Abstract
Background
For patients with programmed death-ligand 1 (PD-L1)-high non-small cell lung cancer (NSCLC), both immune checkpoint inhibitor (ICI) monotherapy and ICI plus chemotherapy are first-line options, yet optimal subgroups for each remain undefined. We evaluated Ki-67 expression as a biomarker for treatment stratification.
Methods
We retrospectively analyzed 334 advanced PD-L1-high NSCLC cases (2018–2024). Patients were stratified by Ki-67 expression, and outcomes with ICI monotherapy versus ICI-chemotherapy were compared using propensity score matching and multivariable Cox models.
Results
After propensity score matching, in the Ki-67 > 30% cohort (n = 166), ICI-chemotherapy was associated with a higher ORR versus ICI monotherapy (32/83, 38.6% vs. 17/83, 20.5%; P = 0.01) and with longer PFS (9.9 vs. 8.4 months; HR 0.51, 95% CI 0.37–0.72; P < 0.001) and OS (22.1 vs. 16.5 months; HR 0.47, 95% CI 0.32–0.70; P < 0.001). In the Ki-67 ≤ 30% cohort (n = 78), ORR was identical (9/39, 23.1%), and no improvement was observed for PFS (8.3 vs. 9.1 months; HR 1.59, 95% CI 0.98–2.59; P = 0.06) or OS (15.7 vs. 20.3 months; HR 1.47, 95% CI 0.85–2.54; P = 0.16). Patients receiving ICI-chemotherapy had more grade ≥ 3 adverse events (40/157, 25.5% vs. 24/177, 13.6%), predominantly hematologic (30/157, 19.1% vs. 3/177, 1.7%).
Conclusion
Ki-67 expression may represent a potential predictive biomarker for identifying PD-L1-high NSCLC patients more likely to benefit from chemo-immunotherapy, though its clinical utility requires confirmation in prospective, randomized studies.
Supplementary Information
The online version contains supplementary material available at 10.1007/s00262-025-04207-9.
Keywords: Ki-67, PD-L1, Non-small cell lung cancer, Immune checkpoint inhibitor
Introduction
Lung cancer remains the leading cause of cancer mortality globally, with non-small cell lung cancer (NSCLC) constituting about 80% of new diagnoses [1]. In recent years, immune checkpoint inhibitors (ICIs) have become the cornerstone of first-line therapy for advanced NSCLC without driver mutations. Both ICI monotherapy (supported by the IMpower110, KEYNOTE-024/042 trials [2–4]) and ICI plus chemotherapy (validated in the KEYNOTE-189/407 studies [5, 6]) are now standard options for patients with programmed death-ligand 1 (PD-L1) ≥ 50% according to current guidelines [7].
Current evidence [8–11] indicates comparable long-term survival between ICI monotherapy and ICI plus chemotherapy in patients with PD-L1 ≥ 50% NSCLC. However, the combination regimen may improve objective response rate (ORR), reduce the risk of early progression (within 3 months) and short-term mortality (within 12 months), making it more suitable for patients with high tumor burden or rapid disease progression. Additional research [12] suggests that cancer cachexia may impair the efficacy of ICI monotherapy, while combination chemotherapy could mitigate the negative impact of cachexia on anti-tumor immunity by rapidly reducing tumor burden and promoting the reversal of cachexia. However, concepts such as tumor burden, progression rate, and cachexia remain ambiguous, as they are external manifestations of tumor proliferative activity influenced by multiple factors, limiting their clinical applicability [13–15]. Therefore, there is an urgent need to identify specific biomarkers that can predict the benefit of combination chemotherapy in patients with high PD-L1 expression.
Ki-67, a key biomarker reflecting tumor cell proliferative activity, holds significant value in prognostic assessment and treatment response prediction for various malignancies, including breast and lung cancers [16, 17]. A high Ki-67 index indicates active tumor cell proliferation and is associated with increased sensitivity to chemotherapy [18]. Clara Helal et al. [19] confirmed in their study on early-stage triple-negative breast cancer that Ki-67 ≥ 30% is a strong independent predictor of pathological complete response (pCR) following neoadjuvant chemo-immunotherapy. This suggests that Ki-67 expression may serve as a potential predictive biomarker to stratify NSCLC patients with PD-L1 ≥ 50% for chemo-immunotherapy. Therefore, we conducted this real-world cohort study to evaluate its role in optimizing treatment selection. We present this article in accordance with the REMARK reporting checklist.
Methods
Study design and patients
We systematically searched the electronic databases at General Hospital of Southern Theater Command to identify eligible patients diagnosed with NSCLC from January 2018 to April 2024. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments, and was approved by the Ethics Committee of our hospital (approval number: NZLLKZ2024159). The inclusion criteria were: (1) histologically or cytologically confirmed lung adenocarcinoma or squamous cell carcinoma; (2) stage IIIB–IV disease (AJCC 8th edition) [20]; (3) PD-L1 tumor proportion score (TPS) ≥ 50% in primary or metastatic lesions; (4) received first-line programmed death-1 (PD-1) inhibitor monotherapy or combination platinum-based chemotherapy with complete clinical data. Exclusion criteria included: (1) non-squamous/non-adenocarcinoma histological subtypes (e.g., neuroendocrine carcinoma, sarcomatoid carcinoma); (2) presence of driver mutations, including EGFR, ALK, ROS1, BRAF V600E, MET Exon 14 skipping, RET, NTRK and HER2 mutations; (3) fewer than 2 cycles of immunotherapy or lack of response assessment; (4) missing Ki-67 data.
Biomarker assessments and molecular profiling
PD-L1 expression was assessed using by the DAKO 22C3 immunohistochemical staining assay, with TPS ≥ 50% defined as high PD-L1 expression. Ki-67 index was determined using MIB-1 monoclonal antibody, with tonsil tissue as the positive control. Two pathologists selected three high-proliferation areas at 200 × magnification, counting ≥ 500 tumor cells to determine the nuclear positivity rate (i.e., Ki-67 index). The Ki-67 cutoff of 30% (high: > 30%; low: ≤ 30%) was selected based on prior lung cancer studies indicating its prognostic significance [21, 22] and was further supported by internal validation in this cohort.
In addition, a subset of enrolled patients underwent routine targeted mutation testing by multiplex real-time polymerase chain reaction (PCR) or next-generation sequencing (NGS) at the discretion of the treating physician. For the exploratory analyses, concomitant mutations were catalogued and classified into four pre-defined immune-related gene baskets—immune-favourable only, immune-unfavourable only, mixed immune favorable and unfavorable, or neither—according to previously published literature [23–27].
Treatment and outcomes
We compared the efficacy of first-line ICI monotherapy (ICI-mono) versus ICI plus chemotherapy (ICI-chemo) in PD-L1 ≥ 50% advanced NSCLC stratified by Ki-67 expression. The ICIs used in our study were all PD-1 inhibitors (e.g., pembrolizumab, camrelizumab, sintilimab). Chemotherapy regimens were selected by physicians based on factors such as histology (e.g., pemetrexed/paclitaxel plus carboplatin/cisplatin).
Tumor response was assessed per the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 by a blinded, independent central review conducted by radiologists and medical oncologists; dual reads were performed, with senior adjudication of any discrepancies to assign the final best overall response. Adverse events (AEs) were captured from medical records and graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (NCI-CTCAE) version 5.0. ORR was defined as the percentage of patients achieving complete response (CR) or partial response (PR). Progression-free survival (PFS) was defined as the time from the start of treatment to tumor progression or death, and overall survival (OS) was defined as the time from the start of first-line treatment to all-cause mortality.
Statistical analysis
Baseline characteristics were compared using Mann–Whitney U (continuous) and Fisher’s exact tests (categorical). Survival distributions were estimated by Kaplan–Meier and compared with log-rank tests. ORR 95% confidence intervals (CIs) were computed by the exact Clopper–Pearson method. We performed 1:1 propensity-score matching (PSM, nearest-neighbour, no replacement, caliper 0.05) within Ki-67 > 30% and ≤ 30% strata using sex, age, smoking history, ECOG-PS, histology, clinical stage, brain and liver metastases; balance was assessed by standardized mean differences (SMDs < 0.1). Cox models provided adjusted hazard ratios (aHRs, 95% CIs); covariates were prespecified by clinical relevance and pre‑match imbalance (SMDs > 0.1), with variance Inflation factor (VIF) screening indicating no concerning collinearity (all < 5). All tests were two-sided, with P < 0.05 considered statistically significant. Analyses were performed using GraphPad Prism 8.0 (GraphPad Software, La Jolla, CA, USA) and SPSS 25.0 (IBM Corp., Armonk, NY, USA).
Results
Patient characteristics
Among 535 consecutive PD-L1-high advanced NSCLC patients treated with first-line PD-1 inhibitor monotherapy or combination chemotherapy between January 2018 and April 2024, 68 were excluded due to non-adenocarcinoma/squamous histology, 20 for harboring driver gene mutations, 47 for receiving fewer than two cycles of immunotherapy or lacking efficacy evaluation, and 66 for missing Ki-67 expression data. The study flow is shown in Fig. 1. In both the overall cohort (n = 334) and the subgroup based on Ki-67 expression (> 30%, n = 227; ≤ 30%, n = 107), baseline characteristics of patients treated with ICI-mono or ICI-chemo showed generally acceptable balance (all P > 0.05, SMD < 0.2); with SMD values for age, ECOG-PS, clinical stage, and liver metastasis exceeding 0.1, as shown in Tables S1 and S2.
Fig. 1.
Flowchart of patient enrollment. PD-L1, programmed death-ligand 1; NSCLC, non-small cell lung cancer; ICI, immune checkpoint inhibitor; ICI-mono, ICI monotherapy; ICI-chemo, ICI plus chemotherapy
Clinical outcomes in the overall cohort
As of April 1, 2025, with a median follow-up of 27.5 months (95% CI: 22.2–32.8), 289 out of 334 patients (86.5%) experienced disease progression, and 217 patients (65.0%) died. The ORR in the combination therapy group was significantly higher than in the monotherapy group (54/157, 34.4% vs. 40/177, 22.6%; P = 0.02; Table S1). Despite this, no survival advantage was observed with combination therapy: PFS 9.4 versus 8.5 months (HR 0.83, 95% CI 0.66–1.05; P = 0.11; Fig. 2A) and OS 19.3 versus 18.4 months (HR 0.82, 95% CI 0.62–1.07; P = 0.14; Fig. 2B).
Fig. 2.
Kaplan–Meier curves for A PFS and B OS in the overall PD-L1 ≥ 50% NSCLC cohort receiving first-line ICI-chemo versus ICI-mono. PFS, progression-free survival; OS, overall survival; ICI, immune checkpoint inhibitor; ICI-mono, ICI monotherapy; ICI-chemo, ICI plus chemotherapy; CI, confidence interval; HR, hazard ratio
Clinical outcomes in the overall cohort stratified by Ki-67 expression
We then compared ICI-mono versus ICI-chemo in the overall cohort stratified by Ki-67 expression. Before fixing the cut-off, we examined alternative thresholds (20%, 25%, 30%, 50%) in sensitivity analyses: at 20% the OS difference within the high Ki-67 stratum was non-significant (P = 0.19), approached significance at 25% (P = 0.09), became significant at 30% (P = 0.001), and was more pronounced at 50% (P < 0.001), at the cost of excluding patients with Ki-67 between 30 and 50% who may still benefit from ICI-chemo. For context, the Supplement includes the cohort’s binned Ki‑67 distribution (Fig. S1) and survival curves for each cutoff (Fig. S2).
Specifically, among patients with Ki-67 > 30%, ICI-chemo yielded a higher ORR than ICI-mono (40/106, 37.7% vs. 27/121, 22.3%; P = 0.01; Table S2) and was associated with longer PFS (9.8 vs. 8.4 months; HR 0.62, 95% CI 0.46–0.82; P = 0.001; Fig. 3A) and OS (22.7 vs. 16.6 months; HR 0.57, 95% CI 0.40–0.80; P = 0.001; Fig. 3B). Conversely, in the Ki-67 ≤ 30% cohort, ORR was similar between treatments (14/51, 27.5% vs. 13/56, 23.2%; P = 0.61; Table S2); ICI-chemo was associated with shorter PFS (8.3 vs. 9.1 months; HR 1.56, 95% CI 1.02–2.36; P = 0.04; Fig. 3C) and a non‑significant trend toward shorter OS (16.3 vs. 20.6 months; HR 1.53, 95% CI 0.97–2.43; P = 0.07; Fig. 3D), with crossing survival curves. Building on these overall-cohort findings with acceptable baseline balance (SMD < 0.2), we then applied propensity score matching and multivariable Cox adjustment at the 30% cutoff to evaluate robustness.
Fig. 3.
Kaplan–Meier curves before propensity-score matching for A PFS and B OS in patients with PD-L1 ≥ 50% and Ki-67 > 30% NSCLC receiving first-line ICI-chemo versus ICI-mono, and C PFS and D OS in the corresponding PD-L1 ≥ 50% and Ki-67 ≤ 30% cohort. PFS, progression-free survival; OS, overall survival; ICI, immune checkpoint inhibitor; ICI-mono, ICI monotherapy; ICI-chemo, ICI plus chemotherapy; CI, confidence interval; HR, hazard ratio
Clinical outcomes in matched patients stratified by Ki-67 expression
After 1:1 propensity score matching was performed separately within the Ki-67 > 30% and ≤ 30% strata, all baseline covariates achieved SMD < 0.1 as shown in Table 1. In the matched Ki-67 > 30% cohort (n = 166), ICI-chemo consistently yielded a higher ORR (32/83, 38.6% vs. 17/83, 20.5%; P = 0.01; Table 1) and longer median PFS (9.9 vs. 8.4 months; HR 0.51, 95% CI 0.37–0.72; P < 0.001; Fig. 4A) and OS (22.1 vs. 16.5 months; HR 0.47, 95% CI 0.32–0.70; P < 0.001; Fig. 4B) than ICI alone. In the matched Ki-67 ≤ 30% cohort (n = 78), ORR was identical (9/39, 23.1%, Table 1); survival curves crossed, with a non‑significant trend toward shorter PFS (8.3 vs. 9.1 months; HR 1.59, 95% CI 0.98–2.59; P = 0.06; Fig. 4C) and OS (15.7 vs. 20.3 months; HR 1.47, 95% CI 0.85–2.54; P = 0.16; Fig. 4D) for ICI-chemo. Thus, after balancing covariates, high Ki-67 remained associated with benefit from ICI-chemo, whereas low Ki-67 showed no advantage and a trend toward harm.
Table 1.
Baseline characteristics stratified by Ki-67 expression after propensity score matching
| Characteristic | Ki-67 > 30% (n = 166) | Ki-67 ≤ 30% (n = 78) | ||||
|---|---|---|---|---|---|---|
| ICI-mono (n = 83) |
ICI-chemo (n = 83) |
P value (SMD) |
ICI-mono (n = 39) |
ICI-chemo (n = 39) |
P value (SMD) |
|
| Age at diagnosis | ||||||
| Median, years (range) | 62 (35–81) | 62 (32–80) | 0.69 (0.06) | 62 (32–81) | 63 (41–84) | 0.93 (0.02) |
| ≥ 65 years, n (%) | 35 (42.2) | 37 (44.6) | 0.75 (0.05) | 16 (41.0) | 15 (38.5) | 0.83 (0.05) |
| Gender, n (%) | 0.73 (0.05) | 0.79 (0.07) | ||||
| Male | 61 (73.5) | 59 (71.1) | 31 (79.5) | 30 (76.9) | ||
| Female | 22 (26.5) | 24 (28.9) | 8 (20.5) | 9 (23.1) | ||
| ECOG-PS, n (%) | 0.65 (0.07) | 0.73 (0.08) | ||||
| 0–1 | 71 (85.5) | 73 (88.0) | 34 (87.2) | 35 (89.7) | ||
| ≥ 2 | 12 (14.5) | 10 (12.0) | 5 (12.8) | 4 (10.3) | ||
| Smoking history, n (%) | 0.87 (0.02) | 1.00 (0.00) | ||||
| Never | 27 (32.5) | 26 (31.3) | 13 (33.3) | 13 (33.3) | ||
| Current or Former | 56 (67.5) | 57 (68.7) | 26 (66.7) | 26 (66.7) | ||
| Histology, n (%) | 0.64 (0.07) | 1.00 (0.00) | ||||
| Adenocarcinoma | 46 (55.4) | 43 (51.8) | 21 (53.8) | 21 (53.8) | ||
| Squamous cell carcinoma | 37 (44.6) | 40 (48.2) | 18 (46.2) | 18 (46.2) | ||
| Stage, n (%) | 0.85 (0.03) | 1.00 (0.00) | ||||
| III | 18 (21.7) | 19 (22.9) | 9 (23.1) | 9 (23.1) | ||
| IV | 65 (78.3) | 64 (77.1) | 30 (76.9) | 30 (76.9) | ||
| Liver metastasis, n (%) | 9 (10.8) | 10 (12.0) | 0.81 (0.04) | 2 (5.1) | 2 (5.1) | 1.00 (0.00) |
| Brain metastasis, n (%) | 12 (14.5) | 10 (12.0) | 0.65 (0.07) | 6 (15.4) | 5 (12.8) | 0.74 (0.07) |
| Best overall response, n (%) | ||||||
| Complete response | 0 (0.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | ||
| Partial response | 17 (20.5) | 32 (38.6) | 9 (23.1) | 9 (23.1) | ||
| Stable disease | 55 (66.3) | 42 (50.6) | 26 (66.7) | 28 (71.9) | ||
| Progressive disease | 11 (13.3) | 9 (10.8) | 4 (10.3) | 2 (5.1) | ||
| ORR | 0.01 | 1.00 | ||||
| n (%) | 17/83 (20.5) | 32/83 (38.6) | 9/39 (23.1) | 9/39 (23.1) | ||
| 95% CI | 12.4–30.8 | 28.1–49.9 | 11.1–39.3 | 11.1–39.3 | ||
ICI, immune checkpoint inhibitor; ICI-mono, ICI monotherapy; ICI-chemo, ICI plus chemotherapy; SMD, standardized mean difference; ECOG-PS, Eastern Cooperative Oncology Group Performance Status; Ki-67, percentage of tumor cells with nuclear staining; ORR: objective response rate; CI, confidence interval
Fig. 4.
Kaplan–Meier curves after propensity-score matching for A PFS and B OS in patients with PD-L1 ≥ 50% and Ki-67 > 30% NSCLC receiving first-line ICI-chemo versus ICI-mono, and C PFS and D OS in the corresponding PD-L1 ≥ 50% and Ki-67 ≤ 30% cohort. PFS, progression-free survival; OS, overall survival; ICI, immune checkpoint inhibitor; ICI-mono, ICI monotherapy; ICI-chemo, ICI plus chemotherapy; CI, confidence interval; HR, hazard ratio
Prognostic factors in matched patients stratified by Ki-67 expression
Multivariable Cox models were fitted in the post-PSM cohorts to estimate aHRs for treatment and key covariates, assessing robustness within the Ki-67 > 30% and ≤ 30% strata; covariates were selected after collinearity screening (Table S3, all VIF < 2), informed by clinical relevance and pre-matching SMDs. In the matched Ki-67 > 30% cohort (Table 2A), chemo-immunotherapy remained independently associated with longer PFS and OS versus ICI monotherapy (PFS: aHR 0.52, 95% CI 0.37–0.74, P < 0.001; OS: aHR 0.51, 95% CI 0.34–0.76, P = 0.001). ECOG-PS 0–1 was independently favorable (PFS: aHR 0.25, 95% CI 0.15–0.42, P < 0.001; OS: aHR 0.23, 95% CI 0.13–0.40, P < 0.001), whereas age < 65 years, clinical stage III (vs. IV), and histology were not significant after adjustment. In the matched Ki-67 ≤ 30% cohort (Table 2B), the treatment regimen was not associated with improved outcomes (PFS: aHR 1.50, 95% CI 0.90–2.50, P = 0.12; OS: aHR 1.37, 95% CI 0.78–2.40, P = 0.27). ECOG-PS 0–1 remained independently favorable (PFS: aHR 0.38, 95% CI 0.17–0.83, P = 0.02; OS: aHR 0.41, 95% CI 0.18–0.94, P = 0.04), while age, stage, and histology were not significant.
Table 2.
(A) Predictors of PFS and OS in matched patients with Ki-67 > 30%. (B) Predictors of PFS and OS in matched patients with Ki-67 ≤ 30%
| Covariates | PFS | OS | ||||||
|---|---|---|---|---|---|---|---|---|
| cHR (95% CI) | P | aHR (95% CI) | P | cHR (95% CI) | P | aHR (95% CI) | P | |
|
Treatment regimen (ICI-chemo vs. ICI-mono) |
0.51 (0.37–0.72) | < 0.001 | 0.52 (0.37–0.74) | < 0.001 | 0.47 (0.32–0.70) | < 0.001 | 0.51 (0.34–0.76) | 0.001 |
|
Age stratification (< 65 vs. ≥ 65) |
0.80 (0.58–1.11) | 0.18 | 0.94 (0.66–1.34) | 0.73 | 0.80 (0.55–1.17) | 0.25 | 1.05 (0.70–1.59) | 0.81 |
|
ECOG-PS (PS 0–1 vs. PS ≥ 2) |
0.24 (0.15–0.39) | < 0.001 | 0.25 (0.15–0.42) | < 0.001 | 0.22 (0.13–0.37) | < 0.001 | 0.23 (0.13–0.40) | < 0.001 |
|
Clinical stage (III vs. IV) |
0.78 (0.52–1.16) | 0.22 | 0.93 (0.61–1.42) | 0.73 | 0.80 (0.50–1.29) | 0.36 | 1.02 (0.62–1.68) | 0.94 |
|
Histology (ADC vs SCC) |
0.90 (0.65–1.26) | 0.55 | 0.79 (0.56–1.12) | 0.18 | 0.90 (0.61–1.31) | 0.57 | 0.76 (0.51–1.13) | 0.17 |
| Covariates | PFS | OS | ||||||
|---|---|---|---|---|---|---|---|---|
| cHR (95% CI) | P | aHR (95% CI) | P | cHR (95% CI) | P | aHR (95% CI) | P | |
|
Treatment regimen (ICI-chemo vs ICI-mono) |
1.59 (0.98–2.59) | 0.06 | 1.50 (0.90–2.50) | 0.12 | 1.47 (0.85–2.54) | 0.16 | 1.37 (0.78–2.40) | 0.27 |
|
Age stratification (< 65 vs. ≥ 65) |
1.21 (0.74–1.98) | 0.45 | 1.26 (0.73–2.19) | 0.41 | 1.22 (0.70–2.13) | 0.48 | 1.31 (0.70–2.46) | 0.41 |
|
ECOG-PS (PS 0–1 vs. PS ≥ 2) |
0.42 (0.21–0.86) | 0.02 | 0.38 (0.17–0.83) | 0.02 | 0.48 (0.22–1.01) | 0.05 | 0.41 (0.18–0.94) | 0.04 |
|
Clinical stage (III vs. IV) |
0.88 (0.49–1.56) | 0.65 | 0.95 (0.49–1.84) | 0.87 | 0.90 (0.47–1.71) | 0.74 | 0.98 (0.47–2.04) | 0.95 |
|
Histology (ADC vs. SCC) |
1.08 (0.67–1.74) | 0.75 | 1.20 (0.70–2.08) | 0.50 | 1.10 (0.64–1.89) | 0.72 | 1.21 (0.66–2.20) | 0.54 |
ICI, immune checkpoint inhibitor; ICI-mono, ICI monotherapy; ICI-chemo, ICI plus chemotherapy; ECOG PS, Eastern Cooperative Oncology Group Performance Status; ADC, adenocarcinoma; SCC, squamous cell carcinoma; cHR, crude hazard ratio; aHR, adjusted hazard ratio; PFS, progression-free survival; OS, overall survival. aHR for treatment regimen was adjusted for age stratification, ECOG-PS, clinical stage and histology
Safety
Adverse event analysis was conducted in the overall enrolled patient cohort, comparing ICI-chemotherapy and ICI monotherapy (Table 3). Any grade adverse events were more frequent with ICI-chemo than with ICI-mono (139/157, 88.5% vs. 107/177, 60.4%), and grade ≥ 3 events likewise (40/157, 25.5% vs. 24/177, 13.6%). Grade ≥ 3 toxicities with the combination regimen were mainly hematologic (30/157, 19.1% vs. 3/177, 1.7%). For non-hematologic events, chemotherapy related symptoms were more frequent with ICI-chemo, including fatigue (57.3% vs. 14.1%), nausea (56.7% vs. 8.5%), alopecia (54.8% vs. 1.1%), and diarrhea (35.0% vs. 15.8%), whereas immune type events were relatively more prominent with ICI monotherapy (rash 18.1%, pruritus 15.8%, hypothyroidism 18.1%). Treatment-related mortality was low and comparable between the two groups (1.1% vs. 1.3%), as were discontinuation rates due to adverse events (6.8% vs. 8.3%).
Table 3.
Treatment-related adverse events with incidence ≥ 10% in the overall cohort
| Event, n (%) | ICI-mono (n = 177) | ICI-chemo (n = 157) | ||
|---|---|---|---|---|
| Any grade | Grade ≥ 3 | Any grade | Grade ≥ 3 | |
| Any | 107 (60.4) | 24 (13.6) | 139 (88.5) | 40 (25.5) |
| Hematologic | 20 (11.3) | 3 (1.7) | 114 (72.6) | 30 (19.1) |
| Anaemia | 12 (6.8) | 1 (0.6) | 85 (54.1) | 15 (9.6) |
| Neutropenia | 8 (4.5) | 2 (1.1) | 62 (39.5) | 21 (13.4) |
| Leukopenia | 7 (4.0) | 1 (0.6) | 42 (26.8) | 10 (6.4) |
| Non-hematologic | 95 (53.7) | 22 (12.4) | 126 (80.2) | 26 (16.6) |
| Fatigue | 25 (14.1) | 4 (2.3) | 90 (57.3) | 15 (9.6) |
| Nausea | 15 (8.5) | 0 (0.0) | 89 (56.7) | 12 (7.6) |
| Alopecia | 2 (1.1) | 0 (0.0) | 86 (54.8) | 0 (0.0) |
| Diarrhea | 28 (15.8) | 5 (2.8) | 55 (35.0) | 5 (3.2) |
| Decreased appetite | 18 (10.2) | 1 (0.6) | 55 (35.0) | 7 (4.5) |
| Constipation | 12 (6.8) | 0 (0.0) | 55 (35.0) | 1 (0.6) |
| Peripheral neuropathy | 5 (2.8) | 1 (0.6) | 50 (31.8) | 3 (1.9) |
| Stomatitis | 10 (5.6) | 0 (0.0) | 45 (28.7) | 2 (1.3) |
| Hypoalbuminemia | 6 (3.4) | 0 (0.0) | 40 (25.5) | 2 (1.3) |
| Vomiting | 2 (1.1) | 0 (0.0) | 35 (22.3) | 6 (3.8) |
| Arthralgia | 15 (8.5) | 4 (2.3) | 33 (21.0) | 2 (1.3) |
| Pruritus | 28 (15.8) | 7 (4.0) | 33 (21.0) | 2 (1.3) |
| Rash | 32 (18.1) | 9 (5.1) | 32 (20.4) | 4 (2.5) |
| ALT/AST increased | 20 (11.3) | 8 (4.5) | 30 (19.1) | 5 (3.2) |
| Hypothyroidism | 32 (18.1) | 12 (6.8) | 28 (17.8) | 1 (0.6) |
| Pneumonia | 18 (10.2) | 10 (5.6) | 20 (12.7) | 8 (5.1) |
| Led to discontinuation | 12 (6.8) | – | 13 (8.3) | – |
| Led to death | 2 (1.1) | – | 2 (1.3) | – |
ICI, immune checkpoint inhibitor; ICI-mono, ICI monotherapy; ICI-chemo, ICI plus chemotherapy
Exploratory analysis of concomitant genomics and Ki-67
To minimize platform- and specimen-related bias, analyses were restricted to adenocarcinoma cases profiled by tissue-based NGS (n = 108; patient selection flowchart in Fig. 5). Using pre-specified, literature-informed baskets (immune-favorable only, immune-unfavorable only, mixed, and neither), the distribution across Ki-67 strata (> 30% vs. ≤ 30%) showed no significant differences (P = 0.74–1.00; Table S4). When collapsing categories into “immune-unfavorable present” versus “immune-favorable only”, Ki-67 treated as a continuous variable exhibited substantial overlap between groups (P = 0.32; Fig. 6A), and basket composition remained stable across Ki-67 bins (P = 0.97; Fig. 6B). At the single-gene or co-mutation level (e.g., KRAS, TP53, STK11 and their common constellations), Ki-67 distributions were likewise not different (P = 0.88; Fig. 6C). Overall, in this cohort of lung adenocarcinoma with PD L1 ≥ 50%, Ki-67 did not show a clear association with the profiled co-mutation patterns at the current sample size.
Fig. 5.
Patient selection for tissue-based NGS in advanced lung adenocarcinoma with PD-L1 ≥ 50%. PD-L1, programmed death-ligand 1; ctDNA, circulating tumor DNA; NGS, next-generation sequencing
Fig. 6.
Exploratory association between Ki-67 and mutation categories in tissue-NGS adenocarcinomas with PD-L1 ≥ 50% (n = 108). A Ki-67 distribution: immune-favorable only versus immune-unfavorable present. B Proportional composition of the two collapsed baskets across Ki-67 bins. C Ki-67 distribution across detailed gene/co-mutation groups. PD-L1, programmed death-ligand 1; NGS, next-generation sequencing
Discussion
This real-world analysis examined PD-L1 ≥ 50% NSCLC and evaluated whether Ki-67 expression stratifies the association between treatment regimen and outcomes after balancing covariates. In patients with Ki-67 > 30%, chemo-immunotherapy was associated with higher response and longer survival than ICI monotherapy, whereas no advantage—and a trend toward harm—was observed in the Ki-67 ≤ 30% cohort.
Consistent with prior studies [8–11], combination therapy improved short-term tumor response in the unstratified PD-L1-high cohort but failed to confer a significant survival advantage (Fig. 2). Building on this, We exploratively used Ki-67 to stratify the prognostic relevance of these two first-line strategies. This aligns with the established stratification value of Ki-67 across various tumors. In breast cancer, Ki-67 helps assess the benefit of perioperative chemotherapy and guides adjuvant escalation therapy (such as abemaciclib) in patients with highly proliferative disease [28, 29]. In neuroendocrine tumors, Ki-67 (WHO grading) informs the choice between cytotoxic, targeted, or PRRT therapies [30]. Similarly, in gliomas and prostate cancer, higher Ki-67 levels are also associated with enhanced responsiveness to cytotoxic or intensive regimens [31–33].
In recent years, multiple studies have suggested that individual genes or composite genomic features may be associated with immunotherapy outcomes in NSCLC, including immune-favorable and immune-unfavorable alterations as well as variants linked to hyperprogression [23–27]. Accordingly, we assessed the relationship between tumor genomic context and Ki-67 expression to inform the stratification relevance of Ki-67 at the genomic level. In adenocarcinomas with tissue-based NGS, we did not observe a clear association between the distribution of Ki-67 expression and major immune-related genes or mutation baskets. Moreover, given the limited genomic data, the potential stratification value of Ki-67 did not appear to be materially modified by basket (Fig. S3), suggesting that Ki-67, as a marker of proliferative activity, may have relatively independent stratification potential. Owing to limited panels, small sample size, and heterogeneous testing platforms, these results are exploratory and require confirmation in larger, standardized prospective cohorts.
Mechanistically, a cautious interpretation is that proliferative state and immune effects may act in concert: in highly proliferative tumors, platinum/taxane chemotherapy is often accompanied by stronger cytotoxic effects and features of immunogenic cell death (ICD), potentially conditioning the microenvironment for PD-1/PD-L1 blockade [34–36]; in low-proliferation tumors, chemotherapy-induced ICD may be limited and transient lymphodepletion may relate to attenuated apparent benefit from checkpoint inhibition, and—given the significantly higher incidence of treatment-related adverse events (Table 3)—the overall risk–benefit profile of combination therapy appears unfavorable in this subgroup.
To enhance generalizability of our findings, we focused on the two major NSCLC histologies—adenocarcinoma and squamous cell carcinoma—excluding subtypes like large-cell neuroendocrine carcinoma, which differ markedly in biology, prognosis, and Ki-67 profiles [37, 38]. Subgroup analyses suggested better outcomes with ICI plus chemotherapy versus ICI monotherapy in both histologies at high Ki-67 (Fig. S4).
This study has several limitations. As a retrospective, single-center analysis, it is susceptible to selection bias in treatment allocation and to residual or unmeasured confounding despite propensity-score matching and multivariable adjustment, and key factors such as comorbidity burden, baseline tumor burden, and potential ECOG-PS misclassification were not fully captured. Time-dependent treatment dynamics (treatment switching, subsequent therapies) were not modeled in primary analyses and could bias survival estimates. Ki-67 measurement is sensitive to pre-analytic handling, antibody clone, platform, and scoring; an “optimal” cutoff is likely a range rather than a fixed point. The 30% threshold was assessed using multiple internal cut-points within this cohort and should be externally validated before broader application. Accordingly, findings should be interpreted as associative and hypothesis-generating. Prospective, multicenter randomized trials with standardized Ki-67 assessment and preplanned time-dependent analyses are needed before clinical application.
Conclusion
Ki-67 expression may represent a potential predictive biomarker for identifying PD-L1-high NSCLC patients more likely to benefit from chemo-immunotherapy, though its clinical utility requires confirmation in prospective, randomized studies.
Supplementary Information
Below is the link to the electronic supplementary material.
Acknowledgements
We thank all the authors who have worked hard to participate in primary studies.
Author contributions
NY, BX and JZ designed this study concept. NY, XC and HW collected the clinical data. NY, PL and BL conducted the statistical analyses. NY and JZ drafted the manuscript. XC, HW and BX provided critical suggestions and revised the manuscript. All authors approved the final version of the manuscript.
Funding
This work was supported by the Basic and Applied Basic Research Program of Guangdong Province (No. 2021A1515220040) and the Key Clinical Research Project of the General Hospital of Southern Theater Command (No. 2024NZB004).
Data availability
The data used in this study are available upon reasonable request from the corresponding author.
Declarations
Conflicts of interest
The authors declare no competing interests.
Ethics approval
The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments, and was approved by the Ethics Committee of the General Hospital of Southern Theater Command (approval number: NZLLKZ2024159). The requirement for informed consent was waived due to the anonymized use of clinical data.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Contributor Information
Bo Xie, Email: xiebodoctor@163.com.
Juan Zhou, Email: juanzhn@163.com.
References
- 1.Siegel RL, Giaquinto AN, Jemal A (2024) Cancer statistics, 2024. CA Cancer J Clin 74:12–49. 10.3322/caac.21820 [DOI] [PubMed] [Google Scholar]
- 2.Jassem J, de Marinis F, Giaccone G et al (2021) Updated overall survival analysis from IMpower110: Atezolizumab versus platinum-based chemotherapy in treatment-naive programmed death-ligand 1-selected NSCLC. J Thoracic oncol off publ Int Assoc Study Lung Cancer. 10.1016/j.jtho.2021.06.019 [Google Scholar]
- 3.Reck M, Rodríguez-Abreu D, Robinson AG et al (2019) Updated analysis of KEYNOTE-024: pembrolizumab versus platinum-based chemotherapy for advanced non-small-cell lung cancer with PD-L1 tumor proportion score of 50% or greater. J Clin Oncol 37:537–546. 10.1200/JCO.18.00149 [DOI] [PubMed] [Google Scholar]
- 4.de Castro G, Kudaba I, Wu Y-L et al (2023) Five-year outcomes with pembrolizumab versus chemotherapy as first-line therapy in patients with non-small-cell lung cancer and programmed death ligand-1 tumor proportion score ≥ 1% in the KEYNOTE-042 study. J Clin Oncol 41:1986–1991. 10.1200/JCO.21.02885 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Garassino MC, Gadgeel S, Speranza G et al (2023) Pembrolizumab plus pemetrexed and platinum in nonsquamous non-small-cell lung cancer: 5-year outcomes from the phase 3 KEYNOTE-189 study. J Clin Oncol 41:1992–1998. 10.1200/JCO.22.01989 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Novello S, Kowalski DM, Luft A et al (2023) Pembrolizumab plus chemotherapy in squamous non-small-cell lung cancer: 5-year update of the phase III KEYNOTE-407 study. J Clin Oncol 41:1999–2006. 10.1200/JCO.22.01990 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Riely GJ, Wood DE, Ettinger DS et al (2024) Non-small cell lung cancer, version 4.2024, NCCN clinical practice guidelines in oncology. J Natl Compr Canc Netw 22:249–274. 10.6004/jnccn.2204.0023 [DOI] [PubMed] [Google Scholar]
- 8.Pathak R, De Lima Lopes G, Yu H et al (2021) Comparative efficacy of chemoimmunotherapy versus immunotherapy for advanced non-small cell lung cancer: a network meta-analysis of randomized trials. Cancer 127:709–719. 10.1002/cncr.33269 [DOI] [PubMed] [Google Scholar]
- 9.Pérol M, Felip E, Dafni U et al (2022) Effectiveness of PD-(L)1 inhibitors alone or in combination with platinum-doublet chemotherapy in first-line (1L) non-squamous non-small-cell lung cancer (Nsq-NSCLC) with PD-L1-high expression using real-world data. Ann Oncol 33:511–521. 10.1016/j.annonc.2022.02.008 [DOI] [PubMed] [Google Scholar]
- 10.Shah M, Mamtani R, Marmarelis ME, Hennessy S (2023) Chemoimmunotherapy vs. immunotherapy for first line treatment of advanced non-small cell lung cancer with a PD-L1 expression ≥50% or ≥90. Clin Lung Cancer 24:235–243. 10.1016/j.cllc.2023.02.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Hektoen HH, Tsuruda KM, Brustugun OT et al (2025) Real-world comparison of pembrolizumab alone and combined with chemotherapy in metastatic lung adenocarcinoma patients with PD-L1 expression ≥50. ESMO Open 10(5):105073. 10.1016/j.esmoop.2025.105073 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Morimoto K, Uchino J, Yokoi T et al (2021) Impact of cancer cachexia on the therapeutic outcome of combined chemoimmunotherapy in patients with non-small cell lung cancer: a retrospective study. Oncoimmunology 10:1950411. 10.1080/2162402X.2021.1950411 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Zhang L, Zheng H, Jiang S-T et al (2024) Worldwide research trends on tumor burden and immunotherapy: a bibliometric analysis. Int J Surg 110:1699–1710. 10.1097/JS9.0000000000001022 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Morita-Tanaka S, Yamada T, Takayama K (2023) The landscape of cancer cachexia in advanced non-small cell lung cancer: a narrative review. Transl Lung Cancer Res 12:168–180. 10.21037/tlcr-22-561 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Shukuya T, Takahashi K, Shintani Y et al (2023) Epidemiology, risk factors and impact of cachexia on patient outcome: results from the Japanese Lung Cancer Registry Study. J Cachexia Sarcopenia Muscle 14:1274–1285. 10.1002/jcsm.13216 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Yang Y, Shao X, Li Z et al (2024) Prognostic heterogeneity of Ki67 in non-small cell lung cancer: a comprehensive reappraisal on immunohistochemistry and transcriptional data. J Cell Mol Med 28:e18521. 10.1111/jcmm.18521 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Zhu X, Chen L, Huang B et al (2020) The prognostic and predictive potential of Ki-67 in triple-negative breast cancer. Sci Rep. 10.1038/s41598-019-57094-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Chen X, He C, Han D et al (2017) The predictive value of Ki-67 before neoadjuvant chemotherapy for breast cancer: a systematic review and meta-analysis. Future Oncol 13:843–857. 10.2217/fon-2016-0420 [DOI] [PubMed] [Google Scholar]
- 19.Helal C, Djerroudi L, Ramtohul T et al (2025) Clinico-pathological factors predicting pathological response in early triple-negative breast cancer. NPJ Breast Cancer 11:15. 10.1038/s41523-025-00729-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Fong KM, Rosenthal A, Giroux DJ et al (2024) The International Association for the study of lung cancer staging project for lung cancer: proposals for the revision of the M descriptors in the Forthcoming Ninth Edition of the TNM classification for lung cancer. J Thorac Oncol 19:786–802. 10.1016/j.jtho.2024.01.019 [DOI] [PubMed] [Google Scholar]
- 21.Yoo J, Jung JH, Lee MA et al (2007) Immunohistochemical analysis of non-small cell lung cancer: correlation with clinical parameters and prognosis. J Korean Med Sci 22:318–325. 10.3346/jkms.2007.22.2.318 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Hommura F, Dosaka-Akita H, Mishina T et al (2000) Prognostic significance of p27KIP1 protein and ki-67 growth fraction in non-small cell lung cancers. Clin Cancer Res 6:4073–4081 [PubMed] [Google Scholar]
- 23.De Giglio A, De Biase D, Favorito V et al (2025) STK11 mutations correlate with poor prognosis for advanced NSCLC treated with first-line immunotherapy or chemo-immunotherapy according to KRAS, TP53, KEAP1, and SMARCA4 status. Lung Cancer 199:108058. 10.1016/j.lungcan.2024.108058 [DOI] [PubMed] [Google Scholar]
- 24.Bai X, Wu D-H, Ma S-C et al (2020) Development and validation of a genomic mutation signature to predict response to PD-1 inhibitors in non-squamous NSCLC: a multicohort study. J Immunother Cancer 8:e000381. 10.1136/jitc-2019-000381 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Gao A, Wang X, Wang J et al (2024) Homologous recombination deficiency status predicts response to immunotherapy-based treatment in non-small cell lung cancer patients. Thorac Cancer 15:1842–1853. 10.1111/1759-7714.15408 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Exposito F, Redrado M, Houry M et al (2023) PTEN loss confers resistance to anti-PD-1 therapy in non-small cell lung cancer by increasing tumor infiltration of regulatory T cells. Cancer Res 83:2513–2526. 10.1158/0008-5472.CAN-22-3023 [DOI] [PubMed] [Google Scholar]
- 27.Sun D, Qian H, Li J, Xing P (2024) Targeting MDM2 in malignancies is a promising strategy for overcoming resistance to anticancer immunotherapy. J Biomed Sci 31:17. 10.1186/s12929-024-01004-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Johnston SRD, Toi M, O’Shaughnessy J et al (2023) Abemaciclib plus endocrine therapy for hormone receptor-positive, HER2-negative, node-positive, high-risk early breast cancer (monarchE): results from a preplanned interim analysis of a randomised, open-label, phase 3 trial. Lancet Oncol 24:77–90. 10.1016/S1470-2045(22)00694-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.van den Ende NS, Nguyen AH, Jager A et al (2023) Triple-negative breast cancer and predictive markers of response to neoadjuvant chemotherapy: a systematic review. Int J Mol Sci. 10.3390/ijms24032969 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Rindi G, Mete O, Uccella S et al (2022) Overview of the 2022 WHO classification of neuroendocrine neoplasms. Endocr Pathol 33:115–154. 10.1007/s12022-022-09708-2 [DOI] [PubMed] [Google Scholar]
- 31.Nabian N, Ghalehtaki R, Zeinalizadeh M et al (2024) State of the neoadjuvant therapy for glioblastoma multiforme-where do we stand? Neuro-Oncol Adv 6:vdae028. 10.1093/noajnl/vdae028 [Google Scholar]
- 32.Dumke R, Dumke C, Eberle F et al (2022) Monocentric evaluation of Ki-67 labeling index in combination with a modified RPA score as a prognostic factor for survival in IDH-wildtype glioblastoma patients treated with radiochemotherapy. Strahlenther Onkol 198:892. 10.1007/s00066-022-01959-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Fan J, Liang H, Gu Y et al (2023) Predictive factors associated with differential pathologic response to neoadjuvant chemohormonal therapy in high-risk localized prostate cancer. Urol Oncol 41:354.e1-354.e9. 10.1016/j.urolonc.2023.05.006 [DOI] [PubMed] [Google Scholar]
- 34.Galluzzi L, Guilbaud E, Schmidt D et al (2024) Targeting immunogenic cell stress and death for cancer therapy. Nat Rev Drug Discov 23:445–460. 10.1038/s41573-024-00920-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Geels SN, Moshensky A, Sousa RS et al (2024) Interruption of the intratumor CD8+ T cell:Treg crosstalk improves the efficacy of PD-1 immunotherapy. Cancer Cell 42:1051-1066.e7. 10.1016/j.ccell.2024.05.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Yi L, Xu Z, Ma T et al (2024) T-cell subsets and cytokines are indicative of neoadjuvant chemoimmunotherapy responses in NSCLC. Cancer Immunol Immunother 73:99. 10.1007/s00262-024-03687-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Centonze G, Maisonneuve P, Simbolo M et al (2023) Lung carcinoid tumours: histology and Ki-67, the eternal rivalry. Histopathology 82:324–339. 10.1111/his.14819 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Yang L, Fan Y, Lu H (2022) Pulmonary large cell neuroendocrine carcinoma. Pathol Oncol Res 28:1610730. 10.3389/pore.2022.1610730 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data used in this study are available upon reasonable request from the corresponding author.