Effect of immunotherapy-infusion time of day on survival of patients with advanced cancers: a study-level meta-analysis

Abstract

Background

Immune checkpoint inhibitors (ICIs) have become the standard of care for numerous malignancies. Emerging evidence suggests that the time of day (ToD) of ICI administration could impact the outcomes of patients with cancer. The consistency of ToD effects on ICI efficacy awaits initial evaluation.

Materials and methods

This meta-analysis integrates progression-free survival (PFS) and overall survival (OS) data from studies with a defined ‘cut-off’ ToD. Hazard ratios (HRs) [95% confidence interval (CI)] of an earlier progression or death according to ‘early’ or ‘late’ ToD of ICIs were collected from each report and pooled.

Results

Thirteen studies involved 1663 patients (Eastern Cooperative Oncology Group performance status 0-1, 83%; males/females, 67%/33%) with non-small-cell lung cancer (47%), renal cell carcinoma (24%), melanoma (20%), urothelial cancer (5%), or esophageal carcinoma (4%). Most patients received anti-programmed cell death protein 1 or anti-programmed death-ligand 1 (98%), and a small proportion also received anti-cytotoxic T-lymphocyte-associated protein 4 (anti-CTLA-4) (18%). ToD cut-offs were 13:00 or 14:00 (i.e. ICI median infusion time), for six studies, and 16:00 or 16:30 (i.e. reported threshold for weaker vaccination responses) for seven studies. Pooled analyses revealed that the early ToD groups had longer OS (HR 0.50, 95% CI 0.42-0.58; P < 0.00001) and PFS (HR 0.51, 95% CI 0.42-0.61; P < 0.00001) compared with the late ToD groups.

Conclusions

Patients with selected metastatic cancers seemed to largely benefit from early ToD ICI infusions, which is consistent with circadian mechanisms in immune-cell functions and trafficking. Prospective randomized trials are needed to establish recommendations for optimal circadian timing of ICI-based cancer therapies.

Highlights

• •This meta-analysis assesses the relationship between ToD of ICI infusion and efficacy during metastatic cancer.
• •All 13 retrospective studies meeting the selection criteria were included.
• •Early ToD infusions of ICIs improve OS, PFS, and response rates.
• •These results are consistent with circadian mechanisms in immune-cell functions and trafficking.
• •Randomized clinical and translational studies are needed to conclusively establish the chronotherapy of ICIs.

Introduction

The time of day (ToD) of administration has not been specified nor evaluated in clinical trials assessing the efficacy and adverse events of checkpoint inhibitors (ICIs), but is anticipated to have occurred mostly in the morning.^1,2 However, findings in several retrospective clinical studies have indicated that ICI infusion ToD is usually spread between 08:00 and 18:00, and could affect treatment efficacy. Such results are in line with those from prior randomized chronotherapy trials involving chemotherapy and radiation therapy, as well as recent meta-analyses and systematic reviews.^3,4 Because of the accumulating evidence of ToD effects on the efficacy of ICIs in retrospective clinical studies, we carried out the first meta-analysis for accurately assessing and quantifying the impact of ToD of ICI infusions on their effectiveness in patients with different advanced cancers.

Indeed, circadian rhythms have been shown to rhythmically regulate organismic and cellular functions along the 24-h scale. Modifications of these rhythms have been associated with the development or the worsening of numerous illnesses, including cancer, cardiovascular, neurological and psychiatric diseases, metabolic syndrome, and immune and inflammatory disorders.5, 6, 7, 8

Circadian rhythms are endogenous and genetically based biological variations with an ∼24-h period. They are generated by molecular clocks that consist in feedback transcription-post-transcription and post-translation regulatory loops involving 15 clock genes within each cell.⁹ The discovery of the fine molecular mechanisms responsible for the ticking of molecular circadian clocks across species by Jeffrey Hall, Michael Rosbasch, and Michael Young was acknowledged by the 2017 Nobel Prize in Physiology or Medicine Award.¹⁰ Amazingly, 1 year later, James Allison and Tasuku Honjo won the 2018 Nobel Prize for Physiology or Medicine for discoveries leading to new approaches in harnessing the immune system to fight cancer, such as ICIs.¹⁰

Circadian rhythms help cells and whole organisms to adapt to cyclical changes in their environment, including the alternation of light and darkness over the 24 h in particular.¹¹ The suprachiasmatic nucleus, the central circadian pacemaker in the hypothalamus, receives and integrates photoperiodic information, and coordinates the molecular clocks in the genome of each cell through the generation of rhythmic physiological and hormonal signals, such as those in core body temperature and cortisol secretion. The circadian timing system rhythmically controls drug absorption, distribution, metabolism, and elimination processes at both tissue and whole organism levels, as well as cellular proliferation and death in healthy cells.¹² Moreover, a one-to-one coupling of the circadian clock with the cell cycle has been demonstrated at the single-cell level for proliferating fibroblasts.¹³ Such coupling may be altered in cancer cells,¹⁴ whose molecular clock and/or its regulatory functions can be disrupted.4, 5, 6, 7, 8 Yet, host circadian rhythms can both regulate cancer cell proliferation and their metastatic potential which seems to be greater at night in patients with breast cancer.^8,15, 16, 17, 18

An autonomous molecular circadian clock has also been identified in most immune cells, whose function and trafficking have long been known to display prominent circadian rhythms in rodents and humans.¹⁹ For instance, in humans, proinflammatory cytokines [e.g. interleukin-1 (IL-1) and tumor necrosis factor-α] circulate at higher levels during the rest span at night, whereas anti-inflammatory cytokines (e.g. IL-4 and IL-10) do so during active periods at daytime.^20,21 Furthermore, the release of glucocorticoids, potent immunosuppressors, peaks at the beginning of the day and reaches its lowest level during the first half of the night.²² Concerning cellular immunity, T lymphocytes (CD8+), a main cellular target of ICIs, are subjected to circadian variations, with notably more intense proliferation during the early afternoon; circulation of B-lymphocytes in the blood is enhanced at the end of the evening, while circulating T and T(CD4) lymphocytes peak near the middle of the night, and processes of recognition and destruction of antigens predominate in the very early morning.16, 17, 18^,23,24 The relevance of such circadian orchestration of the immune system for the efficacy of ICIs is emerging through numerous and recent retrospective clinical studies, whose consistency of results is evaluated here in a comprehensive meta-analysis of available evidence.

Materials and methods

This meta-analysis of published data evaluated the impact of the ToD of ICI infusion on the overall survival (OS) of adult patients with metastatic solid tumors, as a primary endpoint. The secondary endpoints were progression-free survival (PFS) and response rate. Incidence of immune-related adverse events was not included as an endpoint, because their rate was inconsistently reported in the studies, thus preventing any meaningful meta-analysis. PubMed and Cochrane databases were used for the search of published articles (15 August 2023), complemented by a manual exploration of abstracts submitted to the annual meetings (including site-specific ones) of the American Society of Clinical Oncology, the European Society for Medical Oncology, the American Association for Cancer Research, and the Society for Immunotherapy of Cancer from 15 February 2020 to 15 February 2023. The search terms used were ‘immunotherapy, Pembrolizumab, Nivolumab, Ipilimumab, Avelumab, Atezolizumab, Tislelizumab, Durvalumab, Dostarlimab, Tremelimumab, Sintilimab, Camrelizumab, Cemiplimab, Zimberelimab, immune checkpoint’, ‘immunotherapy, anti-programmed cell death protein-1 (PD-1) and its ligand anti-programmed death-ligand 1 (PD-L1), or immune checkpoint inhibitor’, ‘cancer survival’, and ‘time-of-day of infusion, circadian rhythm, circadian timing or chronotherapy’ (Supplementary Table S1, available at https://doi.org/10.1016/j.esmoop.2023.102220).

All the study data were extracted independently by two readers (TL and FL). Disagreements were resolved by discussion with a third reader (PFI). Whenever results in a given study population were reported both in abstract form and as an article by the same investigator team, we only considered the data in the article. The following information was collected: the number of patients, the histological type of cancer, the type of ICIs administered, the daily cut-off time applied for differentiating efficacy according to the ToD of ICI infusions, and the proportion of infusions deemed to be needed for weakening ICI efficacy. The efficacy criteria analyzed were OS, PFS, and response rate, as a function of dichotomization of the patient populations according to each study-specific cut-off time and proportion of infusions with late ToD. The main analysis included all types of metastatic cancers, while a secondary analysis specifically addressed ICI timing effect for non-small-cell lung cancer (NSCLC), renal cell carcinoma, and melanoma since at least two independent reports were available for each patient population.

The analyses were conducted according to the Cochrane method for meta-analyses and computed with Review Manager software (RevMan version 5.4; Cochrane, Oxford, UK). The heterogeneity statistic was evaluated with χ² tests and I² statistics, with a χ² test P < 0.05 defining the presence of heterogeneity. I² assesses heterogeneity among studies, with heterogeneity ranking from perfect homogeneity (0%) to moderate heterogeneity (30%-60%) up to total heterogeneity (100%). Begg’s funnel plots were used to visualize the heterogeneity among studies. A fixed-effects model was used to calculate the cumulative relative risk when interstudy heterogeneity was low; when such heterogeneity was high, a randomized model was applied. The meta-analysis results for OS and PFS are reported as hazard ratio (HR) [95% confidence interval (CI)]. All tests were bilateral, with P <  0.05 defining statistical significance.

Results

The search found 16 studies, including 8 articles in peer-reviewed international journals25, 26, 27, 28, 29, 30, 31, 32 and 8 congress abstracts, which were presented as poster communications at major international congresses.33, 34, 35, 36, 37, 38, 39, 40 Three abstracts were discarded because the data⁴⁰ were updated in a subsequently published article,²⁶ or detailed HR data were not reported.^34,35 All 13 retained study reports consisted of retrospective evaluations of ToD ICI administration effects. They involved patients receiving ICIs for metastatic lung cancer,26, 27, 28^,36,39 malignant melanoma,^25,30,32 urothelial cancer,³³ renal cell carcinoma,^29,37,38 and esophageal carcinoma.³¹ ICIs were anti-PD-1 and/or anti-PD-L1, which were combined with anti-cytotoxic T-lymphocyte-associated protein 4 (anti-CTLA-4) ipilimumab for some patients in five studies^{25,29,30,32,37} (Supplementary Figure S1, available at https://doi.org/10.1016/j.esmoop.2023.102220).

The cut-off time defining early or late ToD ICI infusion was given in each study report, along with a threshold value expressed as the proportion of ICI infusions received before or after that ‘cut-off’ time. A difference of up to 4.5 h characterized the defined cut-off times that discriminated early versus late ToD ICI groups among the 13 studies. The percentage of infusions administered after this threshold that was associated with a weak efficacy of ICIs also differed from 20% to 75% between studies. Most studies considered all the immunotherapy cycles,25, 26, 27, 28, 29^,31,33,36, 37, 38 but a few studies only considered the initial ones.^30,31,39

A pragmatic approach was used in five studies which selected 12:00,³⁹ 12:54,²⁶ 13:00,^30,31,37 or 14:00³² as cut-off points corresponding to the median time of the per-patient median times of reported ICI infusions. The percentage of courses associated with a weaker ICI efficacy after the ToD ‘cut-off’ was 50% in one study³⁸ and 25% in another.³⁷ In two studies, ToD groups differed as a function of whether patients received one to four of the initial four ICI infusions before 12:00 or 13:00 versus all four initial infusions after 12:00 or 13:00.^30,39

The ToD ‘cut-off’ was 16:00 for one study³⁶ and 16:30 for six others.^25,27, 28, 29^,33,38 For those seven studies, the early ToD groups consisted of patients receiving fewer than 20% of their ICI infusions after the cut-off time. The patients in the late ToD groups received at least 20% of their ICI infusions after that cut-off time. The cut-off times and proportions in all the studies included in this meta-analysis are summarized in Figure 1.

Figure 1.

Two-dimensional and bubble plots illustrating the sample sizes of studies according to both the daily cut-off time selected for differentiating early versus late ToD groups (abscissa) and the minimum percentage of ICI infusions administered before this cut-off time (ordinate). The diameter of each bubble is proportional to the number of patients included in a single study, or in an aggregate of studies that chose the same criteria for timing cut-off and proportion of ICI infusions before the cut-off time. Corresponding study reference(s) are shown above or below the relevant bubble. Note the occurrence of two clusters of six and seven studies with respective cut-off times at 12:00-14:00 and 16:00-16:30.

ICI, immune checkpoint inhibitor; ToD, time of day.

The current meta-analysis involved 1663 patients, with 1037 patients (62.4%) and 626 patients (37.6%) being allocated to the early and late ToD infusion groups, respectively. The main characteristics of the 13 retained studies are listed in Table 1 and the main patients’ characteristics are listed in Supplementary Table S1, available at https://doi.org/10.1016/j.esmoop.2023.102220. Most importantly, a statistically significant difference favored early ToD administration for OS, with an HR of 0.50 (95% CI 0.42-0.58; P < 0.00001), based on available data from all 13 studies (Figure 2A). The analyses further revealed significant benefits in both secondary endpoints, whenever such data were available. This was the case for PFS, with early ToD infusions being associated with an HR of 0.51 (95% CI 0.42-0.61; P < 0.00001) from available data reported for seven studies26, 27, 28^,30,31,33,39 (Figure 2B). Similarly, response rates were largely improved in the early ToD groups from the six studies where this endpoint was reported (Supplementary Figure S2, available at https://doi.org/10.1016/j.esmoop.2023.102220).

Table 1.

The main characteristics of the studies included in the meta-analysis

Study	Metastatic cancer	ICI	Infusion cut-off time	Number of patients (B/A)	Outcome measure
Country
Method
Period
Ortego et al.33	Urothelial	Anti-PD-1	16:30	88 (62/26)	OS
Spain and Italy		Anti-PD-L1			PFS
Multicenter, retrospective
2016-2021
Qian et al.25	Melanoma	Anti-PD-1	16:30	146 (73/73)	OS
United States		Anti-CTLA-4
Longitudinal
2012-2021
Gonçalves et al.32	Melanoma	Anti-PD-1	14:00	73 (48/25)	OS
Portugal		Anti-CTLA-4
Retrospective
2016-2022
Yeung et al.30	Melanoma	Anti-PD-1	13:00	121 (98/23)	OS
Canada		Anti-CTLA-4			PFS
Retrospective
2015-2021
Karaboué et al.26	NSCLC	Anti-PD-1	12:54	95 (48/47)	OS
France					PFS
Retrospective
2015-2020
Cortellini et al.27	NSCLC	Anti-PD-1	16:30	180 (136/44)	OS
United States, United Kingdom, and Italy					PFS
Multicenter cohort
2018-2022
Barrios et al.36	NSCLC	Anti-PD-1	16:00	129 (86/43)	OS
Brazil		Anti-PD-L1
Real-world database
2018-2021
Vilalta et al.39	NSCLC	Anti-PD-1	12:00	197 (104/93)	OS
Spain					PFS
Retrospective
2015-2020
Rousseau et al.28	NSCLC	Anti-PD-1	16:30	180 (115/65)	OS
France		Anti-PD-L1			PFS
Retrospective
2014-2021
Dizman et al.29	RCC	Anti-PD-1	16:30	135 (89/46)	OS
United States		Anti-CTLA-4
Retrospective
2018-2022
Fernandez-Mañas et al.38	RCC	Anti-PD-1	16:30	56 (32/24)	OS
Spain		Anti-PD-L1
Retrospective
2014-2022
Patel et al.37	RCC	Anti-PD-1	13:00	201 (119/82)	OS
United States		Anti-CTLA-4
Retrospective
2015-2020
Nomura et al.31	Esophagus	Anti-PD-1	13:00	62 (27/35)	OS
Japan					PFS
Retrospective
2017-2022

CTLA-4, cytotoxic T-lymphocyte-associated protein 4; ICI, immune checkpoint inhibitor; NSCLC, non-small-cell lung cancer; OS, overall survival; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; PFS, progression-free survival; RCC, renal cell carcinoma. A indicates the number of patients receiving infusion after the cut-off, while B indicates the number of patients receiving infusion before the cut-off.

Figure 2.

Meta-analysis results for (A) overall survival and (B) progression-free survival according to the early or late ToD group of ICI infusions.

CI, confidence interval; ICI, immune checkpoint inhibitor; ToD, time of day; IV, inverse variance.

Results from the analyses restricted to metastatic NSCLCs also supported early ToD ICI infusions for prolonging both OS [HR 0.53 (95% CI 0.34-0.80); P = 0.003] and PFS [HR 0.55 (95% CI 0.44-0.69); P < 0.00001; Figure 3A and B]. Analyses restricted to metastatic renal cell carcinoma or melanoma also revealed statistically significant differences in OS that favored early ToD ICI infusion groups (Figure 4).

Figure 3.

Meta-analysis results restricted to patients with metastatic non-small-cell lung cancer for (A) overall survival and (B) progression-free survival according to the early or late ToD group of ICI infusions.

CI, confidence interval; ICI, immune checkpoint inhibitor; ToD, time of day; IV, inverse variance.

Figure 4.

Meta-analysis results restricted to patients with (A) metastatic renal cell carcinoma and (B) melanoma for overall survival according to the early or late ToD group of ICI infusions.

CI, confidence interval; ICI, immune checkpoint inhibitor; ToD, time of day; IV, inverse variance.

A subgroup analysis of OS was carried out according to the ToD ‘cut-off’ of the studies. The benefit of early ToD ICI infusions for studies with a cut-off between 12:00 and 14:00 (HR 0.42, 95% CI 0.33-0.53; P < 0.00001) was higher than that of studies with a cut-off between 16:00 and 16:30 (HR 0.58, 95% CI 0.47-0.72; P < 0.00001; subgroup differences: P = 0.05; Supplementary Figure S3, available at https://doi.org/10.1016/j.esmoop.2023.102220).

Assessment of heterogeneity bias among studies showed that a single study²⁶ fell outside the 95% CI (Supplementary Figure S4, available at https://doi.org/10.1016/j.esmoop.2023.102220), with a major effect in favor of the administration of ≥50% ICI infusions in the morning. The exclusion of this study from the meta-analysis did not alter the survival improvement from early ToD ICI infusions, as shown with an HR of 0.52 (95% CI 0.44-0.61; P < 0.00001; Supplementary Figure S5, available at https://doi.org/10.1016/j.esmoop.2023.102220).

Discussion

This meta-analysis of 13 clinical studies reporting the effects of ICI daily timing for OS revealed consistent and statistically significant prolongations of OS (primary endpoint) as well as PFS and objective response rate (secondary endpoints) in those patients with cancer receiving most ICI administrations in the early rather than in the late part of the day. The near doubling of ICI efficacy through early ToD dosing, based on our three endpoints, stood out quite robustly because the patients were receiving single-agent or combined ICIs, as first- or second-line pharmacotherapy, for different cancer types, and in nine countries from four continents. Moreover, all the studies included in this meta-analysis were reported recently and quickly after the first reports (i.e. within the past 2 years).

However, all study designs involved the retrospective assessment of ToD of ICI administration, each constrained by the availability of slots in the treatment delivery day unit, patient preference, interaction with clinical review, and unplanned deferrals or further investigations required before safe administration. Thus, within-patient differences were found for actual ToD of ICI administration along the course of their immunotherapy. Two main different cut-off times were used to differentiate early versus late ToD groups, that is, 16:00-16:30 or 12:00-14:00. Worse efficacy was associated with the administration of ≥20% infusion cycles after 16:00-16:30 on the one hand, or ≥25%-75% infusion cycles after 12:00-14:00 on the other hand. These respective ToD cut-offs were based on results from a vaccination study that revealed less antibody formation in healthy people vaccinated after 16:00-16:30,^24,25 while the 12:00-12:30 cut-off corresponded to the median cut-off of the individual median timing of ICI infusions in the study population, and to a lesser pharmacologic response in some vaccination studies.⁴¹ In our study, results of an exploratory subgroup analysis suggest better efficacy for infusion before 14:00 rather than before 16:00.

Regarding sensitivity analysis, several authors have analyzed the prognostic influence of the timing cut-off²⁵ or of the proportion of infusions beyond the cut-off time,^26,38 as well as the critical role of ICI timing during the initial treatment months.

The prognosis was all the worse as the proportion of patients treated after the cut-off increased. For the timing cut-off, the prognosis impact was variable between 16:00 and 17:00 with a more significant association for a cut-off at 16:15.²⁵

Karaboué et al.²⁶ included a sensitivity study to examine that criterion by analyzing OS and PFS for three patient groups: at least 67% of infusions given before 12:54, at least 33% of the infusions administered before 12:54 and 33% after that time, and at least 67% of administrations after 12:54. Large and statistically significant differences characterized both OS and PFS of these three groups, that is, median OS, 34.2 (not reached–not reached), 15.3 (8.0-22.7), and 12.4 (4.0-20.7) months, respectively (P = 0.003); and median PFS, 11.1 (0.0-25.5), 5.9 (0.0-13.3), and 3.1 (1.6-4.5) months, respectively (P = 0.002). All sensitivity analyses concurred in showing progressively worse efficacy outcomes with late ToD ICI administration, and improved OS through the administration of more courses before the selected ToD cut-off time. Strikingly, two studies have revealed that early ToD ICIs over the initial 2 or 3 treatment months could be most critical for improving efficacy.^30,31

Moreover, despite the retrospective nature of the 13 study reports, it is striking to note that all the results showing an impact of ToD consistently favored early injections and that not a single study showed improved outcomes through late ToD administration. Finally, for one of the studies,³⁶ survival data were immature, but also favored early ToD.

Nonpharmacological factors, potentially accounting for the observed consistent differences in outcomes according to ICI ToD administration, have been raised. These include social factors, related to a preference for morning day hospital visits in wealthier patients, intrinsically expected to have better outcomes,⁴² and a potential associative bias with the number of administered cycles, with patients having an intrinsically immunotherapy-resistant disease more likely to receive a smaller number of administrations, thus more likely to receive a higher proportion of later ToD infusions.²⁷ In the former case, although an imbalance in sociodemographic factors could not be entirely excluded, it would be unlikely to have impacted OS, PFS, or relative risk.⁴³

The mechanisms at work for anti-PD-1 ICI efficacy largely involve T(CD8)- and dendritic cell-based immune responses, the same as those involved in vaccination responses which have been shown to depend on ToD both in mice⁴⁴ and in humans.⁴¹ According to a randomized study involving people aged >65 years, BCG vaccination yielded significantly better humoral responses following dosing in the morning, between 9:00 and 11:00, compared with vaccination in the afternoon, between 15:00 and 17:00.⁴⁵ People vaccinated in the morning also experienced a better response to influenza or severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) vaccines compared with those vaccinated in the afternoon/evening in observational studies.45, 46, 47 Different criteria defined improved vaccination responses in these three studies, that is, greater antibody response^45,46; higher production of interferon-γ, tumor necrosis factor-α, and IL-1β, supporting increased short-term immunity^46,48; or increased long-term immunity with increased frequencies of memory B cells.⁴⁶ Morning vaccination also produced stronger B-cell, T-cell, monocytes, and dendritic cell responses.⁴¹

Experimental studies have revealed that cancer growth was rhythmically regulated by circadian waves of tumor infiltration with tumor-associated macrophages and dendritic cells, while PD-1 expression was also rhythmically controlled by the molecular clock in these cells.^5,49,50 Moreover the trafficking of circulating lymphocytes from blood to lymph nodes was restricted to the rest span of mice at daytime, which would correspond to the rest span at night in humans.⁵⁰ Taken together, these and other results support the enhanced efficacy of anti-PD-1 ICIs following their administration in the morning, when susceptible immune cells are primarily located in the tumor and its draining lymph nodes. This may explain why ICI daily timing of administration could matter so much for efficacy, despite nivolumab and pembrolizumab, for instance, having plasma half-lives of 2–3 weeks,⁵¹ in agreement with comments highlighting the minimal expected impact of drug pharmacokinetics on these circadian-based findings.⁵² Finally, there were not enough data available to allow for an exploration of possible interactions between sex and ICI timing effects. It will be crucial to determine further the relevance of sex, along with hormonal and biological parameters in prospective trials.

Conclusion

Notwithstanding the fact that morning administrations of ICIs could already be advised at basically no cost and arguably no risk, further randomized clinical and translational studies are needed to conclusively establish recommendations for personalized chronotherapy of ICI-based treatments, to enhance efficacy through patient-specific ToD administration.

Supplementary data

Supplementary Fig S1.

Supplementary Fig 4.

Supplementary Fig 5.