Circadian timing of first-line chemoimmunotherapy is associated with survival in extensive-stage small cell lung cancer

ABSTRACT

Growing evidence suggests that circadian timing influences the efficacy of immune checkpoint inhibitors across multiple malignancies. However, evidence in extensive-stage small cell lung cancer (ES-SCLC) remains limited to a single East Asian cohort, and independent validation in other populations is lacking. We conducted a retrospective single-center cohort study of 107 patients with ES-SCLC treated with first-line platinum–etoposide plus atezolizumab or durvalumab. The time of day of administration (ToDA) was defined as the median start time of the first four immunotherapy infusions, and patients were categorized according to the cohort median ToDA. The primary endpoints were progression-free survival (PFS) and overall survival (OS). The cohort median ToDA was 11:54. Patients treated before the median ToDA had significantly longer median PFS (9.07 vs. 7.23 months, p = 0.002) and OS (21.17 vs. 11.43 months, p = 0.002) than those treated later. In multivariable Cox regression analysis, later administration remained independently associated with shorter PFS (adjusted HR 1.63, 95% CI 1.03–2.58; p = 0.038) and OS (adjusted HR 1.65, 95% CI 1.00–2.71; p = 0.049), whereas objective response rates, treatment-related toxicity, and immune-related adverse events were comparable between groups. Morning administration of first-line chemoimmunotherapy was associated with significantly longer survival without increased toxicity in patients with ES-SCLC. These findings support growing evidence that circadian timing may influence immune checkpoint inhibitor efficacy and warrant prospective evaluation of treatment timing as a readily modifiable strategy to optimize immunotherapy delivery.

1. Introduction

Despite the addition of immune checkpoint inhibitors (ICIs) to platinum-etoposide as first-line therapy, extensive-stage small cell lung cancer (ES-SCLC) remains a highly lethal disease, with a median overall survival (OS) of approximately 12-13 months in pivotal trials such as IMpower133 ¹ and CASPIAN. ² Therefore, the discovery of strategies that modestly improve the efficacy of currently available regimens without adding toxicity or cost is of considerable clinical interest.

Circadian rhythms govern multiple aspects of host immunity that are relevant to ICI activity, including the trafficking of T-cells into lymph nodes, ³ recruitment of cytotoxic T-cells into tumors, ⁴ diurnal regulation of PD-1 expression on tumor-associated macrophages, ^{5 , 6} and rhythmic accumulation of immunosuppressive myeloid populations. ⁷ These observations have given rise to the chronotherapy hypothesis, which states that aligning ICI administration with the active phase of the immune system could amplify therapeutic efficacy. ⁸

Empirical support for this hypothesis is substantial. The landmark MEMOIR study, which used propensity score matching in 299 melanoma patients, showed that receiving fewer than 20% of infusions after 16:30 was associated with near doubling of median OS. ⁹ Subsequent retrospective cohorts have replicated this finding across multiple tumor types and ICI agents, including non-small cell lung cancer (NSCLC), ^10-12 renal cell carcinoma, ¹³ melanoma, urothelial, biliary tract, esophageal and gastric cancers. ^14-17

A 2024 study-level meta-analysis encompassing 13 studies and 1,667 patients reported a pooled OS hazard ratio (HR) of approximately 0.50, favoring morning administration. ¹⁷ The most recent 2026 systematic review and meta-analysis of 29 studies and 6.129 patients further confirmed this finding (pooled HR for OS 0.60, 95% CI 0.51–0.70; pooled HR for progression-free survival (PFS) 0.62, 95% CI 0.54–0.71). ¹⁸

Two recent investigations have substantially strengthened the evidence base. First, in a bi-continental study of 713 patients with advanced NSCLC receiving first-line immuno-chemotherapy in France and China, Huang et al. identified 11:30 as the optimal cut-off time for the time-of-day of administration (ToDA), with morning infusions associated with an adjusted OS HR of 0.47 (95% CI 0.37–0.60). ¹⁹ Second, a U.S. Veterans Health Administration target-trial emulation in 4,688 NSCLC patients applied marginal structural modeling with a chemotherapy negative-control cohort and reported that afternoon infusions were associated with worse OS (HR: 1.15, 95% CI 1.04–1.26, p = 0.004), an effect notably absent in chemotherapy-only controls, providing strong causal evidence that the phenomenon is specific to ICI biology rather than reflecting non-specific confounding. ²⁰ Currently, no published randomized controlled trials have confirmed these observational findings.

Evidence specifically in ES-SCLC is limited to a single Chinese cohort of 397 patients (LungTime-R02), which identified 15:00 as the optimal ToDA threshold, with morning infusions associated with substantially improved OS (HR: 0.37, 95% CI 0.26–0.53) and PFS (HR: 0.48, 95% CI 0.36–0.65). ²¹ Whether this finding is applicable to patients with ES-SCLC managed in different healthcare systems, with different ethnic backgrounds, infusion scheduling practices, and co-medication patterns, is currently unknown. Single-center observational thresholds may reflect institutional scheduling patterns rather than universal biological cut-offs. ^{20 , 22} Consistent with this, heterogeneity in the optimal ToDA thresholds reported across studies has been increasingly recognized. ²² External validation in independent cohorts therefore remains an important objective. ²³

In this single-center Turkish cohort study, we investigated the association between the ToDA of first-line chemoimmunotherapy and clinical outcomes in patients with ES-SCLC. The primary endpoints were OS and PFS; secondary endpoints included objective response rate (ORR) and treatment-related toxicity.

2. Materials and methods

2.1. Study design and patients

We conducted a retrospective cohort study at a single tertiary oncology referral center. Consecutive patients with histologically or cytologically confirmed ES-SCLC treated with first-line platinum (carboplatin or cisplatin) and etoposide combined with an immune checkpoint inhibitor (atezolizumab or durvalumab) between January 2018 and May 2026 were screened for inclusion (Figure S1). Patients were eligible if they had received at least two immunotherapy infusions with documented start times; for patients receiving fewer than four induction cycles, ToDA was calculated from all available induction infusions.

The study was approved by the Istanbul Medipol University Non-Interventional Clinical Research Ethics Committee (Ethics Committee Decision No: 819/14.05.2026). The requirement for written informed consent was waived by the Ethics Committee because of the retrospective nature of the study and the use of anonymized clinical data. The study was conducted in accordance with the principles of the Declaration of Helsinki and reported according to the STROBE guideline for observational studies. ²⁴

2.2. Treatment and follow-up

All patients received standard first-line treatment as per institutional practice and current guidelines. ^{1 , 2} The combined chemoimmunotherapy regimen consisted of carboplatin or cisplatin plus etoposide administered every three weeks for up to four cycles, followed by maintenance immunotherapy until disease progression, unacceptable toxicity, or patient/physician decision. Prophylactic cranial irradiation was offered to selected patients with chemoimmunotherapy response, in accordance with international guidelines. ^{25 , 26} Tumor assessment was performed at baseline and every two cycles using contrast-enhanced computed tomography (and brain magnetic resonance imaging when clinically indicated); the best overall response was classified according to RECIST version 1.1. ²⁷ Follow-up data, including progression status, dates and survival outcomes, were extracted from electronic medical records.

2.3. Definition of time-of-day of administration

For each patient, the start time of every immunotherapy infusion during the first four cycles was extracted from chemotherapy infusion nursing electronic records. ToDA for each patient was calculated as the median of the recorded start times of the first up to four immunotherapy infusions, expressed in decimal hours, consistent with previous ToDA studies, including the LungTime-R02 methodology. ^{10 , 19 , 21} The median was chosen to minimize the influence of occasional scheduling variations between treatment cycles, thereby reducing potential exposure misclassification and better representing each patient's typical infusion time. We focused on the first four cycles for three reasons: (i) these constitute the standard induction phase of first-line chemoimmunotherapy in ES-SCLC, after which response assessment is performed and treatment continues with maintenance immunotherapy alone ²⁵ ; (ii) this early treatment period is considered critical for priming antitumour immunity ¹⁹ ; (iii) it precedes prolonged maintenance exposure and the attainment of pharmacokinetic steady state for anti-PD-L1 agents. ^{19 , 28}

Patients were dichotomized into a “Morning” group (median ToDA < 11.9 hours, corresponding to <11:54) and an “Afternoon” group (median ToDA ≥ 11.9 hours). The 11.9-hour threshold represented the median ToDA of the study cohort and was therefore determined independently of survival outcomes. The resulting threshold was also broadly consistent with the 11:30 cut-off reported in the LungTime bi-continental study. ¹⁹

2.4. Endpoints

The primary endpoints were OS and PFS. OS was defined as the time from the start of the first ICI infusion to death from any cause; PFS was defined as the time from the start of the first ICI infusion to the first documented disease progression or death from any cause, whichever occurred first. Patients without an event of interest were censored at the date of last follow-up. Secondary endpoints were objective response rate (ORR; complete or partial response) and disease control rate (DCR; complete response, partial response, or stable disease) according to RECIST version 1.1, ²⁷ and treatment-related toxicity. Treatment-related adverse events were retrospectively identified from routine clinical documentation and graded according to the Common Terminology Criteria for Adverse Events (CTCAE), version 5.0. ²⁹ Immune-related adverse events (irAEs) were identified based on the treating physician's clinical documentation.

2.5. Baseline covariates

Baseline patient and disease characteristics were extracted from medical records at the date of treatment initiation. These included demographic variables (age, sex), comorbidities (hypertension, diabetes mellitus, ischemic heart disease, chronic obstructive pulmonary disease), smoking status, Eastern Cooperative Oncology Group (ECOG) performance status (PS), metastatic involvement (liver, brain, bone, distant lymph node, adrenal), recurrent versus de novo metastatic status, choice of platinum agent (carboplatin vs cisplatin), choice of immunotherapy agent (atezolizumab vs durvalumab), receipt of prophylactic cranial irradiation (PCI), and the number of immunotherapy cycles delivered.

2.6. Statistical analysis

Baseline characteristics were compared between Morning and Afternoon groups using Mann- Whitney U tests for continuous variables and Pearson’s chi-square or Fisher’s exact tests for categorical variables, as appropriate. ORR and DCR were also compared between groups using Fisher’s exact test.

Median follow-up time was estimated using the reverse Kaplan–Meier method, ³⁰ in which the censoring indicator is reversed and a standard Kaplan–Meier estimator is applied. OS and PFS curves were constructed using the Kaplan–Meier method and compared between groups using the two-sided log-rank test.

Univariable Cox proportional hazards regression models were used to estimate HRs with 95% confidence intervals (CIs) for associations between candidate prognostic factors and PFS or OS. Variables found to be significantly associated with the outcome in univariable analysis (p < 0.05) were subsequently entered into multivariable Cox proportional hazards regression models, alongside the time-of-day group as the variable of primary interest, to estimate adjusted hazard ratios (aHR). Post-baseline variables, including objective response and treatment-related adverse events, were excluded from the multivariable models because treating variables occurring after treatment initiation as fixed covariates could introduce guarantee-time bias.

Infusion-time data were complete for 101 of 107 patients (94.4%). In the remaining six patients (5.6%), ToDA was calculated as the median of the available recorded infusion times because induction treatment was discontinued before completion of four ICI cycles (n = 2) or infusion start times were unavailable for one or two planned induction cycles in nursing records (n = 4). No imputation was performed for missing infusion-time data. All baseline demographic, comorbidity, disease, and treatment variables were complete for all patients.

To assess the extent to which the association between ToDA and survival depended on the binary classification based on the cohort median (<11:54 vs ≥11:54), a cut-point sensitivity analysis was performed. Median ToDA was re-dichotomized using alternative thresholds at 30-minute increments between 11:00 and 13:00 (11:00, 11:30, 12:00, 12:30, and 13:00), and for each threshold separate univariable Cox proportional hazards models were fitted for PFS and OS to estimate HRs with 95% CIs and corresponding p values. These analyzes were intended to assess the dependence of the observed association on threshold selection rather than to identify an optimal clock-time cut-off.

Because ToDA classification incorporated infusion-time information obtained during the first two to four immunotherapy administrations, exposure ascertainment extended beyond treatment initiation, potentially introducing immortal-time (guarantee-time) bias. To address this concern, a landmark analysis was performed. The primary landmark was set at 3 months after treatment initiation (approximately 9–13 weeks), corresponding approximately to completion of the fourth treatment cycle under the standard 3-weekly dosing schedule. Because exact patient-specific fourth-cycle completion dates were not available, the 3-month landmark represented a regimen-based approximation rather than an individual cycle-completion date. For each endpoint, patients who experienced the corresponding event before the landmark (progression or death for PFS; death for OS) were excluded, and only patients remaining at risk at the landmark were retained. Patients retained their original Morning/Afternoon ToDA classification derived from the induction-phase infusions, and PFS and OS were recalculated from the landmark onward. Kaplan–Meier estimation, the log-rank test, and univariable Cox proportional hazards regression were used to compare the groups. The analysis was repeated using 6- and 9-month landmarks as additional robustness analyzes. All statistical analyzes were performed using IBM SPSS Statistics software, version 27.0 (IBM Corp., Armonk, NY, USA). All tests were two-sided, and a p value < 0.05 was considered statistically significant.

3. Results

3.1. Patient characteristics

Between January 2018 and May 2026, 107 patients with ES-SCLC received first-line platinum–etoposide combined with atezolizumab (n = 83, 77.6%) or durvalumab (n = 24, 22.4%). Median age was 63 years (interquartile range [IQR] 56–70), and 73 patients (68.2%) were male. ECOG PS was 0 in 82 patients (76.6%), 1 in 20 (18.7%), and 2 in 5 (4.7%). Liver, bone, and brain metastases were present in 41 (38.3%), 66 (61.7%), and 26 (24.3%) patients, respectively; 35 patients (32.7%) received PCI, and the median number of immunotherapy cycles delivered was 10 (IQR 7–17). Detailed baseline characteristics stratified by ToDA group are presented in Table 1. Baseline characteristics were generally comparable between Morning (n = 56) and Afternoon (n = 51) groups, although statistically significant differences were observed in the presence of bone metastases (p = 0.027), recurrent versus de novo disease status (p = 0.038), and platinum backbone selection (p = 0.037).

Table 1.

Baseline patient, disease and treatment characteristics of 107 patients with extensive-stage small cell lung cancer, overall and stratified by time-of-day of administration (ToDA) group.

Characteristic	Overall (N = 107)	Morning (<11:54, n = 56)	Afternoon (≥11:54, n = 51)	P
Demographics
Age, median (IQR), years	63 (56–70)	−	−	NS
Sex, n (%)				0.741
Male	73 (68.2)	39 (69.6)	34 (66.7)
Female	34 (31.8)	17 (30.4)	17 (33.3)
ECOG performance status, n (%)				0.224
0	82 (76.6)	46 (82.1)	36 (70.6)
1	20 (18.7)	9 (16.1)	11 (21.6)
2	5 (4.7)	1 (1.8)	4 (7.8)
Smoking status, n (%)				0.050
Current smoker	33 (30.8)	20 (35.7)	13 (25.5)
Ex-smoker	66 (61.7)	35 (62.5)	31 (60.8)
Never smoker	8 (7.5)	1 (1.8)	7 (13.7)
Comorbidities
Hypertension	51 (47.7)	28 (50.0)	23 (45.1)	0.612
Diabetes mellitus	24 (22.4)	11 (19.6)	13 (25.5)	0.469
Ischemic heart disease	32 (29.9)	18 (32.1)	14 (27.5)	0.597
Chronic obstructive pulmonary disease	27 (25.2)	13 (23.2)	14 (27.5)	0.614
Disease characteristics
Disease status, n (%)				0.038
De novo metastatic	99 (92.5)	49 (87.5)	50 (98.0)
Recurrent	8 (7.5)	7 (12.5)	1 (2.0)
Metastatic sites, n (%)
Liver	41 (38.3)	17 (30.4)	24 (47.1)	0.076
Bone	66 (61.7)	29 (51.8)	37 (72.5)	0.027
Brain	26 (24.3)	14 (25.0)	12 (23.5)	0.859
Distant lymph node	29 (27.1)	15 (26.8)	14 (27.5)	0.938
Adrenal	17 (15.9)	10 (17.9)	7 (13.7)	0.559
Treatment characteristics
Chemotherapy backbone, n (%)				0.037
Carboplatin	100 (93.5)	55 (98.2)	45 (88.2)
Cisplatin	7 (6.5)	1 (1.8)	6 (11.8)
Immunotherapy agent, n (%)				0.258
Atezolizumab	83 (77.6)	41 (73.2)	42 (82.4)
Durvalumab	24 (22.4)	15 (26.8)	9 (17.6)
ICI cycles, median (IQR)	10 (7–17)	—	—	NS
Prophylactic cranial irradiation, n (%)	35 (32.7)	17 (30.4)	18 (35.3)	0.587
Response and safety
Best overall response, n (%)				0.246
Complete response	11 (10.3)	8 (14.3)	3 (5.9)
Partial response	77 (72.0)	38 (67.9)	39 (76.5)
Stable disease	5 (4.7)	4 (7.1)	1 (2.0)
Progressive disease	14 (13.1)	6 (10.7)	8 (15.7)
Objective response rate, n (%)	88 (82.2)	46 (82.1)	42 (82.4)	1.000
Disease control rate, n (%)	93 (86.9)	50 (89.3)	43 (84.3)	0.446
Any-grade toxicity, n (%)	85 (79.4)	44 (78.6)	41 (80.4)	0.816
Grade 3–4 toxicity, n (%)	27 (25.2)	13 (23.2)	14 (27.5)	0.649
Immune-related adverse event, n (%)	8 (7.5)	5 (8.9)	3 (5.9)	0.718

P values in bold indicate statistical significance (P < 0.05). —, group-specific median (IQR) not displayed; P value reflects the comparison between Morning and Afternoon groups. Objective response rate = complete response + partial response, according to RECIST v1.1. Disease control rate = complete response + partial response + stable disease. NS, not statistically significant. ECOG, Eastern Cooperative Oncology Group; ICI, immune checkpoint inhibitor; IQR, interquartile range; RECIST, Response Evaluation Criteria in Solid Tumors; ToDA, time-of-day of administration.

3.2. Time-of-day of administration distribution and follow-up

The median ToDA across the cohort was 11:54 (mean 12.03 h; standard deviation 1.53 h), with the distribution strongly skewed toward morning administration: only 5.6% (n = 6) of patients received infusions at or after 15:00, and 70.1% (n = 75) received their infusions before 12:30 (Figure 1). Patients were dichotomized into Morning (median ToDA < 11.9 h; n = 56) and Afternoon (≥11.9 h; n = 51) groups. At data cut-off, 85 progression events (79.4%) and 72 deaths (67.3%) had occurred; median follow-up estimated by the reverse Kaplan–Meier method was 29.4 months (95% CI, 20.0–39.8). ³⁰

Figure 1.

Two-panel figure showing the distribution and within-patient consistency of immune checkpoint inhibitor infusion timing in 107 patients. Panel A shows the distribution of per-patient median time of day of administration (ToDA), with a vertical dashed line marking the cohort median cut-off at 11:54. Patients before the cut-off are classified as Morning (n = 56) and those at or after the cut-off as Afternoon (n = 51), with most median infusion times concentrated around midday. Panel B shows individual infusion times across the first up to four treatment cycles for each patient, grouped as All Morning (n = 20), All Afternoon (n = 22), or Mixed (n = 65) according to whether infusion times remained on one side of or crossed the 11:54 cut-off.

3.3. Response

Objective response according to RECIST v1.1 was observed in 88 patients (82.2%), with virtually identical rates in the Morning (46/56, 82.1%) and Afternoon (42/51, 82.4%) groups (Fisher’s exact p > 0.05). DCR was also comparable (89.3% vs 84.3%; p = 0.446) (Table 1).

3.4. Progression-free survival

Median PFS was significantly longer in the Morning group than in the Afternoon group: 9.07 months (95% CI 7.32–10.81) versus 7.23 months (95% CI 6.49–7.98); log-rank p = 0.002 (Figure 2A). In univariable Cox regression analysis, afternoon ToDA was associated with shorter PFS (HR 1.93, 95% CI 1.25–2.98; p = 0.003) (Table 2 and Figure 3). In multivariable Cox regression, afternoon ToDA remained independently associated with shorter PFS (adjusted HR [aHR] 1.63, 95% CI 1.03–2.58; p = 0.038). Liver metastasis was also independently associated with shorter PFS (aHR 1.86, 95% CI 1.11–3.12; p = 0.019). Complete multivariable results for PFS are presented in Table 3 and Figure 4A.

Figure 2.

Two-panel Kaplan–Meier figure comparing survival outcomes between Morning (n = 56) and Afternoon (n = 51) time-of-day administration groups. Panel A shows progression-free survival, with median PFS of 9.07 months in the Morning group and 7.23 months in the Afternoon group (log-rank P = 0.002). Panel B shows overall survival, with median OS of 21.17 and 11.43 months, respectively (log-rank P = 0.002). In both panels, the Morning survival curve remains generally above the Afternoon curve. Shaded areas represent 95% confidence intervals, crosses indicate censored observations, and number-at-risk tables are displayed below the curves.

Table 2.

Univariable analysis of factors associated with progression-free survival (PFS) and overall survival (OS).

Variable	Comparison	PFS HR (95% CI)	P	OS HR (95% CI)	P
ToDA group	Afternoon vs Morning	1.93 (1.25–2.98)	0.003	2.08 (1.29–3.34)	0.003
Demographics
Sex	male vs female	1.05 (0.66–1.67)	0.850	1.05 (0.64–1.74)	0.838
Comorbidities
Hypertension	no vs yes	1.27 (0.83–1.94)	0.279	0.87 (0.54–1.39)	0.554
Diabetes mellitus	no vs yes	0.84 (0.51–1.38)	0.489	0.61 (0.35–1.04)	0.070
Ischemic heart disease	no vs yes	0.93 (0.59–1.47)	0.750	0.56 (0.34–0.93)	0.024
COPD	no vs yes	1.17 (0.72–1.93)	0.524	0.76 (0.45–1.27)	0.290
Smoking status	ex–smoker vs current	0.81 (0.50–1.30)	0.378	0.76 (0.45–1.28)	0.295
	Non–smoker vs current	4.20 (1.83–9.66)	<0.001	2.01 (0.84–4.77)	0.115
Disease characteristics
ECOG PS	1 vs 0	1.21 (0.68–2.13)	0.516	2.40 (1.31–4.39)	0.005
	2 vs 0	3.88 (1.49–10.13)	0.006	7.60 (2.80–20.59)	<0.001
Liver metastasis	no vs yes	0.51 (0.33–0.80)	0.003	0.50 (0.31–0.81)	0.005
Bone metastasis	no vs yes	0.54 (0.34–0.86)	0.009	0.45 (0.27–0.76)	0.003
Brain metastasis	no vs yes	1.47 (0.86–2.53)	0.157	1.51 (0.84–2.73)	0.170
Lymph node metastasis	no vs yes	0.86 (0.54–1.38)	0.538	0.80 (0.48–1.35)	0.402
Adrenal metastasis	no vs yes	0.87 (0.49–1.54)	0.625	1.23 (0.63–2.42)	0.541
Disease onset	Denovo metastatic vs recurrent	1.00 (0.46–2.19)	0.991	1.17 (0.51–2.73)	0.709
Treatment characteristics
Chemotherapy regimen	cisplatin vs carboplatin	0.79 (0.32–1.96)	0.608	0.83 (0.33–2.10)	0.696
ICI agent	durvalumab vs atezolizumab	0.60 (0.35–1.04)	0.068	0.38 (0.17–0.83)	0.016
Any–grade toxicity	no vs yes	1.62 (0.97–2.72)	0.064	0.98 (0.54–1.80)	0.958
Immune–related AE	no vs yes	4.56 (1.43–14.48)	0.010	4.82 (1.18–19.74)	0.029
Objective response	no vs yes	3.04 (1.79–5.14)	<0.001	1.89 (1.04–3.45)	0.036
PCI	no vs yes	1.00 (0.63–1.56)	0.985	0.79 (0.49–1.28)	0.336

P values in bold indicate statistical significance (P < 0.05). The ToDA group row (shaded) is the primary variable of interest. AE, adverse event; CI, confidence interval; COPD, chronic obstructive pulmonary disease; ECOG-PS, Eastern Cooperative Oncology Group-Performance status; HR, hazard ratio; ICI, immune checkpoint inhibitor; OS, overall survival; PCI, prophylactic cranial irradiation; PFS, progression-free survival; ToDA, time-of-day of administration.

Figure 3.

Two-panel forest plot showing univariable Cox regression analyses of factors associated with progression-free survival (PFS) and overall survival (OS) in 107 patients. Panel A shows PFS; significant associations are observed for never-smoking versus current smoking, ECOG performance status 2 versus 0, absence versus presence of liver metastasis, absence versus presence of bone metastasis, absence versus presence of immune-related adverse events, and absence versus presence of objective response. Panel B shows OS; significant associations are observed for absence versus presence of ischemic heart disease, ECOG performance status 1 and 2 versus 0, absence versus presence of liver and bone metastases, durvalumab versus atezolizumab, absence versus presence of immune-related adverse events, and absence versus presence of objective response. Squares show hazard-ratio estimates with horizontal 95% confidence intervals; the vertical dashed line marks a hazard ratio of 1, and statistically significant associations are highlighted in red.

Table 3.

Multivariable Cox proportional hazards regression models for progression-free survival (PFS) and overall survival (OS).

Variable	Comparison	PFS aHR(95% CI)	P	OS aHR(95% CI)	P
ToDA group	Afternoon vs Morning	1.63 (1.03–2.58)	0.038	1.65 (1.00–2.71)	0.049
Smoking status			0.002
	Current vs never-smoker	0.34 (0.15–0.79)	0.012
	Ex-smoker vs never-smoker	0.22 (0.10–0.52)	0.001
ECOG PS			0.257		0.001
	0 vs 2	0.49 (0.18–1.31)	0.155	0.17 (0.06–0.46)	0.001
	1 vs 2	0.40 (0.14–1.19)	0.100	0.31 (0.10–0.97)	0.044
Liver metastasis	Yes vs No	1.86 (1.11–3.12)	0.019	1.27 (0.73–2.21)	0.395
Bone metastasis	Yes vs No	1.37 (0.82–2.29)	0.231	1.76 (0.99–3.13)	0.054
Immunotherapy agent	Atezolizumab vs durvalumab	1.25 (0.71–2.21)	0.448	1.82 (0.81–4.10)	0.149

P values in bold indicate statistical significance (P < 0.05). The ToDA group row (shaded) is the primary variable of interest. Blank cells indicate the variable was not included in that model, as it was not significant in univariable analysis for that outcome. aHR, adjusted hazard ratio; CI, confidence interval; ECOG PS, Eastern Cooperative Oncology Group -Performance Status; OS, overall survival; PFS, progression-free survival; ToDA, time-of-day of administration.

Figure 4.

Two-panel forest plot showing multivariable Cox regression models for progression-free survival (PFS) and overall survival (OS). Afternoon versus Morning time-of-day of administration (ToDA), the primary variable of interest, is associated with higher adjusted hazards for both PFS (aHR 1.63, 95% CI 1.03–2.58; P=0.038) and OS (aHR 1.65, 95% CI 1.00–2.71; P=0.049). In Panel A, current and former smoking versus never-smoking and the presence of liver metastasis are also statistically significant. In Panel B, ECOG performance status 0 and 1 versus 2 are also statistically significant. Diamond markers indicate the ToDA effect, squares indicate other covariates, horizontal lines show 95% confidence intervals, and the vertical dashed line marks an adjusted hazard ratio of 1.

3.5. Overall survival

Median OS was significantly longer in the Morning group than in the Afternoon group: 21.17 months (95% CI 14.76–27.58) versus 11.43 months (95% CI 9.18–13.69); log-rank p = 0.002 (Figure 2B). In univariable Cox regression analysis, afternoon ToDA was associated with shorter OS (HR 2.08, 95% CI 1.29–3.34; p = 0.003) (Table 2 and Figure 3). In multivariable Cox regression, afternoon ToDA remained independently associated with shorter OS (aHR 1.65, 95% CI 1.00–2.71; p = 0.049). ECOG performance status was independently associated with OS (overall p = 0.001), with lower hazards observed for ECOG PS 0 (aHR 0.17, 95% CI 0.06–0.46; p = 0.001) and ECOG PS 1 (aHR 0.31, 95% CI 0.10–0.97; p = 0.044) compared with ECOG PS 2. Complete multivariable results for OS are presented in Table 3 and Figure 4B.

3.6. Sensitivity and landmark analyzes

In cut-point sensitivity analyzes using alternative ToDA thresholds between 11:00 and 13:00, effect estimates varied according to the selected threshold (Figure S2 and Table S1). For PFS, a ToDA < 11:00 was associated with a lower hazard than a ToDA ≥ 11:00 (HR 0.60, 95% CI 0.36–0.99; p = 0.048), whereas the associations at the 11:30, 12:00, 12:30, and 13:00 thresholds were not statistically significant. For OS, none of the alternative thresholds reached statistical significance.

Landmark analyzes showed a consistent association between afternoon ToDA and poorer survival (Figure S3 and Table S2). At the 3-month landmark, 105 patients remained at risk for PFS and 106 for OS. Afternoon ToDA was associated with shorter PFS (HR 2.06, 95% CI 1.33–3.20; p = 0.001) and OS (HR 2.16, 95% CI 1.34–3.49; p = 0.002). Similar associations were observed at the 6-month landmark for PFS (HR 2.02, 95% CI 1.21–3.37; p = 0.008) and OS (HR 2.20, 95% CI 1.32–3.66; p = 0.002). At the 9-month landmark, the association remained significant for OS (HR 2.15, 95% CI 1.21–3.83; p = 0.009), whereas the PFS estimate remained directionally consistent but was not statistically significant (HR 1.58, 95% CI 0.64–3.89; p = 0.323).

3.7. Safety

Any-grade treatment-related toxicity was reported in 85 patients (79.4%), without significant difference between Morning and Afternoon groups (78.6% vs 80.4%; p = 0.816). Grade 3–4 toxicity was observed in 27 patients (25.2%) and immune-related adverse events in 8 patients (7.5%), with similar distributions between groups (p = 0.649 and p = 0.718, respectively) (Table 1). No unexpected safety signals were observed.

4. Discussion

In this single-center retrospective cohort of 107 patients with ES-SCLC treated with first-line platinum–etoposide and an immune checkpoint inhibitor, ToDA of first-line chemoimmunotherapy was significantly associated with survival outcomes. Patients with a median ToDA of the first up to four ICI infusions before 11:54 experienced longer median PFS (9.07 vs 7.23 months) and OS (21.17 vs 11.43 months) than those treated later in the day. These associations remained significant after multivariable adjustment, with an aHR of 1.63 (95% CI 1.03–2.58) for PFS and 1.65 (95% CI 1.00–2.71) for OS in the Afternoon group. Beyond externally validating the directional association between earlier treatment timing and improved survival reported by Huang et al. ²¹ , our findings corroborate and extend the emerging evidence that circadian timing may represent a clinically relevant and potentially modifiable determinant of immune checkpoint inhibitor efficacy across thoracic malignancies.

Our findings are consistent with a growing body of evidence supporting the morning advantage of ICI administration across cancer types. The most recent systematic review and meta-analysis by Inoue et al. of 29 studies and 6,129 patients reported pooled HR of 0.60 for OS and 0.62 for PFS in favor of early administration, ¹⁸ confirming the findings of an earlier study-level meta-analysis by Landré et al. including 13 studies and 1,667 patients. ¹⁷ The largest single dataset published to date, a target-trial emulation by Gonzalez et al. analyzing 4,688 NSCLC patients in the U.S. Veterans Health Administration, applied marginal structural modeling with a chemotherapy negative-control cohort and demonstrated that afternoon administration was associated with worse OS (HR 1.15, 95% CI 1.04-1.26, p = 0.004), while no comparable association was observed in a chemotherapy-only negative-control cohort, supporting the hypothesis that the time-of-day effect is specific to immune checkpoint inhibitor biology rather than reflecting non-specific prognostic differences. ²⁰ The magnitude of the association observed in our ES-SCLC cohort was greater than that reported in these broader analyzes. This may reflect the particularly aggressive biology and poor prognosis of ES-SCLC, the relatively homogeneous platinum–etoposide treatment backbone, the more uniform treatment and infusion-scheduling practices within a single-center setting, and potentially a greater dependence on effective early immune priming during induction therapy. Taken together, the consistency of these findings across different tumor histologies, geographic populations, healthcare systems, and treatment schedules strengthens the biological plausibility of circadian modulation of ICI efficacy, rather than suggesting a center-specific scheduling phenomenon.

To date, the only previous study evaluating ToDA in ES-SCLC is the single-center Chinese LungTime-R02 cohort (n = 397), which identified 15:00 as the optimal ToDA threshold and reported HR of 0.37 (95% CI 0.26–0.53) for OS and 0.48 (95% CI 0.36–0.65) for PFS in favor of morning administration. ²¹ Our findings independently reproduced the same directional association, demonstrating that morning administration of first-line chemoimmunotherapy was associated with significantly longer survival in patients with ES-SCLC, although the cohort-median threshold used for the primary analysis was substantially earlier (11:54 versus 15:00). Several factors may account for this difference. First, the distribution of infusion times differed markedly between the two healthcare systems. In LungTime-R02, approximately 13% of patients received the majority of their infusions after 15:00, whereas in our cohort only 5.6% received any infusion at or beyond this time. The median ToDA in our cohort was 11:54, with most treatments administered between 09:00 and 13:00, making evaluation of a 15:00 threshold less informative in our setting. Second, the distribution of immune checkpoint inhibitors differed between cohorts, with atezolizumab accounting for 77.6% of treatments in our study, whereas LungTime-R02 included a substantially higher proportion of durvalumab-treated patients. Whether these agents differ in their chronopharmacological properties remains unknown but could theoretically influence the optimal ToDA window. Third, population-level differences in circadian behavior, lifestyle, and healthcare practices may also contribute to variation in the optimal timing of treatment. Notably, the bicontinental analysis by Huang et al. of 713 patients with advanced NSCLC treated in France and China identified an optimal ToDA threshold of 11:30, ¹⁹ remarkably close to the 11:54 threshold observed in our cohort and substantially earlier than the 15:00 cut-off reported in LungTime-R02. Accordingly, these differences argue against interpreting any single clock-time threshold as universally applicable across healthcare settings. The clinically relevant therapeutic window remains uncertain and requires prospective evaluation.

Cut-point sensitivity analyzes showed variability in effect estimates across alternative thresholds, indicating that the primary median-based 11:54 classification should not be interpreted as a discrete biological or clinical cut-off. These findings support prospective evaluation of ToDA as a potentially continuous or context-dependent exposure rather than the adoption of a universally applicable clock-time threshold.

Landmark analyzes were performed to assess potential guarantee-time bias arising from the exposure definition. The association remained significant for both PFS and OS at 3 and 6 months and for OS at 9 months, whereas the PFS estimate at 9 months remained directionally consistent but was no longer statistically significant in the substantially smaller cohort. These findings reduce the likelihood that the primary association was solely attributable to guarantee-time bias.

Although our study was not designed to investigate the underlying mechanisms, these observations are consistent with current understanding of circadian immunobiology. Adaptive immune responses exhibit diurnal oscillations, with circadian rhythms regulating multiple components of antitumour immunity relevant to immune checkpoint inhibitor efficacy, including antigen presentation and dendritic cell activation, lymphocyte trafficking, CD8⁺ T-cell infiltration, cytokine secretion, PD-1 expression, and immunosuppressive myeloid populations. ^3-8 Experimental evidence further indicates that the circadian clock modulates antitumour immunity through at least two complementary mechanisms: by shaping the composition of the tumor immune microenvironment, including the rhythmic accumulation of PD-L1-expressing immunosuppressive myeloid cells that constrain cytotoxic T-cell activity ⁷ and independently, through a cell-intrinsic circadian clock within T cells themselves that directly regulates T-cell receptor-dependent activation pathways and the magnitude of the T-cell response. ³¹ Administration of immune checkpoint inhibitors during periods of heightened immune responsiveness may therefore enhance immune priming during the induction phase, potentially contributing to improved long-term antitumour activity.

Notably, objective response rates were virtually identical between treatment groups despite significant improvements in progression-free and overall survival. This pattern is consistent with findings from a recent national neoadjuvant melanoma cohort, in which morning ICI administration was associated with improved long-term survival despite no significant difference in pathological response, suggesting that treatment timing and early tumor response capture distinct aspects of immunotherapy efficacy. These concordant observations across both metastatic and neoadjuvant settings raise the possibility that circadian timing primarily influences the durability and long-term quality of antitumour immune responses rather than the likelihood of achieving an initial tumor response. This mechanistic interpretation remains hypothesis-generating and warrants validation in prospective studies incorporating longitudinal immune profiling and biomarker analyzes. ³²

Our study has several notable strengths. To our knowledge, this is the first study evaluating ToDA of first-line chemoimmunotherapy in ES-SCLC outside East Asia, providing an independent validation of the chronotherapy hypothesis in a distinct healthcare system and population. We also evaluated the dependence of the observed association on the selected ToDA threshold through prespecified cut-point sensitivity analyzes and assessed potential guarantee-time bias using landmark analyzes. The single-center design ensured relatively homogeneous treatment delivery and infusion-scheduling practices, while all patients received guideline-concordant first-line therapy, supporting the real-world applicability of our findings to comparable clinical settings.

This study has several limitations that should be acknowledged. First, the single-center retrospective design introduces the possibility of selection bias, as patients with better performance status, fewer comorbidities, or shorter travel distances may have been preferentially scheduled for morning infusions. ²⁰ Second, although this represents one of the largest ES-SCLC chronotherapy cohorts reported outside East Asia, the sample size (n = 107) remains smaller than that of LungTime-R02 and limits the statistical power of subgroup analyzes. Third, the predominantly morning distribution of infusion times precluded direct evaluation of the 15:00 ToDA threshold proposed by Huang et al. ²¹ Fourth, follow-up duration also differed between groups because patients in the Afternoon cohort were enrolled earlier, potentially influencing median survival estimates despite appropriate time-to-event analyzes. Fifth, because ToDA classification was derived from the median timing of the first two to four ICI infusions, the exposure definition may introduce guarantee-time bias despite defining time zero as the first ICI infusion. Landmark analyzes at 3, 6, and 9 months yielded directionally consistent estimates, reducing—but not eliminating—this concern. Exact per-cycle administration dates were unavailable; therefore, the 3-month landmark represents a regimen-based approximation of fourth-cycle completion rather than a patient-specific cycle date. In addition, variability across alternative ToDA thresholds indicates that the median-based 11:54 classification should not be interpreted as a validated biological or clinical cut-off. The relatively low documented irAE rate (7.5%) should be interpreted in the context of retrospective toxicity ascertainment, as mild, transient, or clinically nonspecific immune-related events may have been under-recorded in routine clinical documentation. Comparisons according to immune checkpoint inhibitor type should be interpreted cautiously given the small durvalumab subgroup (n = 24), potential confounding by indication, and the exploratory nature of this analysis. Finally, as with all observational studies, residual confounding cannot be excluded, and prospective studies or target-trial emulation approaches will be important to further validate these findings. ²⁰

In conclusion, our findings suggest that morning administration of first-line chemoimmunotherapy is associated with significantly longer PFS and OS in patients with ES-SCLC without increasing treatment-related toxicity. These results extend the chronotherapy hypothesis to a non-East Asian ES-SCLC cohort and externally validate the directional association previously reported in East Asia, ²¹ while highlighting that the clinically relevant ToDA window remains uncertain and may vary according to local infusion-time distributions. Prospective multicentre studies are warranted to confirm these findings and define the optimal timing of immunotherapy administration in routine clinical practice.

Supplementary Material

Supplementary Material

supplementary_Figures_revised - Clean.docx

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Data availability statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to their containing information that could compromise the privacy of research participants.

Use of Artificial Intelligence

Artificial intelligence (ChatGPT, OpenAI) was used to assist with language editing and improving the clarity of the manuscript. All scientific content was critically reviewed and approved by the authors, who take full responsibility for the manuscript.

Supplementary material

Supplemental data for this article can be accessed at https://doi.org/10.1080/2162402X.2026.2738262.