Background

Immune checkpoint inhibitors (ICIs), principally represented by monoclonal antibodies targeting programmed death-1 (PD-1) or programmed death-ligand 1 (PD-L1), have transformed the therapeutic paradigm and become the pillar regimens for advanced or metastatic non-small cell lung cancer (NSCLC) [1, 2]. Despite the relative success of immunotherapy over chemotherapy witnessed in the past decade, PD-(L)1 blockade was far from perfect to induce durable immune responses for every patient from the total unselected NSCLC population, even within the PD-L1-positive stratum, calling for an exigent need of reliable immunotherapeutic predictors to promote precision medicine.

Effectiveness of ICI therapy has been associated with the biological peculiarity of tumor microenvironment differed by anatomical locations [3]. By a logical extension of this notion, it is envisioned that the heterogeneity of immunotherapeutic efficacy in the metastatic setting might depend on the interplay of the overall antitumor immunity with the local metastatic-organ microenvironment [4]. Preliminary evidences have demonstrated the negative correlation between survival prospects and distant metastases (e.g., liver metastasis) in NSCLC treated with ICIs [5, 6]. Nevertheless, these studies were conducted in a single-arm setting of ICIs, which lacked a controlled arm of conventional treatment for comparison, resulting in an ignorance of the predictive versus prognostic effect [7]. Besides, the therapeutic impact of each metastatic organ was previously investigated in isolation, but comprehensive analysis that looks at the big picture and incorporates the entire metastatic landscape is still lacking; as tumors could spread to several organs concurrently for patients with advanced stage cancers, outcome studies that focused on single organ metastasis presented limited clinical significance.

On this ground, we assumed that only by integrative delineation of the whole metastatic-organ landscape from both perspectives of prognostic and predictive effects could we systematically assess the impact of organ metastases on ICI therapy. For this purpose, we discriminated the predictive versus prognostic effect of each metastatic organ within a double-arm randomized trial cohort, and further proposed a metastasis-based scoring system (METscore), which incorporated both prognostic and predictive effects of the metastatic-organ landscape, to forecast survival outcomes of advanced NSCLC patients treated with ICI therapy in clinical-trial and real-word cohorts.

Methods

Population

We conducted a retrospective study of individual-participant data from three clinical trials (OAK: NCT02008227, N = 850; FIR: NCT01846416, N = 136; BIRCH: NCT02031458, N = 667) [8,9,10] and a real-word cohort from Nanfang Hospital, Southern Medical University (NFyy: N = 409), giving a total sample size of 2062 advanced-stage NSCLC patients. OAK was a randomized phase III trial comparing atezolizumab versus docetaxel in previously treated NSCLC patients [11, 12]. FIR and BIRCH were both phase II atezolizumab trials in chemotherapy-naïve or previously treated PD-L1-selected NSCLC patients [13, 14]. Detailed information about these three clinical trials has been described in the corresponding studies [11, 13, 14]. Deidentified individual-participant data of the clinical trials were accessed according to Roche’s policy and process for clinical study data sharing. The NFyy cohort included patients diagnosed with metastatic NSCLC who were treated with at least one dose of ICIs as first-line or subsequent regimen, which was collected after the approval of the Institutional Ethical Review Boards of Nanfang Hospital. The clinical characteristics of the NFyy cohort are summarized in Additional file 1: Table S1.

Baseline PD-L1 status of the three trial cohorts was prospectively tested using the SP142 antibody, scored as a percentage of tumor cells (TC0: < 1%; TC1: ≥ 1% but < 5%; TC2: ≥ 5% but < 50%; TC3: ≥ 50%) or tumor-infiltrating immune cells (IC0: < 1%; IC1: ≥ 1% but < 5%; IC2: ≥ 5% but < 10%; IC3: ≥ 10%) with PD-L1 staining [11, 13, 14]. PD-L1 expression of the real-world NFyy cohort was tested before treatment using the 22C3 antibody and calculated as a tumor proportion score (TPS). Patients with PD-L1 TC/IC or TPS of at least 1% were considered positive for the respective immunohistochemistry assays. Among these cohorts, 465 patients in OAK and 367 in NFyy were tested as PD-L1 positive, while patients in PD-L1-selected FIR and BIRCH trials were all confirmed positive at enrollment.

Study design

The workflow of the study was illustrated in Fig. 1. We first took advantage of the OAK trial as it endorsed two arms of patients (immunotherapy versus chemotherapy) so that we were able to delineate the predictive versus prognostic effect of distant metastases. Then, a metastasis-based scoring system, termed METscore, was developed to incorporate both the immunotherapy-specific predictive effect and the treatment-independent prognostic effect of the whole metastatic-organ landscape, thereby forecasting clinical outcomes of ICI therapy. The generalization performance of METscore was tested in single-arm FIR and BIRCH trials and the real-world NFyy cohort. Moreover, the discrimination ability was validated in the first-line ICI setting. Overall survival (OS) was defined as the primary endpoint, and progression-free survival (PFS) was the co-primary endpoint as an addition to OS for first-line cohorts.

Fig. 1
figure 1

Flowchart showing the design of the study. The double-arm OAK trial (atezolizumab versus docetaxel) was used as a discovery cohort to identify the prognostic and predictive effects of metastatic organs. A metastasis-based scoring system, termed METscore, was developed to forecast survival prospects of ICI therapy in the PD-L1-positvive population, and was further validated in PD-L1-selected atezolizumab trials (FIR and BIRCH) and a real-word ICI cohort. Abbreviations: NSCLC, non-small cell lung cancer; ICI, immune checkpoint inhibitor; PD-L1, programmed death-ligand 1; TC, tumor cell, IC: immune cell

Full size image

Prognostic and predictive effects of metastatic-organ landscape

Pretreatment statuses of common distant metastatic organs of NSCLC were included for analysis; those with incidence of < 5% in OAK were not examined. Thus, the metastatic-organ landscape comprised adrenal glands, brain, liver, bone, pleura, pleural effusion, and mediastinum in the present study. These metastatic organs were divided into two categories based on their predictive value for ICI therapy. Accordingly, category I metastatic organs were defined as those with survival benefits of immunotherapy over chemotherapy, while others were defined as category II.

Multivariate Cox proportional hazard regression was performed to delineate the predictive versus prognostic effect of each metastatic organ via the inclusion of metastatic statuses of organs, treatment arm, and treatment-by-organ interaction terms using the randomized OAK data (Additional file 2: Table S2). The model coefficients were scaled from − 5 to 5 and rounded to the nearest integers to facilitate clinical use [15, 16]. The scaled points corresponding to the metastatic statuses of organs delineated the treatment-independent prognostic effects, while those corresponding to treatment-by-organ interactions delineated the immunotherapy-specific predictive effects (Additional file 3: Table S3). The METscore in forecasting outcomes of ICI therapy was then set up to incorporate the prognostic and predictive effects of each metastatic organ by calculating the total points.

Statistical analysis

Kaplan-Meier method and log-rank test were used for comparing survival outcomes among groups, with hazard ratio (HR) and 95% confidence interval (CI) reported. Survival analyses were conducted in GraphPad Prism (version 8.0.1) or R (version 3.6.1) with R packages “survival” and “survminer.” Multivariate Cox proportional hazard regression was implemented using R package “survival.” The optimum cut-point for METscore was determined by maximally selected rank statistics using the R package “maxstat” with a minimal proportion of 20% in each group by setting the parameter “minprop” as 0.2. All statistical tests were two sided, and P ≤ 0.05 was deemed statistically significant.

Results

Specific predictive effect of metastatic-organ landscape for ICI therapy

To delineate the immunotherapeutic predictive effect of organ metastases, we explored the impact of metastatic-organ landscape on survival outcomes in the atezolizumab arm and the docetaxel arm from OAK respectively. In terms of OS, survival prospects were significantly different among patients with various metastatic organs in the atezolizumab-treated population (P = 0.0105; Fig. 2A). Nevertheless, OS was generally similar in the docetaxel-treated population regardless of metastatic organs (P = 0.3742; Fig. 2B). Pairwise comparisons among metastatic organs in the atezolizumab arm showed that adrenal gland metastasis and brain metastasis yielded the best long-term survival benefits, with significantly longer OS compared with metastases to other organs, including liver, bone, and pleural effusion (the lower triangle in Fig. 2C). With regard to the docetaxel arm, however, no significant difference of OS existed between any pair of metastatic organs (the upper triangle in Fig. 2C).

Fig. 2
figure 2

Metastatic-organ landscape as a determinant of survival outcomes specific to immune checkpoint-based therapy. Kaplan-Meier curves comparing overall survival among various metastatic organs in A atezolizumab- and B docetaxel-treated populations. C Pairwise comparisons of overall survival in the atezolizumab- (lower triangle) and docetaxel-treated (upper triangle) populations. Kaplan-Meier curves comparing progression-free survival per metastatic organs in C atezolizumab- and D docetaxel-treated populations. E Pairwise comparisons of progression-free survival in the atezolizumab- (lower triangle) and docetaxel-treated (upper triangle) populations. Data in each cell of C and F represents hazard ratio (95% interval confidence) for the pairwise comparison of row-defining organ versus column-defining organ. Significant results are in bold red font. Abbreviations: Met, metastasis; PEFF, pleural effusion; Media, mediastinum

Full size image

Consistent findings were observed in terms of PFS, where significant survival difference was seen among atezolizumab-treated patients with various metastatic organs (P = 0.0167; Fig. 2D) but not in docetaxel-treated populations (P = 0.8242; Fig. 2E). In agreement with this, there only existed significant differences of PFS in pair-wise comparisons in the atezolizumab arm (the lower triangle in Fig. 2F) but not in the docetaxel arm (the upper triangle in Fig. 2F).

Metastatic-organ landscape as a predictor in a PD-L1 dependent manner

Upon comparing the efficacy following treatment with immunotherapy and chemotherapy, we found an identifiable association between the metastatic-organ landscape and the clinical benefits of atezolizumab versus docetaxel in the total population from OAK (Fig. 3A). On account of the clinical practice of immunotherapy in patients who were PD-L1 positive (≥ 1%), we investigated the predictive effect (immunotherapy versus chemotherapy) of the metastatic-organ landscape stratified by PD-L1 status. Intriguingly, the predictive effect was observed exclusively in the PD-L1-positive population (TC/IC ≥ 1%; Fig. 3B) rather than in the PD-L1-negative population (TC/IC < 1%; Fig. 3C).

Fig. 3
figure 3

Predictive significance of metastatic-organ landscape for efficacy with immunotherapy versus chemotherapy in a PD-L1-dependent fashion. Forest plots showing overall survival (OS) benefits of atezolizumab versus docetaxel per organ metastatic category in A the total, B PD-L1-positive (TC/IC ≥ 1%), and C PD-L1-negative (TC/IC < 1%) populations from OAK trial. Two categories of metastatic organs were identified based on OS benefits of atezolizumab versus docetaxel in the PD-L1-positive population, and metastases to category I organ (adrenal glands, brain, and liver) presented significantly better OS results in atezolizumab arm versus docetaxel arm, while metastases to category II organ (bone, pleura, pleural effusion, and mediastinum) derived no benefit from atezolizumab over docetaxel. Abbreviations: Atezo, atezolizumab; Doce, docetaxel; CI, confidence interval; PD-L1, programmed death-ligand 1; TC, tumor cell; IC, immune cell

Full size image

In general, the predictive significance varied across metastatic organs at different degrees or even conversely within the PD-L1-positive population, which defined two organ categories (Fig. 3). For category I organs, OS benefits of atezolizumab versus docetaxel were found in patients whose tumors metastasized to adrenal glands (HR 0.42, 95% CI 0.25–0.70), brain (HR 0.41, 95% CI 0.19–0.88), and liver (HR 0.57, 95% CI 0.36–0.89) (Fig. 3B). On the contrary, for bone, pleural, pleural effusion, and mediastinum, patients harboring metastasis to any of these category II organs did not benefit from atezolizumab relative to docetaxel (Fig. 3B). Synergistic predictive effect was found among the category I organ metastases (adrenal glands, brain, and liver) within the PD-L1-positive population (Additional file 4: Fig. S1A-F), where the OS benefits of atezolizumab versus docetaxel in patients with metastases to double category I organs (HR 0.32, 95% CI 0.14-0.76, P = 0.0013; Additional file 4: Fig. S1C) were even more pronounced in comparison to those with metastasis to a single category I organ (HR 0.59, 95% CI 0.41-0.85, P = 0.0048; Additional file 4: Fig. S1B).

We also examined the predictive versus prognostic effect of the metastatic statuses of category I organs (Additional file 5: Fig. S2A-D). Remarkably within the PD-L1-positvie population, the presence of the category I organ metastases was associated with decreased OS in the docetaxel arm (P < 0.0001; Additional file 5: Fig. S2B), indicative of the inherent unfavorable prognostic effect of metastases. In contrast, OS was not influenced by the metastatic number of the category I organs in the atezolizumab arm (P = 0.5978; Additional file 5: Fig. S2A), suggesting that the unfavorable prognostic effect was alleviated by the favorable predictive effect specific to immunotherapy. Conversely, the inherent unfavorable prognostic effect of metastases became discernible in PD-L1-negative patients treated with atezolizumab, since the immunotherapeutic predictive effect was absent in this population; consequently, we observed a negative correlation between OS and the number of the category I organ metastases (P = 0.0005; Additional file 5: Fig. S2C).

Incorporating metastatic-organ landscape to forecast ICI therapy

The synergistic value of different metastatic organs inspired us to incorporate the whole landscape for comprehensive assessment. Having identified metastatic-organ landscape as a determinant of both inherent prognosis and immunotherapeutic benefit, we believed that only through incorporating both the treatment-independent prognostic effect and the immunotherapy-specific predictive effect of the whole metastatic-organ landscape could we truly reflect its impact on ICI therapy.

To this end, we proposed a metastasis-based scoring system (METscore) to forecast survival outcomes of advanced-stage NSCLC patients treated with ICI agents using the PD-L1-positive OAK cohort, since the therapeutic impact was particularly witnessed in this stratum (Fig. 4A). On this basis, given a metastatic profile of a certain patient obtained from pretreatment radiologic assessments, two scaled points could be assigned to each metastatic organ, one of which delineated the prognostic effect and the other delineated the predictive effect. The total points of prognostic and predictive effects of the whole metastatic-organ landscape constituted the METscore in forecasting the survival benefits following ICI therapy.

Fig. 4
figure 4

A metastasis-based scoring system (METscore) incorporating prognostic and predictive effects for ICI prediction in the PD-L1-positive population. A Nomogram illustrating scaled points as proxy for prognostic and predictive effects of metastatic organs, and the total points as METscore for the prediction of overall survival of PD-L1-positvive patients following ICI therapy. B Putative cut-points (0 ~ 5) of METscore to stratify PD-L1-positive patients into -High and -Low groups with significantly different OS in the atezolizumab-treated population from OAK. Abbreviations: Met, metastasis; Non, non-metastasis; CI, confidence interval; OS, overall survival; ICI, immune checkpoint inhibitor; PD-L1, programmed cell death-ligand 1

Full size image

A range of putative cut-points (0 ~ 5) for METscore was assessed in the atezolizumab population (Fig. 4B). There was a general pattern of prolonged OS in patients with lower METscore relative to those with higher METscore, and an enlarged OS difference could be expected if more stringent cut-points (4 and 5) were selected (Fig. 4B). For practical application, the optimal cut-point of METscore, corresponding to 3 achieved by the maximally selected rank statistics, was determined by weighing the survival benefits against the minimal proportion of each group. And patients were classified into METscore-High (METscore ≥ 3) and -Low (METscore < 3) categories accordingly. Thereupon, the METscore system and the threshold were eventually locked for performance evaluation throughout the study. The system has been translated into a web-based tool that is freely available to the public (Additional file 6: Fig. S3).

METscore enables identification of therapeutic benefit from checkpoint blockade

We first evaluated the discrimination performance of METscore in the PD-L1-positive OAK cohort. Patients with METscore-Low obtained significantly longer OS than METscore-High counterparts in the atezolizumab-treated population (HR 0.48, 95% CI 0.32–0.72, P < 0.0001; Fig. 5A). By contrast, there was no significant difference in terms of OS between METscore-Low and -High groups in the docetaxel-treated population (HR 0.73, 95% CI 0.49–1.07, P = 0.0805; Fig. 5B). Concurrently, upon direct comparison of survival prospects following treatment with immunotherapy and chemotherapy, OS was demonstrated to favor atezolizumab as compared to docetaxel within the METscore-Low group (HR 0.64, 95% CI 0.49–0.84, P = 0.0011; Additional file 7: Fig. S4A), whereas it was generally similar between the two arms within the METscore-High group (HR 0.98, 95% CI 0.64–1.48, P = 0.9121; Additional file 7: Fig. S4B). These results indicated that METscore enabled noninvasive identification of beneficiaries of ICI therapy, i.e., the METscore-Low patients.

Fig. 5
figure 5

METscore forecasts survival outcomes of checkpoint blockade-based immunotherapy in the PD-L1-positive population. Kaplan-Meier curves showing overall survival according to METscore groups (≥ 3 vs. < 3) in A atezolizumab- and B docetaxel-treated populations from PD-L1-positive OAK cohort. Kaplan-Meier curves showing overall survival according to METscore groups in PD-L1-selected atezolizumab trials, C FIR and D BIRCH, and E NFyy cohort. Abbreviations: HR, hazard ratio; CI, confidence interval; PD-L1, programmed death-ligand 1

Full size image

Generalization of METscore in external clinical-trial and real-world cohorts

The METscore-based system was then externally validated in PD-L1-selected clinical-trial and real-world cohorts. In FIR, the use of METscore allowed stratification of patients into METscore-High and -Low groups with significant difference of OS (HR 0.57, 95% CI 0.33–0.97, P = 0.0350; Fig. 5C). In BIRCH, OS rates consistently favored METscore-Low over METscore-High (HR 0.53, 95% CI 0.40–0.71, P < 0.0001; Fig. 5). On top of that, the generalization performance of METscore was confirmed in NFyy, where patients with METscore-Low had prolonged OS than those with METscore-High (HR 0.60, 95% CI 0.39–0.92, P = 0.0181; Fig. 5E).

Considering that ICI has become the first-line treatment of advanced NSCLC [17], we evaluated the discrimination performance of METscore specifically in the first-line setting. It is noteworthy that survival prospects were significantly longer for METscore-Low patients referenced to METscore-High patients in the first-line FIR/BIRCH cohort in terms of both OS (HR 0.44, 95% CI 0.23–0.83, P = 0.0087; Fig. 6A) and PFS (HR 0.56, 95% CI 0.32–0.95, P = 0.0290; Fig. 6B). The significant survival advantage of METscore-Low over METscore-High groups were replicated in the first-line NFyy cohort according to OS (HR 0.51, 95% CI 0.29–0.90, P = 0.0182; Fig. 6C) and PFS (HR 0.45, 95% CI 0.25–0.80, P = 0.0045; Fig. 6D).

Fig. 6
figure 6

METscore enables classification of survival benefits from first-line checkpoint blockade-based immunotherapy. Kaplan-Meier curves of A overall survival and B progression-free survival according to METscore groups (≥ 3 vs. < 3) for patients receiving atezolizumab as first-line therapy within PD-L1-selected FIR and BIRCH trials. Kaplan-Meier curves of C overall survival and D progression-free survival according to METscore groups for patients receiving ICIs as first-line therapy within the PD-L1-selected NFyy cohort. Abbreviations: ICI, immune checkpoint inhibitor; HR, hazard ratio; CI, confidence interval; PD-L1, programmed death-ligand 1

Full size image

Discussion

Understanding organ-specific tumor immune context has been one of the top ten challenges in cancer immunotherapy [4]. However, even in an era of precision medicine where regimens are increasingly informed by multi-omic data, the role of metastatic organs in affecting the efficacy of immunotherapy remains unclear. In this study, we delineated and incorporated both predictive and prognostic effects of metastatic-organ landscape to forecast survival outcomes of advanced NSCLC patients following ICI therapy, particularly in the PD-L1-positive stratum.

While cancers originating from various organs presented different susceptibility to ICIs, immunotherapeutic efficacy could even vary greatly among advanced patients with the same cancer type based on anatomically differences of metastatic tumor lesions [4]. Until recently, however, the metastatic-organ landscape has not yet been incorporated into the current clinical practice of immunotherapy in the metastatic setting, partially due to the paucity of evidence; this provided a direction for researchers to dissect therapeutic heterogeneity and to accelerate clinical progress from the dimension of metastatic organ-specific antitumor immunity. With years of concerted efforts, association between metastatic organs and ICI efficacy continued to be uncovered [5, 6, 18, 19]. Nonetheless, owing to the single-arm nature of most previous studies without comparison treatment arms, whether the given metastatic organ was purely a treatment-independent prognostic factor or a immunotherapy-specific predictive factor remains controversial [7]; consequently, confusions would be generated when making claims of the predictive effect under the circumstance of single-arm ICI cohort.

In principle, a predictive marker should be claimed preferably in a randomized cohort [7]. Taking advantage of the double-arm randomized OAK trial of immunotherapy versus traditional chemotherapy, we were able to make the distinction between predictive and prognostic effects. Herein, we dissected the predictive versus prognostic effect of each metastatic organ. In particular, the metastatic-organ landscape exerted immunotherapeutic impact predominantly on the PD-L1-positive (TC/IC ≥ 1%) population. This makes sense as PD-L1 ≥ 1% was exactly the indication for first-line immunotherapy of advanced NSCLC on the basis of the National Comprehensive Cancer Network (NCCN) guidelines [17]. Within the PD-L1-positive stratum, metastatic organs were furtherly divided into two categories according to their survival benefits of immunotherapy over chemotherapy, and the predictive and prognostic effects were delineated utilizing a multivariate Cox regression with the inclusion of treatment-by-organ interaction terms.

Of note, the presence of metastasis to category I organs, including adrenal glands, brain, and liver, was identified as a potential predictive marker of sensitivity to immunotherapy, in spite of the coexisting adverse prognostic effect inherently coupled with this malignant biology behavior of tumors (i.e., metastasis). The paradoxical coexistence of the favorable predictive effect and the unfavorable prognostic effect of category I organ metastases (Fig. 4A) helped explain the non-significant association of survival prospects with the number of category I organ metastases in the immunotherapy arm (Additional file 5: Fig. S2A), which was congruent with accumulating evidence supporting that adrenal gland or brain metastasis did not influence patients’ survival in the context of PD-(L)1 blockade of NSCLC [19, 20]. This was reminiscent of our recent study reporting an upregulated blood-based tumor mutation burden and/or PD-L1 expression for theses metastases [20], and another study reporting a favorable response rate of adrenal gland metastatic lesions to ICIs [18]. The “Janus-faced” significance of category I organ metastases has also been recapitulated in liver metastasis, where superior efficacy of immunotherapy over chemotherapy was coupled with a worse prognosis [21,22,23]. Interestingly, the presence of liver metastases in PD-L1-negative patients predicts lack of response to immunotherapy (Fig. 3C). Results from two IMvigor211 (NCT02302807) [24, 25] and IMvigor210 (NCT02108652) [26,27,28] support the finding in a cross-cancer manner, implying that these patients may avoid immunotherapy (Additional file 8: Fig. S5). All these convergence of evidences suggested PD-(L)1 blockade shaped antitumor immunity in a metastatic-organ specific manner.

For first-line therapy of advanced NSCLC without actionable molecular biomarkers, patients tested as PD-L1 positive are recommended to receive immunotherapy according to the NCCN guidelines. Despite that, even in this PD-L1-selected population, the response rate to immunotherapy is mostly less than 50% in first-line trials [29], which prompted us to identify more accurate beneficiaries from immunotherapy through the integration of the organ-specific metastatic landscape. Accordingly, a metastasis-based scoring system was developed to classify PD-L1-positive patients into METscore-High and -Low categories with differential survival benefits derived from ICI agents, which was later on validated in real-life and first-line cohorts. Our results indicate that the METscore system might provide further guidance for therapeutic strategies of advanced NSCLC patients in the context of PD-L1 positive. In view of the efficacy benefits over chemotherapy in METscore-Low patients, immunotherapy might be a preferable alternative for these patients. As for METscore-High patients that account for a relatively small proportion, it is rational to treat them with conventional therapy due to a lack of response to immunotherapy, which could also avoid the risk of developing immune-related toxicity and hyperprogression.

As survival gains have been witnessed in the addition of bevacizumab to immunochemotherapy among first-line populations, including those difficult-to-treated patients harboring liver metastases, this combination strategy is approved for the treatment of metastatic nonsquamous NSCLC without EGFR/ALK genetic alterations [1, 30, 31]. To evaluate the discrimination performance of METscore in this regimen, we applied the system to patients receiving ICIs plus bevacizumab and chemotherapy in the NFyy cohort. We found that OS trended longer in the METscore-Low group compared to its counterpart; concordantly, both OS and PFS rates favored METscore-Low over METscore-High among patients in the first-line setting, albeit not reaching a conventional significant level (Additional file 9: Fig. S6). These findings indicate the METscore system might also play its role in patients receiving bevacizumab in combination with immunochemotherapy. Nonetheless, the findings should be interpreted with caution due to the small sample size; a more holistic assessment in cohorts with immunotherapy-based combination regimens is needed before its reliable utility in the complicated clinical practice.

Moreover, we noted that the effect of the category I metastatic organs in the PD-L1-negative population was the opposite of that in the PD-L1 positive population (Additional file 5: Fig. S2). OS was generally similar regardless of the category I metastases in PD-L1-negative patients treated with docetaxel (P = 0.5766; Additional file 5: Fig. S2D), which is in line with the results illustrated in Fig. 3C. In the PD-L1-negative stratum, patients without metastases to adrenal glands, brain, or liver derived significant clinical benefits from atezolizumab over docetaxel, whereas the superiority was removed or even reversed in patients with these metastases, particularly liver metastasis, resulting in an absence of significant difference. We also evaluated METscore in the PD-L1-negative stratum in OAK without external validation and observed moderate discrimination performance for risk stratification in both arms (Additional file 10: Fig. S7). Since METscore is the sum of scaled points from both predictive effect and prognostic effect (Methods, Fig. 4A), we reasoned that it is the prognostic component of METscore that contribute to its moderate ability in the PD-L1-negative population where the immunotherapeutic predictive component does not work. Therefore, its performance in patients tested as PD-L1 negative or treated with non-ICI therapies cannot be guaranteed.

Limitations existed in this study. Firstly, lung metastasis was not taken into account, since it was not identified as ipsilateral or contralateral in the trial data while our study focused on distant metastases. Secondly, METscore has its own scope of application, namely in the PD-L1-positive population treated with immunotherapy. Additionally, the study was based on post hoc analyses; despite the METscore model was independently validated using clinical-trial and real-word cohort as well as the first-line subsets, prospective validation of the METscore model is still needed in a clinical trial setting to corroborate these findings.

Conclusions

In conclusion, the study delineated the immunotherapeutic predictive versus prognostic effect of metastatic-organ landscape for advanced NSCLC patients in a PD-L1-dependent manner, which helped to improve the understanding of organ-specific antitumor immunity in cancer immunotherapy. Using baseline radiologic assessments, our metastasis-based scoring system incorporating both predictive and prognostic effects of metastatic-organ landscape enabled non-invasive forecast of survival outcomes in PD-L1-positive NSCLC patients following ICI therapy.