Therapeutic outcomes of PD-1 blockade-related regimens versus docetaxel in previously immunotherapy-treated advanced non-small cell lung cancer: a retrospective, exploratory study Authors: Ting-Ting Huang, Xin-Ying Wang, Qin Li, Peng-Fei Song Published: 2026 ABSTRACT AbstractBackgroundThe optimal treatment strategy for patients with advanced non-small cell lung cancer (NSCLC) following progression on initial immune checkpoint inhibitors (ICIs) therapy remained poorly defined. The present study aimed to identify the therapeutic outcomes of PD-1 blockade-based regimens versus docetaxel in previously immunotherapy-treated advanced NSCLC retrospectively.MethodsPatients with advanced NSCLC who failed prior PD-1/PD-L1 blockades therapy between January 2019 and December 2025 were screened retrospectively. A total of 64 patients who received PD-1 blockade-related regimens (PD-1 blockades plus chemotherapy or PD-1 blockades plus anlotinib) were treated as observation group (OG), while 63 patients who received docetaxel single-agent were deemed as control group (CG). Therapeutic outcomes including objective response rate (ORR), disease control rate (DCR), duration of response (DoR), progression-free survival (PFS), overall survival (OS) and treatment related adverse events (TRAEs) were compared between the two groups.ResultsORR of the 64 patients who received PD-1 blockade-based regimens in OG was 26.6% compared with 15.9% of the 63 patients who received docetaxel monotherapy in CG (P = 0.141). DCR was significantly higher in OG at 82.8% compared with 54.0% in CG (P < 0.001). The median DoR of responders in the OG and CG was 7.1 months (95%CI: 2.5–11.7) and 3.1 months (95%CI: 1.9–4.3), respectively (P = 0.056). The median PFS of 64 patients in OG and 63 patients in CG was 6.5 months and 3.3 months, respectively (P = 0.008). Similarly, the median OS was dramatically longer with PD-1 blockade-related regimens versus docetaxel group (median OS: 16.8 vs. 9.5 months, P = 0.013). After adjustment for available prognostic factors including best response to prior ICIs therapy, the treatment effect remained directionally consistent for DCR, PFS and OS, whereas ORR remained statistically non-significant. The adjusted estimates were as follows: ORR, adjusted OR = 1.85, 95%CI: 0.72–4.81, P = 0.209; DCR, adjusted OR = 4.01, 95%CI: 1.56–10.42, P = 0.004; PFS, adjusted HR = 0.59, 95%CI: 0.38–0.89, P = 0.011; and OS, adjusted HR = 0.62, 95%CI: 0.39–0.93, P = 0.020. TRAEs of any grade occurred in 82.8% of patients receiving PD-1 blockade-related regimens vs. 81.0% of patients receiving docetaxel monotherapy, and grade ≥ 3 TRAEs in OG and CG was 42.2% and 34.9%, respectively. Common TRAEs in both groups included fatigue, nausea/vomiting and hematologic toxicity.ConclusionPD-1 blockade-based regimens were associated with promising disease-control and survival outcomes compared with docetaxel monotherapy with an acceptable safety profile among patients with previously ICIs-treated advanced NSCLC in clinical practice. Given the retrospective, non-randomized and exploratory design, prospective clinical trials are warranted to confirm these findings subsequently.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12885-026-16412-5.Keywords: Non-small cell lung cancer, Previously ICIs-treated, PD-1 blockades, Docetaxel, Effectiveness, Safety Background According to GLOBOCAN 2022, lung cancer accounted for almost 2.5 million new cases and 1.8 million deaths globally, underscoring the persistent lethality of this disease despite advances in screening, molecular diagnostics and systemic therapy [1]. In China, there were approximately 1.06 million new cases and 0.73 million deaths in 2022, indicating that lung cancer remained the most common cause of cancer death and one of the most frequently diagnosed malignancies [2]. Non-small cell lung cancer (NSCLC) constituted the vast majority of lung cancers, and a considerable proportion of patients still presented with unresectable locally advanced or metastatic disease at initial diagnosis [3]. Over the past decade, the therapeutic landscape of advanced NSCLC has been transformed by immune checkpoint inhibitors (ICIs), especially agents targeting programmed death-1 (PD-1) or programmed death ligand-1 (PD-L1) [4]. For patients without actionable oncogenic driver alterations, PD-1/PD-L1 blockade, used either alone or in combination with chemotherapy, has become a cornerstone of first-line treatment and has produced durable survival benefits in selected subsets of patients. Despite this progress, the clinical success of first-line immunotherapy was incomplete and unevenly distributed. Only a subset of patients achieved deep or durable responses, whereas many others experienced primary resistance, early progression, or eventual acquired resistance after an initial period of benefit [5]. In addition, some patients discontinued treatment because of immune-related adverse events rather than unequivocal radiographic progression, creating a clinically heterogeneous population of previously immunotherapy-treated patients whose subsequent management was often challenging [6]. The post-ICIs setting therefore represented a growing and clinically heterogeneous population for whom optimal sequencing was not fully defined [7]. Historically, docetaxel served as one of the principal benchmarks for previously treated advanced NSCLC and remained a pragmatic control regimen in the post-immunotherapy setting [8]. Although the addition of ramucirumab to docetaxel improved survival outcomes over docetaxel alone in the phase III REVEL trial, the absolute benefit was modest, and real-world adoption might be influenced by access, toxicity considerations, and regional practice patterns [9]. In patients previously exposed to immunotherapy, retrospective studies of salvage chemotherapy suggested that conventional chemotherapy remained active but far from optimal with limited disease control and relatively short progression-free intervals [10]. The multicenter CLARITY study and related real-world reports reinforced the notion that chemotherapy after immunotherapy was feasible but did not fully overcome the poor prognosis of ICI-pretreated disease [11]. More recently, newer agents such as datopotamab deruxtecan challenged docetaxel in the pretreated NSCLC space: in the phase III TROPION-Lung01 study, datopotamab deruxtecan significantly improved progression-free survival (PFS) versus docetaxel, but overall survival (OS) did not reach statistical significance at the primary analysis. Taken together, these data suggested that docetaxel remained an important clinical comparator not because it was ideal, but because truly superior and broadly validated alternatives were still being defined [12]. Against this background, rechallenge or continuation of ICIs emerged as an attractive but controversial therapeutic concept. The rationale was intuitive: if resistance to prior immunotherapy was heterogeneous rather than absolute, some patients might still derive benefit from renewed PD-1/PD-L1 inhibition under a different treatment context. However, the clinical literature remained mixed. Systematic reviews and meta-analyses suggested that ICIs rechallenge might yield clinically meaningful benefit in selected patients with advanced NSCLC, and a broader meta-analysis across solid tumors also suggested that rechallenge might be feasible without a dramatic increase in severe immune-related toxicity [13]. At the same time, these studies consistently highlighted substantial heterogeneity in patient selection, prior response, treatment-free interval, reason for discontinuation of the initial ICIs, and the composition of the rechallenge regimen. In particular, simple ICIs monotherapy rechallenge after progression generally showed limited activity in unselected patients, whereas growing retrospective evidence suggested that outcomes might be improved in selected cases with combination-based strategies, thereby shifting attention from mere re-exposure to regimen design [14]. Therefore, the key question was no longer whether post-ICIs immunotherapy could ever work, but rather in whom, under what biological conditions, and in combination with which partner drugs it might retain clinical value. There was also a strong biological rationale for combining PD-1 blockade with chemotherapy or antiangiogenic therapy after prior immunotherapy exposure. Chemotherapy was no longer viewed solely as a cytotoxic modality: growing mechanistic evidence indicated that selected cytotoxic agents might improve immunogenic cell death, enhance tumor antigen release, facilitate antigen presentation, and partially remodel the tumor microenvironment in ways that supported antitumor T-cell activity [15]. Likewise, vascular endothelial growth factor (VEGF)-driven angiogenesis was increasingly understood as an immunoregulatory process rather than a purely vascular one. VEGF might suppress dendritic-cell maturation, reduce effective T-cell trafficking into tumors, and promote an immunosuppressive microenvironment. From this perspective, antiangiogenic agents might exert an immune-modulating effect that could help recondition resistant tumors and restore, at least in part, their sensitivity to ICIs [16]. These mechanistic considerations provided a plausible foundation for post-ICIs combination strategies based on PD-1 blockade plus chemotherapy or PD-1 blockade plus antiangiogenic therapy. In NSCLC, these concepts had already begun to generate clinical signals, although the available evidence was still dominated by retrospective and non-randomized studies. Several Chinese real-world analyses reported encouraging outcomes with anlotinib plus PD-1 inhibitors or with broader anlotinib-related regimens in previously treated or previously immunotherapy-treated advanced NSCLC [7]. These studies were important because they reflected a treatment pattern that was particularly relevant in China, where anlotinib was accessible and widely used in later-line practice. Unfortunately, CONTACT-01, a phase Ⅲ trial comparing atezolizumab plus cabozantinib with docetaxel after prior ICIs and chemotherapy exposure, improved PFS but failed to show an OS advantage (median OS: 10.7 vs. 10.5 months, P = 0.367) [17]. This negative OS result was highly instructive: it suggested that the concept of immunotherapy plus antiangiogenic therapy after prior immunotherapy could not be generalized indiscriminately, and that the success of post-ICIs strategies was likely to depend on the specific regimen, line of therapy, patient population, and clinical context. This evolving evidence also highlighted that studies in this area must be positioned carefully. Recent publications already explored anlotinib plus PD-1 blockade, anlotinib-related regimens, and ICIs rechallenge in previously immunotherapy-treated NSCLC, and highly similar real-world questions had only recently begun to appear in the literature [18]. Therefore, the value of an additional study lied less in claiming absolute novelty and more in providing incremental, clinically grounded comparative evidence using clearly defined treatment groups and real-world outcomes. In that context, retrospective comparative studies remained meaningful when they addressed a practical question faced by oncologists every day: whether maintaining PD-1 blockade as part of a combination strategy offered a more favorable balance of effectiveness and safety than reverting to conventional single-agent chemotherapy after prior immunotherapy failure. In the present study, the treatment strategy of interest was not ICIs rechallenge in the abstract, but a clinically specific comparison between PD-1 blockade-based regimens and single-agent docetaxel in patients with previously immunotherapy-treated advanced NSCLC. PD-1 blockade-based group included two routinely encountered real-world patterns: PD-1 blockade plus chemotherapy and PD-1 blockade plus anlotinib, whereas the control group received only docetaxel monotherapy. This design was clinically relevant because it compared a PD-1 maintained in combination strategy against a familiar chemotherapy benchmark in a population for whom prospective treatment standards remained unsettled. Accordingly, this retrospective exploratory study was conducted to evaluate the effectiveness and safety of PD-1 blockade-based regimens versus docetaxel monotherapy in previously immunotherapy-treated advanced NSCLC. Materials and methods Study protocol and eligible criteria The present study was a retrospective, observational cohort study designed to compare the effectiveness and safety of PD-1 blockade–based regimens versus docetaxel monotherapy among patients with previously immunotherapy-treated advanced NSCLC. We retrospectively reviewed the medical records of patients treated between January 2019 and December 2025. Eligible patients had advanced NSCLC and had previously received at least one cycle of PD-1/PD-L1 inhibitor-based therapy before subsequent treatment with either a PD-1 blockade-based regimen or docetaxel monotherapy in routine clinical practice. Patients were eligible for inclusion if they met all of the following criteria: (1) histologically confirmed stage ⅢB–Ⅳ NSCLC; (2) age ≥ 18 years; (3) Eastern cooperative oncology group (ECOG) performance status of 0–2; (4) prior exposure to ICIs for at least one cycle, followed by investigator-assessed treatment failure due to radiographic/clinical progression or intolerance; (5) receipt of at least one cycle of post-ICIs treatment consisting of either a PD-1 blockade-based regimen or docetaxel monotherapy; (6) at least one measurable lesion according to Response Evaluation Criteria in Solid Tumors (RECIST), version 1.1; (7) availability of adequate clinical data for baseline characteristics, treatment exposure, tumor response, adverse events, and follow-up. The exclusion criteria were as follows: (1) concurrent malignant disease other than NSCLC; (2) severe life-threatening comorbidities that could materially affect survival assessment; (3) actionable driver alterations, including EGFR, ALK or ROS1, for which targeted therapy was indicated; (4) receipt of other concomitant systemic antitumor agents during the study treatment period; (5) incomplete clinical records that prevented reliable evaluation of efficacy or safety outcomes. Ultimately, A total of 127 patients satisfied the eligibility criteria and were included in the final analysis. Of these, 64 patients received PD-1 blockade-based regimens and 63 received docetaxel monotherapy. The patient selection process was shown in Fig. 1. Because of the retrospective real-world design, treatment assignment was not randomized and was determined by the treating physician together with patient preference and clinical judgment. Study profile of this retrospective study The study protocol was reviewed and approved by the Ethics Committee of the First People’s Hospital of Lian Yun Gang (approval No. LW-20260302001-01). The study was conducted in accordance with the Declaration of Helsinki. Given the retrospective nature of the study and the use of anonymized clinical data, the requirement for informed consent was waived by the ethics committee. Treatment exposure and group definition Patients were assigned to the observation group (OG) if they received a PD-1 blockade-related regimen after failure of prior ICIs therapy. Two treatment patterns were included in this group: PD-1 blockade plus chemotherapy and PD-1 blockade plus anlotinib. The PD-1 inhibitors used in the OG were commercially available agents approved in China, including tislelizumab, sintilimab, camrelizumab, and pembrolizumab. These agents were administered intravenously at standard every-3-week doses in routine practice, typically on day 1 of each 21-day cycle, and treatment was continued until disease progression or unacceptable toxicity. For patients treated with PD-1 blockade plus chemotherapy, the chemotherapy partner was selected according to histologic subtype and routine clinical practice. Pemetrexed at 500 mg/m² on day 1 of each 21-day cycle was generally used for non-squamous NSCLC, whereas docetaxel at 75 mg/m² on day 1 of each 21-day cycle was used in other appropriate cases. For patients treated with PD-1 blockade plus anlotinib, anlotinib was administered orally at 8–12 mg once daily on days 1–14 of a 21-day cycle. Combination treatment was continued until disease progression or unacceptable toxicity. Patients were assigned to the control group (CG) if they received docetaxel monotherapy after prior ICIs failure. In the CG, docetaxel was administered at a standard dose of 75 mg/m² on day 1 of each 21-day cycle with dose modifications permitted when clinically indicated. Treatment was continued until disease progression or unacceptable toxicity. Dose delays, interruptions, and dose reductions in both groups were recorded according to routine clinical documentation. Sample size consideration Sample size of this study was determined according to the primary endpoint of OS in study using PASS 15.0 software retrospectively. Previous study indicated that docetaxel monotherapy achieved a median OS of 8.5 months among patients with previously-treated NSCLC [19]. This study assumed that PD-1 blockade–based regimen might prolong the median OS from 8.5 months to 15.5 months. A two-sided log-rank test with an overall sample size of 107 subjects (53 in the CG and 54 in the OG) achieved 80% power at a two-sided α of 0.05 significance level to detect a hazard ratio (HR) of 0.55. The study lasted for 60 months periods of which subject screening occurred in 48 months retrospectively. Considering a 15% drop-out rate, 63 patients should be screened in each group. Ultimately, 64 and 63 patients were participated in OG and CG, respectively. Data collection and outcomes assessment Baseline demographic and clinical data were extracted from the electronic medical record system. Variables collected for analysis included age, sex, smoking history, ECOG performance status, histologic subtype, disease stage, metastatic sites, use of local therapy, number of previous systemic treatment lines, type of prior immunotherapy, reason for prior ICIs discontinuation, duration of prior ICIs exposure, and burden of metastatic disease. Data on tumor PD-L1 expression and best response to prior ICIs treatment were extracted from the medical records when available. However, these variables were not available for some patients. The primary endpoint of the study was OS. Secondary endpoints included objective response rate (ORR), disease control rate (DCR), duration of response (DoR), PFS and safety profile. Tumor assessment was performed every 6–8 weeks using computed tomography or magnetic resonance imaging, according to routine clinical follow-up. Radiographic response was evaluated using RECIST version 1.1. Best overall response was categorized as complete response (CR), partial response (PR), stable disease (SD), or progressive disease (PD). ORR was defined as the proportion of patients whose best overall response was CR or PR. DCR was defined as the proportion of patients whose best overall response was CR, PR or SD. Patients who developed clinical progression or intolerable toxicity before the first post-baseline radiologic assessment were not considered evaluable for radiographic response and were counted as non-responders in the ORR and DCR analyses. This approach was applied to preserve a consistent denominator across treatment groups in the retrospective intent-to-treat efficacy analysis. For patients who achieved an objective response, DoR was defined as the interval from the first documented CR or PR to disease progression or death from any cause. PFS was defined as the time from initiation of study treatment to radiographic progression or death, whichever occurred first. OS was defined as the time from initiation of study treatment to death from any cause. Patients who were alive and progression-free at the last available follow-up were censored on that date [20]. Survival follow-up was obtained through outpatient records and/or telephone contact. The data cut-off date for the present analysis was March 15, 2026. Safety assessment Safety was evaluated by reviewing treatment-related adverse events (TRAEs) documented in the medical records during study treatment. Adverse events were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE), version 5.0. The incidence of any-grade TRAEs and grade ≥ 3 TRAEs was summarized in each treatment group. Particular attention was paid to immune-related adverse events in the PD-1 blockade-based group, including pneumonitis, hepatitis, and endocrinopathies, as well as to conventional cytotoxic toxicities such as myelosuppression and peripheral neuropathy. Events leading to treatment discontinuation and treatment-related deaths were also recorded. Statistical analysis All statistical analyses were performed using SPSS version 25.0 and Stata version 14.0. Efficacy analyses were conducted on a retrospective intent-to-treat basis, with all eligible patients included in the denominator of the corresponding group. Categorical variables were summarized as numbers and percentages and were compared using the chi-square test or Fisher’s exact test, as appropriate. Continuous variables were compared using the Mann–Whitney U test when appropriate for data distribution. Time-to-event endpoints were analyzed using the Kaplan–Meier method. Median survival times and corresponding 95% confidence intervals (CIs) were estimated for DoR, PFS, and OS, and survival curves were compared using the log-rank test. Univariable Cox proportional hazards models were used to estimate hazard ratio (HR) and 95%CIs for PFS and OS in the PD-1 blockade-based group relative to the docetaxel group. Given the clinical relevance of prior ICIs sensitivity, adjusted analyses were performed for the reported efficacy endpoints whenever statistically appropriate. Multivariable logistic regression models were used to estimate adjusted odds ratios (ORs) and 95%CIs for ORR and DCR. Multivariable Cox proportional hazards models were used to estimate adjusted HRs and 95%CIs for PFS and OS. Treatment group was included as the main explanatory variable. Covariates entered into the adjusted models included ECOG performance status, pathological stage, use of local therapy, number of previous systemic treatment lines, duration of prior ICIs exposure, best response to prior ICIs therapy and number of metastatic lesions. Best response to prior ICIs therapy was incorporated because it represented a clinically relevant determinant of subsequent treatment selection and post-ICIs outcomes. DoR was analyzed descriptively using the Kaplan-Meier method among responders only, because the number of patients with objective response was limited and multivariable modeling in this responder-only subset would be statistically unstable. All tests were two-sided, and P < 0.05 was considered statistically significant. Given the exploratory nature of the study, all inferential analyses were interpreted cautiously. Results Baseline characteristics A total of 64 patients in the OG and 63 patients in the CG were included in this study. Baseline demographics and disease characteristics were generally balanced between the two groups as shown in Table 1. The median age was 66 years (range: 22–82 years) in the OG and 65 years (range: 19–79 years) in the CG. Both groups were predominantly male (65.6% and 61.9% male) and had a high proportion of smokers (~ 87%) reflecting typical NSCLC demographics. The majority of patients had ECOG performance status 0–1 at treatment initiation (68.7% vs. 71.4%). Nearly all patients had stage Ⅳ (89.1% vs. 87.3%), a small minority presented with stage Ⅲb. Adenocarcinoma slightly exceeded squamous in both groups (56.3% vs. 54.0%). Combination with local therapy was low in both arms (approximately 20% in each). In the OG, 43.8% had only 1 prior line (typically an ICIs-based first-line) and 56.2% had ≥ 2 prior lines, whereas in the docetaxel arm 41.3% had 1 line and 58.7% had ≥ 2 lines before the study therapy. All patients had received at least one PD-1/PD-L1 inhibitor previously; the type of prior immunotherapy was similar between groups (79.7% PD-1 blockades and 20.3% PD-L1 blockades in OG vs. 84.1% and 15.9% in CG, P = 0.516). The reason for discontinuation of prior immunotherapy was progression in the majority of patients (approximately 70%) and about 30% was immune-related toxicity or intolerance. Duration of previous ICIs treatment < 6 months was observed in 54.7% and 49.2% in OG and CG, respectively. The number of metastatic lesions was comparable; about 72%-75% of patients in each group had ≤ 3 distinct metastatic lesions. PD-L1 expression was available in 75 patients. PD-L1 tumor proportion score (TPS) distribution was comparable between the observation and control groups. PD-L1 TPS ≥ 50% was observed in 10 patients (15.6%) in the observation group and 8 patients (12.7%) in the control group; PD-L1 TPS < 50% was observed in 31 patients (48.4%) and 26 patients (41.2%), respectively; and PD-L1 TPS was not available in 23 patients (36.0%) and 29 patients (46.1%), respectively (χ²=1.345, P = 0.510). In the observation and control groups, CR/PR of best response to prior ICIs therapy was observed in 18 patients (28.1%) and 16 patients (25.4%), SD in 20 patients (31.3%) and 18 patients (28.6%), PD in 8 patients (12.5%) and 10 patients (15.9%), and unavailable response data in 18 patients (28.1%) and 19 patients (30.2%), respectively. No significant differences in any baseline variable were observed, indicating the two groups were sufficiently comparable for outcome analysis (all P > 0.5, Table 1). Table 1Comparison of baseline clinical characteristics between observation group and control groupBaseline characteristicsObservation group (N = 64)Control group (N = 63)χ² P Age (years) Median (range)66 (22–82)65 (19–79)NA0.616Gender Male42 (65.6)39 (61.9)0.1900.663 Female22 (34.4)24 (38.1)ECOG performance status 0–144 (68.7)45 (71.4)0.1090.742 220 (31.3)18 (28.6)Pathological Staging Ⅲb7 (10.9)8 (12.7)0.0950.759 Ⅳ57 (89.1)55 (87.3)Histology Adenocarcinoma36 (56.3)34 (54.0)0.0670.796 Squamous carcinoma28 (43.7)29 (46.0)Smoking Status Non-smoker8 (12.5)8 (12.7)0.0010.973 Smoker56 (87.5)55 (87.3)Combination with local therapy Yes13 (20.3)15 (23.8)0.2260.635 No51 (79.7)48 (76.2)Previous treatment lines 128 (43.8)26 (41.3)0.0800.777 ≥ 236 (56.2)37 (58.7)Failure status of previous ICIs Progression46 (71.9)44 (69.8)0.0640.801 Intolerance18 (28.1)19 (30.2)Previous ICIs regimens PD-151 (79.7)53 (84.1)0.4220.516 PD-L113 (20.3)10 (15.9)Duration of previous ICIs treatment < 6 months35 (54.7)31 (49.2)0.3820.536 ≥ 6 months29 (45.3)32 (50.8)Best response to prior ICIs CR/PR18 (28.1)16 (25.4) SD20 (31.3)18 (28.6)0.4640.927 PD8 (12.5)10 (15.9) NA18 (28.1)19 (30.2)Number of metastatic lesions ≤ 346 (71.9)47 (74.6)0.1210.728 > 318 (28.1)16 (25.4)PD-L1 TPS ≥ 50%10 (15.6)8 (12.7) < 50%31 (48.4)26 (41.2)1.3450.510 NA23 (36.0)29 (46.1)ECOG Eastern Cooperative Oncology Group, ICIs immune checkpoint inhibitors, PD-1 Programmed cell death protein 1, PD-L1 Programmed cell death ligand 1, TPS tumor proportion score, NA not available Comparison of baseline clinical characteristics between observation group and control group ECOG Eastern Cooperative Oncology Group, ICIs immune checkpoint inhibitors, PD-1 Programmed cell death protein 1, PD-L1 Programmed cell death ligand 1, TPS tumor proportion score, NA not available Within the PD-1 blockade-related group, 29 patients (45%) received PD-1 blockade plus chemotherapy and 35 (54.7%) received PD-1 blockade plus anlotinib. Among those treated with chemo-immunotherapy rechallenge, docetaxel was the chemotherapy partner in 17 patients and pemetrexed in 12. The PD-1 inhibitors used in this group included tislelizumab (39.1%), sintilimab (29.7%), camrelizumab (17.2%), and pembrolizumab (14.1%). All patients in the control group received single-agent docetaxel, and none received concurrent antiangiogenic therapy. Tumor response Tumor response was evaluated in all enrolled patients on an intent-to-treat basis. Patients who developed clinical progression or intolerable toxicity before the first radiologic assessment were counted as non-responders in the response analysis. No CR were observed in either group. In the PD-1 blockade-related group, 17 patients achieved a PR, 36 had SD, 6 had PD, and 5 were unevaluable for radiographic response, resulting in an ORR of 26.6% (95%CI: 16.3%-39.1%) and a DCR of 82.8% (95%CI: 71.3%-91.1%). In the docetaxel group, 10 patients achieved a PR, 24 had SD, 22 had PD, and 7 were unevaluable, corresponding to an ORR of 15.9% (95%CI: 7.9%-27.3%) and a DCR of 54.0% (95%CI: 40.9%-66.6%). Although ORR was numerically higher with PD-1 blockade-related treatment than with docetaxel monotherapy, the difference did not reach statistical significance (26.6% vs. 15.9%, χ²=2.167, P = 0.141). By contrast, disease control was significantly better in the PD-1 blockade-related group, with a DCR of 82.8% compared with 54.0% in the docetaxel group (χ²=12.242, P < 0.001). As shown in Fig. 2, the waterfall plot of best percentage change in target lesions from baseline further illustrated a more favorable pattern of tumor shrinkage in the PD-1 blockade-related group, with mean change values of -16.1% versus 10.1% in the docetaxel group. Noteworthily, one 51-year-old male patient diagnosed with squamous carcinoma who was intolerable to previous ICIs combined with chemotherapy previously and received pembrolizumab combined with anlotinib in for over 20 months and this patient was still in the treatment of pembrolizumab plus anlotinib at the date of data cut-off, radiological results indicated that the patient’s target lesions showed a dramatical reduction of over 60%. The radiological scan comparison before and after treatment was shown in Fig. 3. Best change in waterfall plot of target lesion from baseline after treatment of PD-1 blockade-based regimens and docetaxel monotherapy Radiological results of the changes for target lesions in the lung site of a male patient before and after the treatment of pembrolizumab plus anlotinib Axial contrast-enhanced chest CT images show a left thoracic lesion adjacent to the pleura (red arrows). Compared with the baseline scan (left), the follow-up scan after treatment (right) demonstrates a marked reduction in lesion size, indicating a favorable radiographic response Duration of response was evaluated among patients who achieved an objective response. As exhibited in Fig. 4, median DoR was 7.1 months (95%CI: 2.5–11.7) in the PD-1 blockade-related group and 3.1 months (95%CI: 1.9–4.3) in the docetaxel group, showing a strong numerical advantage for PD-1 blockade-based treatment and a borderline difference by log-rank testing (χ²=3.660, P = 0.056). Responses in the PD-1 blockade-related group also appeared more durable over time, with a 12-month DoR rate of 40.3%, compared with 10.0% in the docetaxel group. Comparison of Kaplan-Meier Duration of response survival curve between PD-1 blockade-based regimens and docetaxel monotherapy Progression-free survival At the data cutoff date of March 15, 2026, the median follow-up duration was 14.5 months (range: 0.9–36.8 months) in the OG and 9.3 months (range: 0.7–33.4 months) in the CG. By that time, 42 progression or death events (65.6%) had occurred in the OG and 49 (77.8%) in the CG. Kaplan-Meier curves for PFS were shown in Fig. 5. The curves began to separate early and remained apart over time. Median PFS was 6.5 months (95% CI, 3.1–9.9 months) in the OG and 3.3 months (95% CI, 2.2–4.4 months) in the CG. The corresponding 12-month PFS rates were 42.7% and 22.7%, respectively. Overall, PD-1 blockade-related regimens were associated with a significantly longer PFS than docetaxel monotherapy (χ²=7.101, P = 0.008). In univariable Cox regression analysis, PD-1 blockade-related treatment was associated with a lower risk of progression or death relative to docetaxel monotherapy (HR = 0.57, 95%CI: 0.38–0.87). Comparison of Kaplan-Meier Progression-free survival curve between PD-1 blockade-based regimens and docetaxel monotherapy Overall survival OS analysis also favored the OG. At the time of analysis, 37 death events (57.8%) had occurred in the OG and 44 (69.8%) in the CG. The Kaplan-Meier OS curves were presented in Fig. 6. Median OS was 16.8 months (95%CI: 13.5–20.1 months) in the OG and 9.5 months (95%CI: 4.8–14.2 months) in the CG. The 24-month OS rates were 40.6% and 30.7%, respectively. The difference in OS between groups was statistically significant (χ²=6.113, P = 0.013). Univariable Cox regression analysis showed that PD-1 blockade-based regimens were associated with a reduced risk of death compared with docetaxel monotherapy (HR = 0.58, 95%CI: 0.37–0.90). Among the 64 patients receiving PD-1 blockade-based regimens, 46 patients had discontinued prior ICIs therapy because of disease progression and 18 because of treatment-related intolerance. Median OS was 22.3 months (95%CI: 11.96–32.64) in the intolerance subgroup and 16.5 months (95%CI: 11.04–21.60) in the progression subgroup. Although OS was numerically longer in the intolerance subgroup, the difference was not statistically significant (log-rank P = 0.274, supplementary Figure S1). In the docetaxel monotherapy group, OS curves stratified by prior ICIs discontinuation reason were nearly overlapping and were therefore not displayed as an additional supplementary figure. Comparison of Kaplan-Meier Overall survival curve between PD-1 blockade-based regimens and docetaxel monotherapy Adjusted analysis of reported efficacy endpoints To further address the potential influence of prior ICIs sensitivity, adjusted analyses were performed for the reported efficacy endpoints by incorporating best response to prior ICIs therapy and other major baseline prognostic factors into the models. As shown in Table 2, in the multivariable logistic regression models, ORR remained statistically non-significant after adjustment (adjusted OR = 1.85, 95%CI: 0.72–4.81; P = 0.209), whereas PD-1 blockade-related regimens remained associated with a higher probability of disease control compared with docetaxel monotherapy (adjusted OR for DCR = 4.01, 95%CI: 1.56–10.42; P = 0.004). In the multivariable Cox regression models, PD-1 blockade-related regimens remained associated with improved PFS (adjusted HR = 0.59, 95%CI: 0.38–0.89; P = 0.011) and OS (adjusted HR = 0.62, 95%CI: 0.39–0.93; P = 0.020). These findings indicated that the observed favorable direction of treatment effect was not fully explained by available baseline prognostic factors, including best response to prior ICIs therapy. However, residual confounding could not be excluded because of the retrospective, non-randomized design. Table 2Adjusted analysis of reported efficacy endpoints after incorporation of best response to prior ICIs therapyEndpointsStatistical modelEffect measureAdjusted estimate95%CI P ORRMultivariable logistic regressionAdjusted OR1.850.72–4.810.209DCRMultivariable logistic regressionAdjusted OR4.011.56–10.420.004PFSMultivariable Cox regressionAdjusted HR0.590.38–0.890.011OSMultivariable Cox regressionAdjusted HR0.620.39–0.930.020Adjusted models included treatment group, ECOG performance status, pathological stage, use of local therapy, previous treatment lines, duration of prior ICIs treatment, best response to prior ICIs therapy and number of metastatic lesionsORR objective response rate, DCR disease control rate, PFS progression-free survival, OS overall survival, OR odds ratio, HR hazard ratio, CI confidence interval, ICIs immune checkpoint inhibitors Adjusted analysis of reported efficacy endpoints after incorporation of best response to prior ICIs therapy Adjusted models included treatment group, ECOG performance status, pathological stage, use of local therapy, previous treatment lines, duration of prior ICIs treatment, best response to prior ICIs therapy and number of metastatic lesions ORR objective response rate, DCR disease control rate, PFS progression-free survival, OS overall survival, OR odds ratio, HR hazard ratio, CI confidence interval, ICIs immune checkpoint inhibitors Safety All patients who received at least one dose of the study treatment were included and collected in safety analysis retrospectively. As shown in Table 3, In the observation group, 53 patients (82.8%) who received PD-1 blockade-based regimens experienced any grade TRAEs, which was similar to the 51 patients (81.0%) experienced any grade TRAEs in the docetaxel group. The incidence of grade ≥ 3 TRAEs was slightly higher in OG at 42.2% (27/64) compared to 34.9% (22/63) in CG, but this difference was not statistically significant (χ²=0.708, P = 0.400). Table 3Safety profile of patients in observation group and control groupAdverse eventsObservation group (N = 64)Control group (N = 63)All grades (N, %)Grade ≥ 3 (N, %)All grades (N, %)Grade ≥ 3 (N, %)Any grade53 (82.8)27 (42.2)51 (81.0)22 (34.9)Fatigue34 (53.1)7 (10.9)31 (49.2)6 (9.5)Nausea and vomiting23 (35.9)8 (12.5)30 (47.6)8 (12.7)Hematologic toxicity19 (29.7)8 (12.5)25 (39.7)12 (19.0)Hepatotoxicity16 (25.0)4 (6.3)19 (30.2)6 (9.5)Alopecia8 (12.5)1 (1.6)15 (23.8)4 (6.3)Constipation9 (14.1)2 (3.1)12 (19.0)2 (3.2)Hypertension11 (17.2)6 (9.4)2 (3.2)0 (0.0)Rash9 (14.1)3 (4.7)3 (4.8)0 (0.0)Stomatitis5 (7.8)0 (0.0)9 (14.3)0 (0.0)Hand-foot syndrome7 (10.9)2 (3.1)2 (3.2)0 (0.0)Pneumonia6 (9.4)1 (1.6)1 (1.6)0 (0.0) Safety profile of patients in observation group and control group Among the 64 patients who received PD-1 blockade-based regimens, the most common TRAEs of any grade (> 10%) were fatigue (53.1%), nausea and vomiting (35.9%), hematologic toxicity (29.7%), hepatotoxicity (25.0%), hypertension (17.2%), constipation (14.1%), alopecia (12.5%), hand-foot syndrome (10.9%). These aligned with known toxicities of anlotinib (hypertension and hand–foot syndrome), PD-1 inhibitors (hepatotoxicity and rash) and chemotherapy (nausea and vomiting, hematologic toxicity, alopecia). And the comment TRAEs (incidence > 10%) among the 63 patients who received docetaxel monotherapy were fatigue (49.2%), nausea and vomiting (47.6%), hematologic toxicity (39.7%), hepatotoxicity (30.2%), alopecia (23.8%), constipation (19.0%) and stomatitis (14.3%). Overall, the toxicity profile observed in the OG was consistent with the known adverse-event spectrum of PD-1 inhibitors, chemotherapy and anlotinib. No treatment-related deaths were observed in either group. To address the heterogeneity within the PD-1 blockade-based group, exploratory three groups analysis were performed by separating patients into PD-1 blockade plus chemotherapy, PD-1 blockade plus anlotinib and docetaxel monotherapy group. Baseline characteristics were generally comparable among the three groups as shown in Supplementary Table S1. In the three-group OS analysis, median OS was 16.2 months in the PD-1 blockade plus chemotherapy group, 17.7 months in the PD-1 blockade plus anlotinib group, and 9.5 months in the docetaxel group with an overall log-rank P value of 0.039 (Supplementary Figure S2). The three-group safety analysis showed different toxicity profile across regimens: hematologic toxicity was more frequent in the PD-1 blockade plus chemotherapy group and hypertension and hand-foot syndrome were more frequent in the PD-1 blockade plus anlotinib group (Supplementary Table S2). Discussion In this retrospective exploratory study, PD-1 blockade-related regimens were associated with more favorable clinical outcomes than docetaxel monotherapy in patients with advanced NSCLC previously treated with ICIs. Although the difference in ORR did not reach statistical significance, PD-1 blockade-related treatment achieved a substantially higher DCR and was further associated with prolonged PFS and OS, while maintaining a broadly manageable safety profile. Taken together, these findings supported the clinical plausibility that reintroducing PD-1 blockade as part of a combination strategy might still confer benefit in a subset of patients after prior ICIs failure, and that this approach might compare favorably with conventional single-agent docetaxel in routine practice. The relevance of this question has increased considerably in recent years. As first-line treatment for advanced NSCLC without targetable driver alterations has increasingly shifted toward PD-1/PD-L1 inhibitor-based therapy, either alone or in combination with chemotherapy, post-immunotherapy treatment has become an increasingly common and clinically important scenario rather than a relatively uncommon salvage setting [21]. However, despite the rapid expansion of immunotherapy in earlier lines, the optimal strategy after prior ICIs exposure remains incompletely defined. Available evidence suggests that ICIs rechallenge should not be regarded as a uniformly effective approach for all patients, especially for unselected populations who experience progression during initial immunotherapy; rather, its value appears to depend on patient selection, prior sensitivity to immunotherapy, treatment context, and the specific combination partner used [22]. From the perspective of cohort composition, the present study appears clinically representative of the real-world post-immunotherapy population encountered in China. Most patients had stage IV disease, over half had already received at least two prior lines of therapy, and the majority had previously been exposed to PD-1 inhibitors rather than PD-L1 inhibitors. This pattern is broadly consistent with the recent treatment environment in China, where domestically available PD-1 inhibitors have been widely incorporated into routine NSCLC care [7, 23]. In addition, nearly one-third of patients in our cohort discontinued prior ICIs because of intolerance rather than unequivocal progression. This is noteworthy because, in actual clinical practice, subsequent treatment decisions are not made solely after radiographic resistance; they are also shaped by toxicity, performance status, patient preference, and access to later-line options. In that sense, our cohort reflects the therapeutic complexity of real-world post-ICIs NSCLC more accurately than highly selected clinical trial populations. The efficacy findings of the present study were also broadly aligned with the growing body of retrospective evidence on PD-1/PD-L1 rechallenge or continuation strategies. In particular, our ORR and DCR results were close to those reported in several recent Chinese real-world studies. Dou et al. reported that anlotinib plus PD-1 blockade in previously immunotherapy-treated advanced NSCLC yielded an ORR of 23.9% and a DCR of 85.1% [23]. More recently, Zheng et al. described two PD-1 rechallenge strategies: PD-1 plus chemotherapy and PD-1 plus anlotinib—using a contemporaneous docetaxel cohort as contextual reference; their reported ORRs, DCRs, median PFS, and median OS were highly comparable to the efficacy pattern observed in our study [7]. Similarly, Fang et al. found that anlotinib-related regimens achieved a DCR of 79.6%, median PFS of 6.3 months, and median OS of 16.1 months, all of which were superior to single-agent chemotherapy in their exploratory real-world cohort [18]. Against this background, our results did not stand in isolation; rather, they reinforced an emerging and increasingly coherent signal that PD-1 blockade-based combinations might provide meaningful disease control in selected patients after prior immunotherapy exposure. An important point in our data was that the improvement in DCR was more pronounced than the improvement in ORR. This pattern deserved attention. In the salvage setting, especially after prior ICIs exposure, achieving prolonged disease stabilization might be nearly as clinically relevant as obtaining formal tumor shrinkage, because many patients were heavily pretreated, vulnerable to symptomatic deterioration, and at risk of rapid progression [24]. Therefore, the capacity of a regimen to maintain disease control might translate into tangible clinical benefit even when the rate of objective radiographic response was modest. This might partly explain why the present study showed clearer separation for DCR, PFS and OS than for ORR. A similar phenomenon has been observed in other real-world analyses in which immunotherapy-based combinations after progression improved survival-related endpoints even when ORR differences were relatively limited [25]. The survival findings in our cohort were particularly noteworthy. Median PFS and OS were both longer in the PD-1 blockade-related group than in the docetaxel group. After incorporation of best response to prior ICIs therapy and other available prognostic factors, the adjusted treatment effects remained directionally consistent: ORR remained non-significant, whereas DCR, PFS and OS continued to favor PD-1 blockade-related regimens. While these results should be interpreted cautiously because of the retrospective design, they were clinically meaningful in light of the limited outcomes typically seen with docetaxel after prior immunotherapy. In a small real-world study focusing on chemotherapy after immunotherapy, Assi et al. reported that docetaxel achieved only modest activity with a response rate of 24% and median PFS of approximately 3 months [26]. These figures were not substantially different from the outcomes observed in the docetaxel arm of our study and supported the view that chemotherapy alone remained suboptimal for many patients once resistance to immunotherapy emerged [27]. At the same time, our results should not be interpreted as proof that any immunotherapy-containing salvage regimen might outperform docetaxel. In the phase III CONTACT-01 trial, atezolizumab plus cabozantinib improved PFS compared with docetaxel but failed to demonstrate an OS advantage, underscoring that post-immunotherapy combination strategies were not uniformly successful and that regimen-specific factors mattered [17]. Likewise, in the broader previously treated NSCLC space, TROPION-Lung01 showed that datopotamab deruxtecan improved PFS versus docetaxel but did not achieve a statistically significant OS benefit in the overall population, further illustrating how difficult it remained to surpass docetaxel convincingly in later lines [12]. Against this backdrop, the adjusted DCR, PFS and OS signals observed in our study were encouraging, but they should be viewed as hypothesis-generating rather than definitive. Another important issue concerned patient selection. The literature increasingly suggested that outcomes after ICIs rechallenge were heterogeneous and that prior benefit from immunotherapy might influence subsequent sensitivity. In our analysis, best response to prior ICIs therapy was balanced between the two groups, and documented prior ICIs clinical benefit (CR/PR/SD) was associated with improved OS. These results suggested that prior ICIs sensitivity might influence subsequent outcomes after PD-1 blockade-based treatment. Therefore, although the observed survival advantage was not explained by baseline imbalance in prior ICIs response, residual confounding related to immune sensitivity could not be completely excluded. The findings should therefore be interpreted as exploratory and hypothesis-generating rather than definitive evidence of treatment superiority. Previous study concluded that rechallenge immunotherapy appeared to have limited efficacy in unselected patients who progressed during initial immunotherapy, but might be more promising in better-defined subgroups, particularly those who previously derived clinical benefit or relapsed after treatment completion rather than demonstrating clear primary resistance [28]. In a retrospective study by Yan et al., ICIs rechallenge did not improve outcomes across the entire cohort, but patients who had responded to first-line ICIs showed significantly better results than those who were resistant to initial immunotherapy [14]. Similarly, Cheng et al. reported that among patients with driver gene-negative NSCLC, continued immunotherapy beyond progression appeared to be more effective in those who had achieved prior disease control, including PR or SD, during initial immunotherapy [29]. More recently, Wang et al. applied criteria for acquired immune resistance and again suggested that certain biologically selected patients might still derive benefit from ICIs rechallenge [30]. Taken together, these data suggested that post-ICIs NSCLC should not be treated as a homogeneous category. Instead, the biological distinction between primary resistance, acquired resistance, and intolerance-related discontinuation might be central to understand who remained suitable for a PD-1-based salvage strategy. This issue was highly relevant to the interpretation of our own study. In our cohort, approximately half of the patients had a prior ICIs treatment duration of less than 6 months, while the remainder had more prolonged exposure. Although duration of prior ICIs therapy was not independently associated with OS in the multivariable model, the study was not powered to detect more nuanced interactions between previous immunotherapy course and rechallenge efficacy. The reason for prior ICIs discontinuation might also influence the interpretation of the present findings. Patients who discontinued prior ICIs because of disease progression might represent a population with primary or acquired immune resistance, whereas patients who discontinued because of treatment-related intolerance might retain some degree of biological sensitivity to PD-1 blockade. In our exploratory subgroup analysis within the PD-1 blockade-based group, patients who discontinued prior ICIs because of intolerance showed numerically longer OS than those who discontinued because of progression, but the difference was not statistically significant. By contrast, no apparent OS separation was observed between these subgroups in the docetaxel group. Because the intolerance subgroup was small, these findings should be interpreted descriptively, and future prospective studies should stratify patients according to the reason for prior ICIs discontinuation. It remained quite plausible that duration of prior benefit, best response to prior ICIs, cause of discontinuation, and the interval between initial ICIs discontinuation and retreatment all modified the probability of benefit from rechallenge [31]. Accordingly, one of the most important implications of our work was not simply that PD-1 blockade-related regimens outperformed docetaxel in the overall cohort, but also that future prospective studies should move beyond broad treatment comparisons and instead refine patient selection using clinically meaningful markers of prior immunotherapy sensitivity [32]. The biological rationale for PD-1 blockade-related combinations after initial ICIs failure was also increasingly compelling. Resistance to immunotherapy in NSCLC was mechanistically heterogeneous and involved tumor-intrinsic, immune-microenvironmental, and host-related factors, including impaired antigen presentation, T-cell exhaustion, interferon signaling alterations, immunosuppressive myeloid populations, and exclusion of effector lymphocytes from the tumor microenvironment [33]. Under these conditions, simple retreatment with ICIs monotherapy might be insufficient, because the same resistance architecture that limited the first course of treatment might still be present. By contrast, combination-based rechallenge strategies might alter the tumor ecosystem in ways that partially restored immunotherapy sensitivity. Chemotherapy could promote immunogenic cell death, enhance neoantigen release, increase antigen presentation, and reduce immunosuppressive cellular populations, thereby creating a more permissive immune context for renewed PD-1 blockade [34]. This helped explain why several retrospective studies, including ours, suggested that immunotherapy rechallenge might be more rational when delivered as part of a combination regimen rather than as simple monotherapy re-exposure. A similar rationale applied to the antiangiogenic component. VEGF-driven angiogenesis was understood not only as a vascular process, but also as a major mediator of immune escape [35]. Aberrant tumor vasculature impaired T-cell trafficking, promoted hypoxia, and supported an immunosuppressive tumor microenvironment enriched with regulatory T cells, myeloid-derived suppressor cells, and dysfunctional antigen-presenting cells [36]. Antiangiogenic therapy might help normalize tumor vessels, improve immune-cell infiltration, and counteract VEGF-mediated immunosuppression, thereby creating a biological basis for synergy with PD-1 blockade [37]. This concept was also supported by real-world evidence: in the large retrospective study by Li et al., patients who continued immunotherapy after progression appeared to fare best when immunotherapy was combined with antiangiogenic therapy, outperforming immunotherapy plus chemotherapy in both PFS and OS [25]. On the other hand, CONTACT-01 reminded us that not all PD-(L)1 plus antiangiogenic combinations could translate into an OS advantage [17]. Therefore, the relevance of our findings lied not in claiming that antiangiogenic-ICIs combinations were universally superior, but in suggesting that certain regimen platforms: especially those tailored to local practice patterns such as anlotinib-based treatment in China—deserved further prospective evaluation. The safety profile observed in our study was also clinically important. One of the main concerns with rechallenge strategies was the possibility of recurrent or more severe immune-related toxicity, particularly in patients who discontinued initial ICIs therapy because of intolerance [38]. In the present cohort, however, PD-1 blockade-related regimens did not produce a statistically significant increase in overall or grade ≥ 3 TRAEs compared with docetaxel monotherapy, and no treatment-related deaths were observed. The toxicity spectrum was largely predictable and consistent with the known adverse-event profiles of the component agents, including hypertension and hand-foot syndrome with anlotinib and rash, pneumonitis, or hepatic abnormalities with PD-1 blockade. This general pattern was consistent with recent real-world reports. Dou et al. described anlotinib plus PD-1 blockade as tolerable in previously immunotherapy-treated advanced NSCLC [23]. Zheng et al. similarly found manageable toxicity for both PD-1 plus chemotherapy and PD-1 plus anlotinib with grade ≥ 3 TRAEs in the low-40% range and no treatment-related deaths [7]. Feng et al. also reported that ICIs rechallenge in advanced NSCLC showed no new safety signals overall [39]. These convergent data suggested that, when appropriately selected and monitored, PD-1-related rechallenge strategies might be delivered without an obviously prohibitive safety penalty. Nevertheless, the safety implications should not be oversimplified. Reviews of ICIs rechallenge emphasized that recurrence of irAEs remained a genuine concern and that rechallenge decisions should be individualized according to the pattern and severity of prior toxicity, the urgency of disease control, and the availability of alternative treatment options [40]. In particular, retrospective datasets were vulnerable to under-reporting of lower-grade toxicities and might incompletely capture delayed immune-related events [41]. Therefore, although our study was reassuring, it should not be interpreted as evidence that rechallenge was uniformly safe in all patients who were intolerant to prior ICIs. Instead, our findings supported the more measured conclusion that PD-1 blockade-related salvage regimens appeared feasible in routine practice when accompanied by careful surveillance for hepatic toxicity, skin toxicity, pneumonitis, hypertension, and hand-foot syndrome. The PD-1 blockade-based group included two distinct strategies: PD-1 blockade plus chemotherapy and PD-1 blockade plus anlotinib. These regimens differed in biological rationale and toxicity profile. To address this heterogeneity, we performed exploratory three-group analysis comparing PD-1 blockade plus chemotherapy, PD-1 blockade plus anlotinib and docetaxel monotherapy. Both PD-1 blockade-based subgroups showed numerically longer OS than docetaxel monotherapy, while their safety profiles differed: more hematologic toxicity in the chemotherapy-combination group and more hypertension and hand-foot syndrome in the anlotinib-combination group. Nevertheless, these subgroup analyses were limited by small sample size and should be interpreted descriptively rather than as definitive evidence of superiority of one PD-1 blockade-based strategy over another. In addition, platinum-refractory status, prior taxane exposure and exact time from first-line initiation to subsequent treatment were not uniformly available, leaving residual confounding. Therefore, the findings should be considered exploratory and hypothesis-generating rather than definitive evidence of treatment superiority. Several strengths of the present study deserved emphasis. First, unlike many earlier reports that examined only a single rechallenge regimen without a comparator, our study directly contrasted PD-1 blockade-related treatment with a contemporaneous docetaxel group. This enhanced the clinical interpretability of the results because the practical decision in the post-immunotherapy setting was often not whether rechallenge was interesting in theory, but whether it performed better than standard chemotherapy in real life. Second, the observation group incorporated two of the most relevant real-world strategies used in China—PD-1 plus chemotherapy and PD-1 plus anlotinib-thereby increasing the practical relevance of the data. Third, the consistency of benefit across disease control and survival endpoints, together with the persistence of favorable treatment effects after adjusted logistic and Cox regression analyses, made the overall signal more persuasive than a single endpoint result considered in isolation. The study had important limitations that should be stated clearly. Residual confounding should be acknowledged as an important limitation of this retrospective, non-randomized study. Because treatment allocation was based on physician judgment, patient characteristics, and routine clinical decision-making rather than randomization, selection bias could not be avoided and might have favored patients selected for PD-1 blockade-based regimens. Although baseline characteristics were compared and multivariable logistic and Cox adjustment was performed for available prognostic factors, treatment selection in routine practice might still have been influenced by unmeasured clinical factors, including tumor burden, comorbidities, prior depth of response to ICIs, treatment-free interval, subsequent therapies, and comprehensive molecular features. Although best response to prior ICI therapy was additionally collected and incorporated into the baseline comparison and adjusted efficacy analyses, this information remained unavailable in a proportion of patients. Moreover, exact radiographic time to progression on prior immunotherapy was not uniformly available in the retrospective records. Duration of prior ICIs treatment was collected as an available clinical surrogate, but it could not fully distinguish primary resistance, acquired resistance, and intolerance-related discontinuation. Therefore, residual confounding related to prior ICIs sensitivity and disease tempo could not be completely excluded. Propensity score matching was considered but was not performed because of the modest sample size and the risk of further reducing the effective analytical population after matching. Instead, multivariable Cox adjustment and exploratory subgroup analyses were performed using available clinically relevant variables. However, these approaches could not fully eliminate residual confounding from unmeasured or incompletely recorded factors. Therefore, the observed survival advantage of PD-1 blockade-based regimens over docetaxel monotherapy should be interpreted as exploratory and hypothesis-generating rather than definitive evidence of treatment superiority. Additionally, the relatively small sample size also limited the robustness of the survival analyses, subgroup analyses and multivariable Cox models. Although multivariable adjustment was performed using available clinically relevant variables, the limited number of patients and events might have affected model stability and increased statistical uncertainty. Therefore, the observed survival signal should be interpreted cautiously and regarded as exploratory and hypothesis-generating. Another important limitation was the incomplete availability of molecular-marker data. Although patients with actionable EGFR, ALK or ROS1 alterations were excluded and PD-L1 TPS was added to the baseline comparison, other potentially relevant biomarkers were not systematically tested. PD-L1 expression remained the most routinely used predictive biomarker for initial PD-1/PD-L1 inhibitor therapy in metastatic NSCLC, but its role in the rechallenge setting was less clearly defined. Moreover, prior clinical benefit from ICIs might reflect greater immune sensitivity, but it should be regarded as a clinical surrogate rather than a validated biomarker. Other potentially relevant biomarkers, including KRAS, STK11, KEAP1, tumor mutational burden, circulating tumor DNA and immune-inflammatory markers, were not systematically tested. Therefore, the present study could not fully evaluate how molecular or immune-inflammatory biomarkers should guide the selection of PD-1 blockade-based rechallenge strategies. Future prospective studies should incorporate standardized PD-L1 testing, next-generation sequencing, tumor mutational burden, circulating tumor DNA and immune-inflammatory markers to better refine patient selection for PD-1 blockade-based rechallenge strategies [42]. Finally, subsequent therapies after progression were not fully standardized and might have influenced OS, as was common in observational oncology studies. From a clinical-practice perspective, the present findings suggest that PD-1 blockade-based combination regimens may represent a potential treatment option for selected patients with advanced NSCLC after prior ICIs exposure, particularly when standard later-line options are limited and docetaxel monotherapy is expected to provide only modest benefit. However, this study should not be interpreted as establishing a new standard of care. The retrospective single-center design, non-randomized treatment allocation, potential selection bias, heterogeneity of PD-1 plus chemotherapy and PD-1 plus anlotinib regimens, incomplete biomarker information, lack of detailed prior ICIs response data and modest sample size all limit the strength and generalizability of the conclusions. Therefore, the clinical implication of this study is mainly hypothesis-generating: PD-1 blockade-based rechallenge strategies may deserve further prospective evaluation in larger controlled studies. Conclusion In this retrospective exploratory cohort, PD-1 blockade-related combination regimens were associated with favorable disease-control and survival outcomes compared with docetaxel monotherapy among patients with previously immunotherapy-treated advanced NSCLC. The revised adjusted efficacy analyses and three-group analyses provided a more transparent description of the treatment effect after incorporation of best response to prior ICIs therapy and also described PD-1 plus chemotherapy and PD-1 plus anlotinib as distinct strategies. Nevertheless, because of the non-randomized design, small sample size, treatment heterogeneity, incomplete biomarker and prior-response data, and possible residual confounding, the findings should be considered hypothesis-generating. Therefore, these findings should not be interpreted as establishing the superiority of a single PD-1 blockade-based regimen. Rather, they suggested that PD-1 blockade plus chemotherapy and PD-1 blockade plus anlotinib might warrant further prospective evaluation as post-ICIs treatment strategies in selected patients with advanced NSCLC. Supplementary Information Supplementary Material 1. Acknowledgements The authors thank all patients and their families for their contribution to this study, as well as the clinical staff of the First People’s Hospital of Lian Yun Gang for their support in data collection and follow-up. Abbreviations Control group Confidence interval Complete response Common Terminology Criteria for Adverse Events Disease control rate Duration of response Eastern Cooperative Oncology Group Hazard ratio Immune checkpoint inhibitors Immune-related adverse events Non-small cell lung cancer Observation group Objective response rate Overall survival Progressive disease Progression-free survival Partial response Stable disease Treatment-related adverse events Vascular endothelial growth factor Authors' contributions Ting-Ting Huang conceived and designed the study, Xin-Ying Wang and Qin Li collected and analyzed data, Ting-Ting Huang performed the statistical analysis, Ting-Ting Huang and Peng-Fei Song drafted the manuscript, Peng-Fei Song polished the language, all authors reviewed and approved the final version of the manuscript. Funding This work was supported by the Jiangsu Provincial Elderly Health Research Project (No. LKM2024043). Data availability The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the ethics committee of the First People’s Hospital of Lian Yun Gang (approved number: LW-20260302001-01). The requirement for informed consent was waived because of the retrospective nature of the study. All procedures were performed in accordance with the Declaration of Helsinki. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.