shifted toward a Th17 phenotype, as inferred from the increased expression of IL17A, IL26, IL23R, and BATF [151]. An analysis of intestinal group 3 innate lymphoid cells (ILC3s), the innate counterpart of Th17 cells, found RORgt expression was low at ZT13, although there were no differences in the number of ILC3s [153]. Antagonizing RORgt as a transcription factor repressed the Th17-mediated pro-inflammatory function [154]. In summary, Th17 cells promote autoimmune disorders, including irAEs in ICI, and Th17-mediated inflammation is diurnally regulated. Therefore, administering ICI in the early morning could reduce the development of irAEs, potentially due to the lower presence of Th17 cells during this time. Detecting such an effect in patients would depend on identifying the optimal infusion timing window and selecting appropriately granular clinical measures of toxicity (e.g., tissue-specific), as existing retrospective studies relied on heterogeneous timing cutoffs and toxicity-related outcomes.

Melatonin is a hormone released from the pineal gland and functions to synchronize the circadian rhythm and sleep. Melatonin was also shown to modulate both the human immune system and the tumor microenvironment, exerting a pro-immunogenic effect on T cells, Tregs, IL-10, and TAM and MDSC polarization as well as contributing to the regulatory modulation of Th17 cells (extensively reviewed by Mu and Najafi) [155]. The effects of melatonin on the immune system provide another potential mechanism linking the circadian rhythm and sleep to ICI efficacy and toxicity.

In humans, endogenous melatonin levels increase shortly after the start of the dark cycle (18:00), peak during sleep (between 2:00 and 4:00), and gradually return to baseline at awakening (~8:00) [156]. However, the endogenous secretion and exogenous intake of melatonin are different, as the over-the-counter supplement increases melatonin levels as much as 10–100 times higher than the peak level during sleep [156]. Endogenous melatonin regulates sleep, but exogenous melatonin may be used to enhance immune system function. A preliminary clinical study [157] showed that patients with solid cancers had better disease control on a high-dose melatonin/anti-PD-1 combination compared to anti-PD-1 alone. A recent murine study showed that a melatonin/anti-PD-L1 combination reduced tumor growth and metastasis compared to anti-PD-L1 alone after radiofrequency ablation [158]. Melatonin was also shown to inhibit PD-L1 expression in tumor cells [158,159,160,161]. In summary, melatonin is a regulator of sleep and serves as a bridge between the circadian rhythm and the immune system, and these studies support the further investigation of exogenous melatonin as a potential treatment modality to enhance the anti-tumor efficacy of ICIs.

Although drugs with short half-lives are generally considered to be ideal for chronotherapeutic applications, the studies reviewed here suggest that even modalities with longer half-lives (on the order of weeks) may be effective chronotherapeutics. This may seem counterintuitive, as drugs with longer half-lives often obscure circadian effects [162]; however, additional factors play a role, including the time it takes the drug to reach steady-state concentrations in the body and the circadian rhythmicity of target cells or molecules.

Depending on the drug, it takes 4 to 18 weeks for ICIs to reach steady-state concentrations [163]; therefore, there is a crucial pre-steady-state window in which the time of day of administration may be leveraged to maximize effectiveness and improve patient outcomes. In addition, multiple studies reviewed here focused on the time-of-day effect of initial infusions. Specifically, Nomura et al. found that administering initial infusions in the morning resulted in improved patient survival and response rates but that the effect of time-of-day treatment was lost when considering all courses [35]. This provides additional evidence for the special consideration of the time-of-day administration of the first ICI infusions.

Another potential explanation of this paradoxical circadian effect is the rhythmicity of target cells. It was suggested that T cells are more susceptible to activation or stimulation early in the day; therefore, administering ICI therapy when T cells are most ready to be activated may increase efficacy regardless of the half-life of the drug or whether steady state has been reached [164]. While current pharmacokinetic and pharmacodynamic models cannot explain this phenomenon, these ideas provide potential avenues of exploration for future studies.

Our understanding of the circadian rhythm and its impact on the immune system, tumor cells, and drug pharmacology has increased dramatically in the past decade [67]. In the instance of ICIs, as circadian control of the immune system is better understood, it becomes possible to design clinical trials of ICIs that incorporate chronobiology principles, including a patient’s personal chronobiology. Differences in circadian rhythm are common between sexes [67] and between chronotypes (i.e., morning vs. evening person). For example, females exhibit a greater number of rhythmically expressed genes across multiple tissues [165] and a higher amplitude of rhythmic plasma melatonin levels [166] than males. Disrupted sleep/wake cycles also impact personal chronobiology. For instance, inflammatory cytokines were more highly expressed in night-shift workers [167], and exposure to bright light during normal sleeping hours caused a phase-shift in circadian gene expression in the PBMCs of healthy volunteers [168]. These differences may have an important impact on treatment efficacy in chronotherapy for cancer.

Although the time-of-day effect of ICI therapy has not yet been established, there may be opportunities in the future to apply personal chronobiology principles to its administration. Randomized controlled chronotherapy trials including measurements of personal molecular circadian mechanisms were published for other cancer treatment modalities. For example, in a trial [169] evaluating radiotherapy timing (morning or evening) on breast cancer outcomes, it was found that morning treatment is associated with increased late-effect toxicity. However, that correlation was much more significant if patients were stratified by polymorphisms in specific circadian rhythm genes (PER3 and NOCT) [169]. Another trial of 166 patients with glioblastoma investigating the chemotherapy drug temozolomide (TMZ) found a significant difference in efficacy when treatment was given in the morning vs. the evening [170]. The increase in overall survival was driven by patients (N = 56) with methylated (i.e., silenced) promoter regions for MGMT (O-6-Methylguanine-DNA Methyltransferase). MGMT is the protein responsible for repairing TMZ-induced DNA damage, and its expression oscillates throughout the day. The patients who had MGMT expression suppressed at the time of treatment responded better to treatment. These examples show how individual chronobiology can play a critical role in therapy outcomes.

As this field expands, the opportunity will arise to include the evaluation and consideration of patients’ individual circadian rhythms. The current gold standard for identifying an internal circadian phase is dim light melatonin onset (DLMO), a proxy for the phase of the master circadian clock in the SCN [171,172]. The measurement of DLMO requires the individual to remain in clinic for up to 24 h while their melatonin levels are monitored via repeated blood or saliva sampling in a dim-light setting. For various reasons, this is impractical for patients with cancer, and cancer clinics are not set up to carry out these specialized protocols.

Fortunately, various alternatives are being developed. Several recent reviews describe efforts in this field [67,171,173,174]; therefore, we will only briefly highlight these methods. Personal circadian rhythm may be evaluated by wearable devices that take the continuous measurement of parameters such as core or surface body temperature, heart rate, activity, and light exposure [175,176,177]. These have a much lower patient burden compared to DLMO assessment because the devices can be worn at home. Another strategy for evaluating individual circadian time is to infer the circadian phase from a sample taken at a single point in time (snapshot methods). This is advantageous because it can be performed once in the clinic during a regularly scheduled visit. A variety of methods have been developed, including those using one to two samples of easily accessible tissues, fluids, or cell types, such as whole blood or blood-isolated monocytes, hair follicles, oral mucosa, or skin [173,178,179,180], and those applying machine learning methods to identify the molecular chronotype of a tissue sample [181]. Population-based methods combine a traditional around-the-clock sampling scheme with snapshot methods. A small number of individuals are monitored over the course of 24 h, and the remaining individuals are sampled at a single point in time. The single timepoint data are then modeled/matched to the around-the-clock data [182,183].

Although these methods are promising and address some of the practical difficulties of implementing DLMO in a cancer clinic setting, both wearable devices and snapshot methods need optimization to match the performance of DLMO, which remains the gold standard for evaluating personal circadian rhythms. In addition, many of these recently developed approaches have not been validated against DLMO or other gold standard methods, a prerequisite to using them to assess a personal circadian phase in the clinic for chronotherapeutic applications. However, in the absence of information on individual circadian rhythms, population-level data (e.g., sex differences in circadian biology) could be evaluated for their utility in personalizing treatment approaches.

The understanding of the efficacy and toxicity mechanisms behind early ICI infusion and what timing should be considered “early” is important and remains to be elucidated before translation into the clinic. In this section, we highlighted the clinical challenges and logistics, particularly as they pertain to healthcare in the United States, for allocating patients on ICI to exclusively fit in the early/morning period for their infusion.

A typical treatment day for a patient on ICI consists of a lab visit, a consultation visit with a clinician, and a 3 h period for infusion, in that order. A clinician must address patient safety (through bloodwork), symptoms, and concerns before the patient can receive another cycle of treatment. Thus, planning to fit many patients within the “early morning” window can be difficult, especially if clinicians have a fixed clinic schedule. However, these requirements do not need to happen on the same day. It is possible for a patient to obtain their labs, have their provider visits, and then come back the subsequent day to start treatment if the timing happened to be too late during the day. Unfortunately, attending clinic multiple days in a row will be excessively burdensome for some patients, particularly those who need to travel farther for care, have caregiver responsibilities, and/or are not able to take time off work. Patients who are able to visit the clinic multiple days in a row may still face challenges that will increase the burden of care, including increased personal financial costs (such as the cost of staying overnight in a hotel) and increased travel to and from the clinic.

In addition, such a visit schedule could be a logistical challenge for the clinic and providers. Treatment in late afternoons or early evenings may often not be possible due to staff scheduling needs and the need for patient monitoring during infusions. However, some of the studies reviewed here demonstrated better efficacy outcomes when grouping patients based on the timing of initial infusions (first infusion, first four infusions, first 3 months) [35,41]; if future studies corroborate the importance of time-of-day administration for initial infusions more so than total infusions, then initial infusions could be prioritized for morning administration, lowering the burden on the clinic and patients.

Additionally, lifestyle and environmental factors can cause an individual and their chronobiology to shift and vary. For example, what constitutes “lights on” and “lights off” (and thus our activity and resting states) is the natural light–dark cycle from the sun (e.g., sunrise and sunset), which can vary depending on the yearly seasons. In addition, artificial lighting exposure can also have an influence on our circadian clock [184]. Control of one’s circadian rhythm, such as consistent sleep–wake cycles, may also be difficult due to a variety of factors (e.g., age, extent and variability of activity during the wake cycle, and sleep-related needs and symptoms due to disease). As a result, measuring and understanding the influence of chronotherapy is challenging due to the extreme personalized variations in patients.

However, as our knowledge of chronotherapy increases and the effect of early day infusion is clarified, we can foster a paradigm shift toward the inclusion of chronotherapy within personalized medicine. In the future, bloodwork performed before each ICI treatment cycle could be paired with circadian rhythm assessment to optimize delivery of therapy. The inclusion of chronotherapy in ICI treatment regimens will also require clinicians, hospital administrators, and insurance companies to collaborate to mitigate the increase in patient burden caused by more restrictive infusion timing or by increases in clinical assessments (such as circadian rhythm evaluation). Future studies may support the use of melatonin beyond the treatment of sleep difficulties to a tool to align the circadian rhythm to the optimal time window to receive ICI infusion (in addition to the immunogenic benefits of melatonin). Emphasizing sleep patterns and circadian rhythm management may also increase patients’ sense of agency over their care. Adjusting infusion schedules within the healthcare system to prioritize morning infusion timing for ICIs may be a particular challenge, but specific implementation hurdles would depend on the outcome of future research to understand the relative importance of initial infusion timing vs. overall infusions.