Purpose:

Immune-checkpoint inhibitors (ICI) have improved the survival of patients with non–small cell lung cancer (NSCLC). However, only a subset of patients benefit from ICIs, and the role of PD-L1 as predictive biomarker is still debated. A plasma immune-related miRNA-signature classifier (MSC) was established in lung cancer screening settings to identify the lethal form of the disease in early stages. In this exploratory study, we tested the efficacy of the MSC as prognostic marker in patients with advanced NSCLC treated with ICIs.

Experimental Design:

The MSC risk level was prospectively assessed in a consecutive series of 140 patients with NSCLC before starting treatment with ICIs. Overall response rate (ORR), progression-free survival (PFS), and overall survival (OS) in strata of PD-L1 and MSC alone or combined were considered as endpoints. Multiple plasma samples to monitor MSC risk level during treatment were also profiled.

Results:

Adequate tissue and plasma samples were available from 111 (79%) and 104 (75%) patients with NSCLC, respectively. MSC risk level was associated with ORR (P = 0.0009), PFS [multivariate HR = 0.31; 95% confidence interval (CI), 0.17–0.56; P = 0.0001], and OS (multivariate HR = 0.33; 95% CI, 0.18–0.59; P = 0.0002). The combination of MSC and PD-L1 stratified patients into three risk groups having 39%–18%–0% 1-year PFS (P < 0.0001) and 88%–44%–0% 1-year OS (P < 0.0001), according to the presence of 2–1–0 favorable markers. During treatment, MSC risk level decreased or remained low until tumor progression in patients with responsive or stable disease.

Conclusions:

The plasma MSC test could supplement PD-L1 tumor expression to identify a subgroup of patients with advanced lung cancer with worse ORR, PFS, and OS in immunotherapy regimens.

Translational Relevance

Non–small cell lung cancer (NSCLC) is among the most responsive tumors to PD-1/L1 axis blockade therapy. Nonetheless, the true clinical benefit remains limited. Although PD-L1 expression by cancer cells and high tumor mutational burden (TMB) can be predictive of a favorable response, many patients who are negative for PD-L1 or have low TMB actually respond. To identify biomarkers to improve patient selection, we focused on circulating miRNAs. These small molecules released into the bloodstream by both nonimmune and immune cells play important roles in immune regulation. Here, we prospectively tested the efficacy of the already established plasma miRNA signature classifier (MSC), either alone or combined with PD-L1, as a prognostic biomarker in patients with advanced NSCLC treated with immune-checkpoint inhibitors. The results showed that no patient with MSC high-risk level responded to immunotherapy. Moreover, the combination of MSC and PD-L1 enabled the identification of a subgroup of patients reaching median progression-free and overall survival in 2 and 5 months, respectively.

Despite improvements in early diagnosis and new therapeutic strategies, the overall survival (OS) of patients with lung cancer remains very low (1). The influence of the micro- and macro-environment on cancer development is currently gaining increasing attention (2). Epithelial cancers, such as lung cancer, are now considered not simply the result of the abnormal growth of cancer cells but rather of complex interactions between cancer cells and stromal components (2, 3).

In this context, immune-checkpoint inhibitors (ICI) targeting CTLA4 and the PD-1/L1 axis responsible for tumor immune evasion have drastically increased both progression-free survival (PFS) and OS in patients with non–small cell lung cancer (NSCLC; refs. 4–6). However, only a subset of patients responds to ICIs, and PD-L1 expression exhibits limited efficacy as a predictive biomarker (7). More recently, a high tumor mutational burden (TMB), as a surrogate for immunogenic tumor neoantigen presentation (8), was associated with a favorable response to ICIs in patients with NSCLC irrespective of PD-L1 expression in both tissue and plasma samples (9–11).

To identify biomarkers for early lung cancer detection, we have trained and largely validated a circulating miRNA signature classifier (MSC) that can discriminate lung cancer with more aggressive features in patients with lung cancer enrolled in low-dose CT screening trials (12, 13). The MSC is composed of 24 miRNAs whose reciprocal ratios stratify subjects into three risk levels, with the MSC high-risk patients having worse prognoses than MSC intermediate- or low-risk patients (14). We have also shown that the MSC risk level was independent of tumor histology and mutational burden (15).

Because cells release miRNAs packaged in extracellular vesicles, these small circulating molecules may plausibly contribute to cell-to-cell communication by eliciting different functions according to the recipient cell type (16, 17). Indeed, alterations in miRNA levels might lead to protumorigenic phenotypic changes in stromal cells such as fibroblasts (18), macrophages (19), myeloid cells (20), dendritic cells (21), and T cells (22). Specifically for miRNAs composing the MSC, their modulation in plasma was consistent with an immunosuppressive conversion of immune cells, including neutrophils and macrophages (23). Circulating miRNAs can thus be considered robust and reliable biomarkers reflecting changes in tumor–host interaction (24).

The development of a minimally invasive blood-based tool able to supplement current biomarkers to identify patients who could benefit or not from ICI treatment could be crucial to improve the efficacy of these drugs. Here, we report the results of a prospective exploratory study to evaluate the association between plasma MSC risk level and response to ICIs in terms of overall response rate (ORR), PFS, and OS in patients with advanced NSCLC.

Study design and patient characteristics

A consecutive series of 140 patients with NSCLC was administered with ICIs from July 2015 to May 2018 as a first-line (n = 33), maintenance (n = 5), second-line (n = 71), or further line of treatment (n = 31). In detail, 126 patients were treated with anti-PD-1 (95 nivolumab and 31 pembrolizumab), 12 with anti-PD-L1 (3 avelumab, 3 atezolizumab, and 6 durvalumab), and 2 with combined anti-PD-L1 and anti-CTLA4 (durvalumab + tremelimumab). Presence of liver metastases was observed in 21 (15%) patients. Plasma and tissue samples were all collected prior to starting immunotherapy and on treatment in a subset of patients. Samples were prospectively analyzed to establish the plasma MSC risk level and tumor PD-L1 expression.

All patients were periodically evaluated until September 2018, and the response to treatment was measured by radiological evaluation every 8 weeks after treatment initiation. According to the RECIST 1.1 best response criteria, patients were classified as responders (R), patients with stable disease (SD), and with disease progression (P). The study complied with the Declaration of Helsinki. All experimental protocols were approved by the Internal Review and Ethics Boards of the Istituto Nazionale Tumori of Milan (Milan, Italy) and all patients provided informed consent.

miRNA profiling of plasma samples

Whole blood was collected in 10 mL K2EDTA Vacutainer tubes and the plasma was separated by two centrifugation steps at 1,258 × g and 4°C for 10 minutes. Using a mirVana PARIS Kit (Thermo Fisher Scientific) or Maxwell RSC miRNA Tissue Kit (Promega), total RNA was extracted from 200 μL plasma samples and eluted in 50 μL of buffer. miRNA expression was determined by qRT-PCR using a 384-well microfluidic Custom Taq Array MicroRNA Card (Thermo Fisher Scientific) containing probes for the 24 miRNAs of interest, which were spotted in duplicate according to the protocol: miR-101-3p, miR-106a-5p, miR-126-5p, miR-133a, miR-140-3p, miR-140-5p, miR-142-3p, miR-145-5p, miR-148a-3p, miR-15b-5p, miR-16-5p, miR-17-5p, miR-197-3p, miR-19b-3p, miR-21-5p, miR-221-3p, miR-28-3p, miR-30b-5p, miR-30c-5p, miR-320a, miR-451a, miR-486-5p, miR-660-5p, and miR-92a-3p (MiRBase ID - v21). The Ct mean values to set an automatic baseline and a fixed threshold of 0.15 were then extrapolated using ViiA7 RUO Software (Thermo Fisher Scientific).

MSC algorithm

The MSC test was performed following previously reported standard operating procedures with fixed parameters for prospective studies (25). Briefly, the effect of hemolysis on the MSC test was evaluated by both spectrophotometric and molecular analyses. In each plasma sample, the level of free hemoglobin was initially measured by the 414 nm/375 nm absorbance ratio (26). After miRNA profiling, 16 ratios between four hemolysis-related (miR-16-5p, miR-451a, miR-486-5p, and miR-92a-3p) and four unrelated miRNAs (miR-126-5p, miR-15b-5p, miR-221-3p, and miR-30b-5p) were then determined. Samples with a 414 nm/375 nm absorbance ratio higher than 1.4 and in which at least 50% of the miRNA ratios exceeded the respective cut-off values were considered hemolyzed and excluded from further analysis. For all samples passing quality control, the four signatures of risk of disease (RD), presence of disease (PD), risk of aggressive disease (RAD), and presence of aggressive disease (PAD) were determined (25). The respective MSC risk level was attributed to each sample as follows: low risk for RDneg ∩ PDneg ∩ RADneg ∩ PADneg; intermediate risk for RDpos ∪ PDpos ∩ RADneg ∩ PADneg; or high risk for RADpos ∪ PADpos (13).

PD-L1 IHC analysis

According to the kit instruction, 2.5/3-μm–thick sections were cut from paraffin blocks, dried, dewaxed, rehydrated, and unmasked with Dako PT-link EnVision FLEX Target Retrieval Solution (low pH, 30 minutes, 98°C). PD-L1 mAB 22C3 (Dako) was incubated with the EnVision FLEX+ Detection Kit (Dako) in the Autostainer System (Dako). PD-L1 expression was measured as the percentage of positive neoplastic cells (PNC) and patients stratified in three groups: PNC < 1%, PNC 1%–49%, and PNC ≥ 50%. In a group of 43 patients with available material, PD-L1 expression was confirmed by using the mAB SP263 (Ventana Medical Systems) on an automated Benchmark Ultra Platform (Ventana Medical Systems) according to the manufacturer's protocol.

Statistical analyses

The continuous variables were given as the mean values ± SD and the categorical variables as numbers and percentages. Cohen kappa statistic was used to analyze the interrater agreement of the categorical variables.

The ORR was estimated as the percentage of R among all patients in respective class. The relative risk of response (RR) and corresponding 95% confidence interval (CI) were calculated by 2 × 2 contingency table and Fisher exact test was adopted to calculate P values (27). A 0.5 value was added to all cells to avoid computational problems caused by 0 value (28).

For PFS and OS endpoints, the time-to-event occurrence was computed from the start date of treatment to the date when the event was recorded or was censored at the date of the last follow-up assessment in event-free patients. One-year survival curves were estimated using the Kaplan–Meier method and were compared by the log-rank test. Patients who discontinued therapy after one cycle due to adverse effects or clinical deterioration were not considered for PFS analysis.

The crude and adjusted HRs and the corresponding 95% CIs were estimated using Cox proportional hazard models. Multivariate analyses were performed adjusting models for age, gender, smoking status, and presence of liver metastasis. Because treatment with ICIs as first-line therapy was approved only for advanced NSCLC with PD-L1 expression on at least 50% of tumor cells (29), line of therapy was considered as covariate only in multivariate analysis in strata of MSC risk level.

All tests were two-sided, and a P value of less than 0.05 was considered statistically significant. Statistical analyses were performed using SAS 9.4 (SAS Institute), and the Kaplan–Meier curves were obtained using GraphPad Prism version 5.02 statistical software.

Molecular and clinicopathologic findings of the prospective series

A consecutive series of 140 advanced (all stages III–IV) NSCLC cases treated with ICIs was prospectively analyzed for the plasma MSC test and monitored for up to 3 years. The median PFS and OS for the whole series were 3.0 and 8.1 months, respectively (Supplementary Fig. S1).

As reported in Table 1, female patients accounted for 34% of the series, and the overall average age was 66.2 years. The main histology type was adenocarcinoma, accounting for 65% of tumors. Twenty patients (14%) were never smokers, 78 (56%) former smokers, and 42 (30%) current smokers. The MSC was evaluable in 111 (79%) patients: 26 (19%) were MSC high risk, 51 (36%) were MSC intermediate, and 34 (24%) were MSC low risk. A total of 29 (20%) samples were not analyzable due to high hemolysis levels (25). Adequate tumor tissue samples for PD-L1 expression by 22C3 IHC assay were available for 104 (74%) patients: PNCs were <1% in 30 (21%) patients, between 1% and 49% in 36 (26%) patients, and ≥50% in 38 (27%) patients. For the 43 patients analyzed with both clones 22C3 and SP263, categorical data were consistent and included 17 with PNCs < 1%, 21 with PNCs 1%–49%, and 5 with PNCs ≥ 50%. Response to treatment was assessed according to RECIST 1.1 criteria: 26 (19%) were classified as R, 33 (24%) as SD, 67 (48%) as P, and 14 (10%) discontinued therapy after one cycle due to adverse effects (N = 5) or clinical deterioration (N = 9).

Table 1.

Clinicopathologic and molecular characteristics of 140 patients with advanced lung cancer treated with immune-checkpoint inhibitors

140 Patients with lung cancer treated with checkpoint inhibitors, N (%)
Gender  
 Female 48 (34%) 
Age (years) 66.2 ± 9.9 
Smoking habit  
 Never 20 (14%) 
 Former 78 (56%) 
 Current 42 (30%) 
Histology  
 Adenocarcinoma 91 (65%) 
 Squamous carcinoma 34 (24%) 
 Others 15 (11%) 
Stage  
 III 30 (21%) 
 IV 110 (79%) 
Plasma MSC risk level  
 High 26 (19%) 
 Intermediate 51 (36%) 
 Low 34 (24%) 
 Not evaluable 29 (21%) 
PD-L1 expression  
 PNC < 1% 30 (21%) 
 PNC 1%–49% 36 (26%) 
 PNC ≥ 50% 38 (27%) 
 Not evaluable 36 (26%) 
Overall response rate to ICIs  
 Responsive 26 (19%) 
 Stable disease 33 (24%) 
 Disease progression 67 (48%) 
 Not evaluated 14 (10%) 
140 Patients with lung cancer treated with checkpoint inhibitors, N (%)
Gender  
 Female 48 (34%) 
Age (years) 66.2 ± 9.9 
Smoking habit  
 Never 20 (14%) 
 Former 78 (56%) 
 Current 42 (30%) 
Histology  
 Adenocarcinoma 91 (65%) 
 Squamous carcinoma 34 (24%) 
 Others 15 (11%) 
Stage  
 III 30 (21%) 
 IV 110 (79%) 
Plasma MSC risk level  
 High 26 (19%) 
 Intermediate 51 (36%) 
 Low 34 (24%) 
 Not evaluable 29 (21%) 
PD-L1 expression  
 PNC < 1% 30 (21%) 
 PNC 1%–49% 36 (26%) 
 PNC ≥ 50% 38 (27%) 
 Not evaluable 36 (26%) 
Overall response rate to ICIs  
 Responsive 26 (19%) 
 Stable disease 33 (24%) 
 Disease progression 67 (48%) 
 Not evaluated 14 (10%) 

NOTE: For continuous variables, average and respective SDs are reported.

Overall, 131 (94%) patients had at least one plasma or tissue sample suitable for marker evaluation, while both samples resulted adequate for 84 (60%) patients (Fig. 1). No interrater agreement (κ = −0.07) between MSC and PD-L1 was observed (Supplementary Table S1).

Figure 1.

CONSORT diagram of the prospective Apollo study evaluating the plasma MSC and PD-L1 expression on tumor cells in 140 patients with advanced lung cancer treated with ICIs.

Figure 1.

CONSORT diagram of the prospective Apollo study evaluating the plasma MSC and PD-L1 expression on tumor cells in 140 patients with advanced lung cancer treated with ICIs.

Close modal

Plasma MSC risk level and PD-L1 tumor expression are associated with ORR

ORR in the whole series was 19% (26/140). No one of the 25 patients with NSCLC with adequate plasma sample classified as R were MSC high-risk level at the baseline. Indeed, among the 26 MSC high-risk patients 5 (19%) were SD, 16 (62%) P, and 5 (19%) discontinued ICI treatment after one cycle due to adverse effects (N = 2) or clinical deterioration (N = 3). Moreover, an opposite trend between R (0%–22%–38%) and P (62%–40%–35%) was observed among patients with MSC high–intermediate–low risk level. Overall, a significant reduction in ORR (RR = 0.07; 95% CI, 0.00–1.05; P = 0.0009) was observed by comparing MSC high- versus intermediate- and low-risk patients with advanced NSCLC (Table 2). Considering PD-L1 tumor expression, a lower ORR was observed in patients with PNC < 50% (RR = 0.41; 95% CI, 0.20–0.83; P = 0.0158).

Table 2.

Results on ORR and corresponding RR

All patientsPatients with both markers available
NORRRR95% CIPNORRRR95% CIP
MSC risk level 111     84     
 High 26 0% 0.07 0.00–1.05 0.0009 19 0% 0.07 0.00–1.16 0.0028 
 Intermediate/low 85 28% (Reference)  65 34% (Reference)  
PD-L1 expression 104     84     
 PNC < 50% 66 15% 0.41 0.20–0.83 0.0158 50 18% 0.47 0.23–0.98 0.0463 
 PNC ≥ 50% 38 37% (Reference)  34 38% (Reference)  
MSC and PD-L1 131     84     
 0 Favorable markers 35 3% 0.11 0.02–0.78 0.0024 12 0% 0.12 0.01–1.93 0.0357 
 1–2 Favorable markers 96 26% (Reference)  72 31% (Reference)  
All patientsPatients with both markers available
NORRRR95% CIPNORRRR95% CIP
MSC risk level 111     84     
 High 26 0% 0.07 0.00–1.05 0.0009 19 0% 0.07 0.00–1.16 0.0028 
 Intermediate/low 85 28% (Reference)  65 34% (Reference)  
PD-L1 expression 104     84     
 PNC < 50% 66 15% 0.41 0.20–0.83 0.0158 50 18% 0.47 0.23–0.98 0.0463 
 PNC ≥ 50% 38 37% (Reference)  34 38% (Reference)  
MSC and PD-L1 131     84     
 0 Favorable markers 35 3% 0.11 0.02–0.78 0.0024 12 0% 0.12 0.01–1.93 0.0357 
 1–2 Favorable markers 96 26% (Reference)  72 31% (Reference)  

NOTE: P values by two-tailed Fisher exact probability test are reported.

The two markers were first combined considering the 131 patients with any data available, in terms of the presence or absence of any favorable marker, that is, MSC intermediate- and low-risk level and/or PD-L1 ≥ 50%. In this scenario, the ORR was significantly lower in patients with advanced NSCLC with no favorable markers as compared with those with 1–2 favorable markers (RR = 0.11; 95% CI, 0.02–0.78; P = 0.0024). Similar results were observed when considering the subgroup of 84 patients with advanced NSCLC with both MSC and PD-L1 available (Table 2).

Plasma MSC risk level and PD-L1 tumor expression are associated with patients' survival

Patients' survival was evaluated according to plasma MSC risk level and PD-L1 expression on tumor cells both alone and combined. As shown in Fig. 2A, lower 1-year PFS was observed in the 21 patients with lung cancer with a MSC high-risk level than in the 47 with MSC intermediate and the 33 with MSC low-risk levels (P < 0.0001). As expected, the 34 patients with PD-L1 tumor expression ≥50% showed a better outcome than the 28 and 35 patients with PD-L1 signal <1% and 1%–49% (P < 0.0001; Fig. 2B). Similar results were observed when considering OS (Fig. 3A and B).

Figure 2.

Kaplan–Meier curves reporting the PFS of patients with advanced lung cancer treated with ICIs in strata defined by the MSC and PD-L1 tumor expression. Analysis of 101 subjects with available plasma MSC risk level: 21 high versus 47 intermediate versus 33 low (A); 97 subjects with available PD-L1 expression on tumor cells: 28 with PD-L1 <1% versus 35 with PD-L1 1%–49% versus 34 with PD-L1 ≥ 50% (B); 120 subjects with at least one marker available: 32 with no favorable markers versus 88 with any favorable marker (MSC intermediate or low and/or PD-L1 ≥ 50%; C); and 78 subjects with both markers available stratified according to the presence of 0, 1, or 2 favorable markers (D). P values by log-rank test are reported.

Figure 2.

Kaplan–Meier curves reporting the PFS of patients with advanced lung cancer treated with ICIs in strata defined by the MSC and PD-L1 tumor expression. Analysis of 101 subjects with available plasma MSC risk level: 21 high versus 47 intermediate versus 33 low (A); 97 subjects with available PD-L1 expression on tumor cells: 28 with PD-L1 <1% versus 35 with PD-L1 1%–49% versus 34 with PD-L1 ≥ 50% (B); 120 subjects with at least one marker available: 32 with no favorable markers versus 88 with any favorable marker (MSC intermediate or low and/or PD-L1 ≥ 50%; C); and 78 subjects with both markers available stratified according to the presence of 0, 1, or 2 favorable markers (D). P values by log-rank test are reported.

Close modal
Figure 3.

Kaplan–Meier curves reporting the OS of patients with advanced lung cancer treated with ICIs in strata defined by the MSC and PD-L1 tumor expression. Analysis of 111 subjects with available plasma MSC risk level: 26 high versus 51 intermediate versus 34 low (A); 104 subjects with available PD-L1 expression on tumor cells: 30 with PD-L1 <1% versus 36 with PD-L1 1%–49% versus 38 with PD-L1 ≥ 50% (B); 131 subjects with at least one marker available: 35 with no favorable markers versus 96 with any favorable marker (MSC intermediate or low and/or PD-L1 ≥ 50%; C); and 84 subjects with both markers available stratified according to the presence of 0, 1, or 2 favorable markers (D). P values by log-rank test are reported.

Figure 3.

Kaplan–Meier curves reporting the OS of patients with advanced lung cancer treated with ICIs in strata defined by the MSC and PD-L1 tumor expression. Analysis of 111 subjects with available plasma MSC risk level: 26 high versus 51 intermediate versus 34 low (A); 104 subjects with available PD-L1 expression on tumor cells: 30 with PD-L1 <1% versus 36 with PD-L1 1%–49% versus 38 with PD-L1 ≥ 50% (B); 131 subjects with at least one marker available: 35 with no favorable markers versus 96 with any favorable marker (MSC intermediate or low and/or PD-L1 ≥ 50%; C); and 84 subjects with both markers available stratified according to the presence of 0, 1, or 2 favorable markers (D). P values by log-rank test are reported.

Close modal

Combining the two markers in the whole series, the 1-year PFS was 24% in 88 patients with at least one favorable marker, while 32 patients with no favorable markers had disease progression within 5 months since ICI starting (Fig. 2C; P < 0.0001). Considering OS, 52% of subjects with favorable markers and the 11% without favorable markers were still alive 1 year after starting ICIs treatment (Fig. 3C; P < 0.0001).

Analysis was then restricted to the patients with both MSC and PD-L1 available data. This subset of patients could be stratified into three distinct risk groups showing 39%–18%–0% 1-year PFS (P < 0.0001) and 88%–44%–0% 1-year OS (P < 0.0001), according to the presence of 2, 1, or 0 favorable markers, respectively (Fig. 2D and 3D). For the mid group, no main differences in terms of OS were observed if the favorable marker was MSC intermediate and low or PD-L1 <50% (Supplementary Fig. S2).

Multivariate analysis confirmed data on patients' survival

PFS and OS in strata of each marker were then analyzed by Cox proportional hazard models in univariate and multivariate analyses. The distribution of covariates across MSC risk level and PD-L1 tumor expression are described in Supplementary Table S2. In the whole series, MSC risk level, PD-L1 tumor expression, as well as their combination remained significantly associated with both PFS and OS by univariate and multivariate analyses (Supplementary Table S3).

In the subgroup of patients with both markers available, patients with MSC intermediate-/low-risk level reported a significant reduction in disease progression and mortality as compared with those with high MSC risk level (Table 3). The corresponding multivariate HRs were 0.35 (95% CI, 0.18–0.70; P = 0.0026) and 0.28 (95% CI, 0.12–0.58; P = 0.0007). Significant reduction in disease progression (multivariate HR = 0.35; 95% CI, 0.19–0.63; P = 0.0006) and mortality (multivariate HR = 0.43; 95% CI, 0.21–0.88; P = 0.0211) was also observed in patients with PNC ≥50% as compared with those with PNC <50%. When the two markers were considered together, patients with at least one favorable marker reported a significant lower probability in disease progression (multivariate HR = 0.25; 95% CI, 0.12–0.56; P = 0.0006) and mortality (multivariate HR = 0.28; 95% CI, 0.12–0.65; P = 0.0034), as compared with those with no favorable markers.

Table 3.

Results from the Cox proportional hazards models on PFS and OS in the subset of patients with both plasma MSC risk level and PD-L1 tumor expression available

PFSOS
NHR (95% CI)PHRa (95% CI)PNHR (95% CI)PHRa (95% CI)P
MSC risk level 78     84     
 High 16   19   
 Intermediate/low 62 0.33 (0.18–0.62) 0.0006 0.35 (0.18–0.70) 0.0026 65 0.29 (0.15–0.58) 0.0004 0.28 (0.12–0.58) 0.0007 
PD-L1 expression 78     84     
 PNC <50% 48   50   
 PNC ≥50% 30 0.37 (0.20–0.66) 0.0009 0.35 (0.19–0.63) 0.0006 34 0.39 (0.19–0.79) 0.0093 0.43 (0.21–0.88) 0.0211 
MSC and PD-L1 78     84     
 0 Favorable markers 12   12   
 1–2 Favorable markers 66 0.24 (0.11–0.49) 0.0001 0.25 (0.12–0.56) 0.0006 72 0.27 (0.13–0.58) 0.0008 0.28 (0.12–0.65) 0.0034 
PFSOS
NHR (95% CI)PHRa (95% CI)PNHR (95% CI)PHRa (95% CI)P
MSC risk level 78     84     
 High 16   19   
 Intermediate/low 62 0.33 (0.18–0.62) 0.0006 0.35 (0.18–0.70) 0.0026 65 0.29 (0.15–0.58) 0.0004 0.28 (0.12–0.58) 0.0007 
PD-L1 expression 78     84     
 PNC <50% 48   50   
 PNC ≥50% 30 0.37 (0.20–0.66) 0.0009 0.35 (0.19–0.63) 0.0006 34 0.39 (0.19–0.79) 0.0093 0.43 (0.21–0.88) 0.0211 
MSC and PD-L1 78     84     
 0 Favorable markers 12   12   
 1–2 Favorable markers 66 0.24 (0.11–0.49) 0.0001 0.25 (0.12–0.56) 0.0006 72 0.27 (0.13–0.58) 0.0008 0.28 (0.12–0.65) 0.0034 

aAdjusted for age, sex, smoking status, presence of liver metastasis, MSC risk level, and line of therapy.

Plasma MSC risk level to monitor responsive and stable disease during treatment with ICIs

For a subset of patients' representative of each class of response to ICIs, longitudinally collected plasma samples were analyzed to monitor changes in MSC risk level during treatment: 4 R, 5 SD, and 6 P were included. In 2 R patients the MSC decreased from intermediate- to low-risk level after starting ICI treatment (Supplementary Fig. S3A and S3B). Two R MSC low-risk patients remained MSC low risk in response to ICIs and in one of these MSC risk increased at tumor progression (Supplementary Fig. S3C and S3D). A similar fluctuation of MSC risk level was observed in the 5 SD patients: 3 with MSC low-risk level remained low until or shortly before disease progression (Supplementary Fig. S3E–S3G). One MSC high risk decreased to intermediate risk and one who remained MSC intermediate risk had disease progression 2 weeks later (Supplementary Fig. S3H and S3I). On the other hand, in patients with diseases progression, the changes in the MSC risk level during treatment were not uniform across the six subjects analyzed (Supplementary Fig. S3J–S3O).

With the advent of ICIs, new insights into the crosstalk between tumor and the host involving both innate and the adaptive immunity, are emerging (30). Indeed, looking for molecules that reflect an impaired crosstalk could represent a promising strategy to identify diagnostic, prognostic, and predictive biomarkers. PD-L1 is the only biomarker currently used in clinical practice but its role is still debated: different assays and cutoffs have been used in the clinical trials (29, 31, 32); nonetheless ICIs efficacy was achieved also in PD-L1–negative tumors (33, 34). More recently, in patients with advanced NSCLC enrolled in the CheckMate 227 clinical trial, TMB showed to be an independent PFS predictive factor, regardless of PD-L1 expression (9). In addition, a TMB blood-based assay (bTMB) was standardized in two large retrospective series and was still predictive for PFS. However, bTMB failed to predict OS and almost the 25% of patients did not achieved the minimum amount of circulating tumor DNA (ctDNA) for optimal assay performance or did not pass quality control steps (10).

In this study, we have shown for the first time that a circulating miRNA signature classifier with prognostic value can supplement PD-L1 to identify patients with worse ORR, PFS, and OS in a consecutive series of 140 advanced NSCLC cases treated with ICIs. A recent article identified a signature of seven miRNAs associated with 6-month OS in a retrospective series composed of 20 (training) and 31 (validation) patients with advanced NSCLC treated with nivolumab (35). No overlap between the seven miRNAs and the 24 composing the MSC was observed.

Because the use of ORR and PFS as surrogate endpoints of OS is still a controversial issue in immunotherapy settings (36), we presented the results according to all three endpoints. In a first data analysis according to ORR, the MSC test alone seemed to be able to identify a subset of nonresponding patients corresponding to the 19% of the whole series. Nevertheless, the identification of a subgroup of patients with 0% 1-year PFS and OS was achieved only by combining MSC and PD-L1. The added value of these markers was also confirmed by multivariate analysis, where the adjusted HR improved from 0.35 (PFS) and 0.43 (OS) of PD-L1 alone, to 0.25 (PFS) and 0.28 (OS) when combined with the MSC.

As for TMB, some limitations have also been identified in this study. No adequate tumor tissue sample was available for PD-L1 expression in 26% of patients. In addition, the MSC testing efficiency was limited by the sensitivity of the assay to hemolysis (25), which affected 20% of the plasma samples analyzed. Nevertheless, when tissue and plasma samples were combined, the percentage of patients with adequate material for biomarker evaluation increased to 94%. Because ICI treatment is rapidly becoming a first-line therapy for NSCLC (29, 31), tissue and liquid biopsies are likely to become more easily accessible. Indeed in this series, the percentage of inadequate plasma and tissue samples in first-line treated patients decreased to 13% and 3%, respectively.

Liquid biopsy based on ctDNA was recently assessed to monitor the response to immunotherapy in patients with advanced NSCLC (37–39). Three studies measured the frequencies of mutations identified in the tumor tissue in plasma collected from a total of 73 patients with advanced NSCLC by targeted next-generation sequencing or droplet digital PCR. At baseline, ctDNA was found in the plasma samples of 42 (58%) patients and was not associated with response to ICIs. On the other hand, the analysis of longitudinally collected plasma samples demonstrated that ctDNA levels decreased in responding patients and increased in nonresponding patients, thus supporting its use as an “on treatment” biomarker. In our study, analysis on longitudinally collected plasma samples suggested that MSC risk level follows tumor response to treatment in R and SD patients, although not homogeneous in 6 P patients. These preliminary results must be further validated in larger series.

Immunotherapy with ICIs may provide long-term benefits in approximately 25% of patients with NSCLC (40). Nonetheless, tumor progression or even hyperprogression is observed in nonresponding patients (41, 42). This first exploratory prospective study suggests that circulating miRNAs can supplement standard tissue biopsy in the clinical practice. Larger and possibly randomized studies are needed to establish the efficacy of MSC to select a subgroup of patients who do not benefit of ICI treatment.

D. Signorelli is a consultant/advisory board member for AstraZeneca. M.C. Garassino reports receiving commercial research grants from MSD and Ely Lilly and speakers bureau honoraria from Bristol-Myers Squibb, MSD, Ely Lilly, AstraZeneca, Roche, Boehringer Ingelheim, Novartis, Bayer, Pfizer, Sanofy, and Italfarmaco, and is a consultant/advisory board member for AstraZeneca, Roche, Bristol-Myers Squibb, MSD, Eli Lilly, Novartis, and Bayer. G. Sozzi, M.C. Garassino, U. Pastorino, and M. Boeri are co-inventors and co-owners of a patent regarding the new use of the MSC test in combination with PD-L1 tumor expression to identify advanced NSCLC patients with worse survival in immunotherapy regimens. No potential conflicts of interest were disclosed by the other authors.

Conception and design: M. Boeri, M. Milione, U. Pastorino, M.C. Garassino, G. Sozzi

Development of methodology: M. Boeri, M. Milione, C. Verri, G. Centonze, M.C. Garassino

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Proto, D. Signorelli, G. Lo Russo, M.C. Garassino

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Boeri, M. Milione, C. Proto, D. Signorelli, G. Lo Russo, C. Galeone, U. Pastorino, M.C. Garassino, G. Sozzi

Writing, review, and/or revision of the manuscript: M. Boeri, M. Milione, C. Proto, D. Signorelli, G. Lo Russo, C. Galeone, C. Verri, A. Martinetti, E. Sottotetti, M.C. Garassino, G. Sozzi

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Proto, D. Signorelli, G. Lo Russo, M. Mensah, A. Martinetti, E. Sottotetti, M.C. Garassino

Study supervision: M. Boeri, M. Milione, C. Proto, D. Signorelli, G. Lo Russo, M.C. Garassino, G. Sozzi

This work was supported by Italian Ministry of Health (5 × 1000 Funds – 2014 and GR-2016-02361849); investigator grants No. 18812 (to G. Sozzi) from the Italian Association for Cancer Research; grant UO1 CA166905 from the NCI. M. Boeri was supported by a Fondazione Pezcoller Fellowship. The authors thank Dr. G. Apolone for his valuable suggestions, Dr. C. Borzi for technical assistance, and the granting agencies: the Italian Ministry of Health, the Italian Association for Cancer Research, the US NCI, and Fondazione Pezcoller for financial support. The authors acknowledge the editorial assistance provided by the Nature Research Editing Service.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Silva
M
,
Pastorino
U
,
Sverzellati
N
. 
Lung cancer screening with low-dose CT in Europe: strength and weakness of diverse independent screening trials
.
Clin Radiol
2017
;
72
:
389
400
.
2.
Quail
DF
,
Joyce
JA
. 
Microenvironmental regulation of tumor progression and metastasis
.
Nat Med
2013
;
19
:
1423
37
.
3.
McGranahan
N
,
Furness
AJ
,
Rosenthal
R
,
Ramskov
S
,
Lyngaa
R
,
Saini
SK
, et al
Clonal neoantigens elicit T cell immunoreactivity and sensitivity to immune checkpoint blockade
.
Science
2016
;
351
:
1463
9
.
4.
Antonia
SJ
,
Villegas
A
,
Daniel
D
,
Vicente
D
,
Murakami
S
,
Hui
R
, et al
Durvalumab after chemoradiotherapy in stage III non-small-cell lung cancer
.
N Engl J Med
2017
;
377
:
1919
29
.
5.
Vokes
EE
,
Ready
N
,
Felip
E
,
Horn
L
,
Burgio
MA
,
Antonia
SJ
, et al
Nivolumab versus docetaxel in previously treated advanced non-small cell lung cancer (CheckMate 017 and CheckMate 057): 3-year update and outcomes in patients with liver metastases
.
Ann Oncol
2018
;
29
:
959
65
.
6.
Gandhi
L
,
Rodriguez-Abreu
D
,
Gadgeel
S
,
Esteban
E
,
Felip
E
,
De
AF
, et al
Pembrolizumab plus chemotherapy in metastatic non–small-cell lung cancer
.
N Engl J Med
2018
;
378
:
2078
92
.
7.
Rizvi
H
,
Sanchez-Vega
F
,
La
K
,
Chatila
W
,
Jonsson
P
,
Halpenny
D
, et al
Molecular determinants of response to anti-programmed cell death (PD)-1 and anti-programmed death-ligand 1 (PD-L1) blockade in patients with non-small-cell lung cancer profiled with targeted next-generation sequencing
.
J Clin Oncol
2018
;
36
:
633
41
.
8.
Rizvi
NA
,
Hellmann
MD
,
Snyder
A
,
Kvistborg
P
,
Makarov
V
,
Havel
JJ
, et al
Cancer immunology. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer
.
Science
2015
;
348
:
124
8
.
9.
Hellmann
MD
,
Ciuleanu
TE
,
Pluzanski
A
,
Lee
JS
,
Otterson
GA
,
Audigier-Valette
C
, et al
Nivolumab plus ipilimumab in lung cancer with a high tumor mutational burden
.
N Engl J Med
2018
;
378
:
2093
104
.
10.
Gandara
DR
,
Paul
SM
,
Kowanetz
M
,
Schleifman
E
,
Zou
W
,
Li
Y
, et al
Blood-based tumor mutational burden as a predictor of clinical benefit in non-small-cell lung cancer patients treated with atezolizumab
.
Nat Med
2018
;
24
:
1441
8
.
11.
Yu
H
,
Chen
Z
,
Ballman
K
,
Watson
MA
,
Govindan
R
,
Lanc
I
, et al
Correlation of PD-L1 expression with tumor mutation burden and gene signatures for prognosis in early-stage squamous cell lung carcinoma
.
J Thorac Oncol
2019
;
14
:
25
36
.
12.
Boeri
M
,
Verri
C
,
Conte
D
,
Roz
L
,
Modena
P
,
Facchinetti
F
, et al
MicroRNA signatures in tissues and plasma predict development and prognosis of computed tomography detected lung cancer
.
Proc Natl Acad Sci U S A
2011
;
108
:
3713
8
.
13.
Sozzi
G
,
Boeri
M
,
Rossi
M
,
Verri
C
,
Suatoni
P
,
Bravi
F
, et al
Clinical utility of a plasma-based miRNA signature classifier within computed tomography lung cancer screening: a correlative MILD trial study
.
J Clin Oncol
2014
;
32
:
768
73
.
14.
Sestini
S
,
Boeri
M
,
Marchiano
A
,
Pelosi
G
,
Galeone
C
,
Verri
C
, et al
Circulating microRNA signature as liquid-biopsy to monitor lung cancer in low-dose computed tomography screening
.
Oncotarget
2015
;
20
:
32868
77
.
15.
Verri
C
,
Borzi
C
,
Holscher
T
,
Dugo
M
,
Devecchi
A
,
Drake
K
, et al
Mutational profile from targeted NGS predicts survival in LDCT screening-detected lung cancers
.
J Thorac Oncol
2017
;
12
:
922
31
.
16.
Fernandez-Messina
L
,
Gutierrez-Vazquez
C
,
Rivas-Garcia
E
,
Sanchez-Madrid
F
,
de la
FH
. 
Immunomodulatory role of microRNAs transferred by extracellular vesicles
.
Biol Cell
2015
;
107
:
61
77
.
17.
Bayraktar
R
,
Van
RK
,
Calin
GA
. 
Cell-to-cell communication: microRNAs as hormones
.
Mol Oncol
2017
;
11
:
1673
86
.
18.
Bronisz
A
,
Godlewski
J
,
Wallace
JA
,
Merchant
AS
,
Nowicki
MO
,
Mathsyaraja
H
, et al
Reprogramming of the tumour microenvironment by stromal PTEN-regulated miR-320
.
Nat Cell Biol
2012
;
14
:
159
67
.
19.
Baer
C
,
Squadrito
ML
,
Laoui
D
,
Thompson
D
,
Hansen
SK
,
Kiialainen
A
, et al
Suppression of microRNA activity amplifies IFN-gamma-induced macrophage activation and promotes anti-tumour immunity
.
Nat Cell Biol
2016
;
18
:
790
802
.
20.
Li
L
,
Zhang
J
,
Diao
W
,
Wang
D
,
Wei
Y
,
Zhang
CY
, et al
MicroRNA-155 and microRNA-21 promote the expansion of functional myeloid-derived suppressor cells
.
J Immunol
2014
;
192
:
1034
43
.
21.
Pyfferoen
L
,
Brabants
E
,
Everaert
C
,
De
CN
,
Heyns
K
,
Deswarte
K
, et al
The transcriptome of lung tumor-infiltrating dendritic cells reveals a tumor-supporting phenotype and a microRNA signature with negative impact on clinical outcome
.
Oncoimmunology
2016
;
6
:
e1253655
.
22.
Smigielska-Czepiel
K
,
van den
BA
,
Jellema
P
,
van der Lei
RJ
,
Bijzet
J
,
Kluiver
J
, et al
Comprehensive analysis of miRNA expression in T-cell subsets of rheumatoid arthritis patients reveals defined signatures of naive and memory Tregs
.
Genes Immun
2014
;
15
:
115
25
.
23.
Fortunato
O
,
Borzi
C
,
Milione
M
,
Centonze
G
,
Conte
D
,
Boeri
M
, et al
Circulating mir-320a promotes immunosuppressive macrophages M2 phenotype associated with lung cancer risk
.
Int J Cancer
2018
Nov 13 [Epub ahead of print]
.
24.
Kohlhapp
FJ
,
Mitra
AK
,
Lengyel
E
,
Peter
ME
. 
MicroRNAs as mediators and communicators between cancer cells and the tumor microenvironment
.
Oncogene
2015
;
34
:
5857
68
.
25.
Mensah
M
,
Borzi
C
,
Verri
C
,
Suatoni
P
,
Conte
D
,
Pastorino
U
, et al
MicroRNA based liquid biopsy: the experience of the plasma miRNA signature classifier (MSC) for lung cancer screening
.
J Vis Exp
2017
;
10
:
56326
.
26.
Fortunato
O
,
Boeri
M
,
Verri
C
,
Conte
D
,
Mensah
M
,
Suatoni
P
, et al
Assessment of circulating microRNAs in plasma of lung cancer patients
.
Molecules
2014
;
19
:
3038
54
.
27.
Nakayama
T
,
Zaman
MM
,
Tanaka
H
. 
Reporting of attributable and relative risks, 1966–97
.
Lancet
1998
;
351
:
1179
.
28.
Deeks
JJ
,
Higgins
PTJ
, on behalf of the Statistical Methods Group of The Cochrane Collaboration
. 
Statistical algorithms in Review Manager 5.
Available from:
https://training.cochrane.org/handbook/statistical-methods-revman5.
29.
Reck
M
. 
Pembrolizumab as first-line therapy for metastatic non-small-cell lung cancer
.
Immunotherapy
2018
;
10
:
93
105
.
30.
Butt
AQ
,
Mills
KH
. 
Immunosuppressive networks and checkpoints controlling antitumor immunity and their blockade in the development of cancer immunotherapeutics and vaccines
.
Oncogene
2014
;
33
:
4623
31
.
31.
Carbone
DP
,
Reck
M
,
Paz-Ares
L
,
Creelan
B
,
Horn
L
,
Steins
M
, et al
First-line nivolumab in stage IV or recurrent non-small-cell lung cancer
.
N Engl J Med
2017
;
376
:
2415
26
.
32.
Tsao
MS
,
Kerr
KM
,
Kockx
M
,
Beasley
MB
,
Borczuk
AC
,
Botling
J
, et al
PD-L1 immunohistochemistry comparability study in real-life clinical samples: results of Blueprint phase 2 project
.
J Thorac Oncol
2018
;
13
:
1302
11
.
33.
Felip Font
E
,
Gettinger
SN
,
Burgio
MA
,
Antonia
SJ
,
Holgado
E
,
Spigel
DR
, et al
Three-year follow-up from CheckMate 017/057: nivolumab versus docetaxel in patients with previously treated advanced non-small cell lung cancer (NSCLC)
.
Ann Oncol
2017
;
28
:
v460–96
.
34.
Fehrenbacher
L
,
von
PJ
,
Park
K
,
Rittmeyer
A
,
Gandara
DR
,
Ponce
AS
, et al
Updated efficacy analysis including secondary population results for OAK: a randomized phase III study of atezolizumab versus docetaxel in patients with previously treated advanced non-small cell lung cancer
.
J Thorac Oncol
2018
;
13
:
1156
70
.
35.
Halvorsen
AR
,
Sandhu
V
,
Sprauten
M
,
Flote
VG
,
Kure
EH
,
Brustugun
OT
, et al
Circulating microRNAs associated with prolonged overall survival in lung cancer patients treated with nivolumab
.
Acta Oncol
2018
;
57
:
1225
31
.
36.
Mushti
SL
,
Mulkey
F
,
Sridhara
R
. 
Evaluation of overall response rate and progression-free survival as potential surrogate endpoints for overall survival in immunotherapy trials
.
Clin Cancer Res
2018
;
24
:
2268
75
.
37.
Cabel
L
,
Riva
F
,
Servois
V
,
Livartowski
A
,
Daniel
C
,
Rampanou
A
, et al
Circulating tumor DNA changes for early monitoring of anti-PD1 immunotherapy: a proof-of-concept study
.
Ann Oncol
2017
;
28
:
1996
2001
.
38.
Iijima
Y
,
Hirotsu
Y
,
Amemiya
K
,
Ooka
Y
,
Mochizuki
H
,
Oyama
T
, et al
Very early response of circulating tumour-derived DNA in plasma predicts efficacy of nivolumab treatment in patients with non-small cell lung cancer
.
Eur J Cancer
2017
;
86
:
349
57
.
39.
Goldberg
SB
,
Narayan
A
,
Kole
AJ
,
Decker
RH
,
Teysir
J
,
Carriero
NJ
, et al
Early assessment of lung cancer immunotherapy response via circulating tumor DNA
.
Clin Cancer Res
2018
;
24
:
1872
80
.
40.
Assi
HI
,
Kamphorst
AO
,
Moukalled
NM
,
Ramalingam
SS
. 
Immune checkpoint inhibitors in advanced non-small cell lung cancer
.
Cancer
2018
;
124
:
248
61
.
41.
Champiat
S
,
Dercle
L
,
Ammari
S
,
Massard
C
,
Hollebecque
A
,
Postel-Vinay
S
, et al
Hyperprogressive disease is a new pattern of progression in cancer patients treated by anti-PD-1/PD-L1
.
Clin Cancer Res
2017
;
23
:
1920
8
.
42.
Lo Russo
G
,
Moro
M
,
Sommariva
M
,
Cancila
V
,
Boeri
M
,
Centonze
G
, et al
Antibody-Fc/FcR interaction on macrophages as a mechanism for hyperprogressive disease in non-small cell lung cancer subsequent to PD-1/PD-L1 blockade
.
Clin Cancer Res
2018
Sep 11 [Epub ahead of print]
.