2. PD-L1 Expression
The expression of PD-L1 in cases of SCLC is reported to be less frequent than in cases of NSCLC. The scarce cellularity of SCLC specimens limits the ability to detect PD-L1 
. In spite of previous reports that showed a positive correlation between PD-L1 expression levels with a limited disease (LD) stage and a favourable outcome, currently, the role of PD-L1 expression in SCLC patients is controversial 
. Only about a third of patients who were enrolled in the IMPower133 trial (137/403) had evaluable tumour tissue. For those patients with adequate quantities of tumour material, the PD-L1 expression level was <1% in tumour cells in almost all cases (129/137), while the PD-L1 expression level in immune cells was <1% in about half of the cases (68/137). No correlations between PD-L1 expression levels in tumour cells or immune cells and clinical outcomes have been found in patients treated with chemotherapy plus atezolizumab (Table 1
). Conversely, patients with both PD-L1-negative tumours and immune cells showed an improvement in OS rate (median OS, 10.2 months versus 8.3 months, respectively; HR, 0.51; 95% CI, 0.30–0.89) and PFS rate (median PFS, 5.4 versus 4.2 months; HR, 0.52; 95% CI, 0.31–0.88) receiving chemotherapy plus atezolizumab versus chemotherapy plus placebo. Therefore, these data suggest that PD-L1 expression is not a predictive biomarker in patients with SCLC receiving chemotherapy plus ICIs. An OS benefit was observed in patients with PD-L1 expression ≥5%; however, the number of patients in this subgroup was very low 
. Concerning the use of atezolizumab in the second-line setting, the IFCT-1603 phase II RCT compared atezolizumab versus chemotherapy. In this study, the tumour PD-L1 expression level was evaluated using the SP-142 assay. Unfortunately, of the 53 evaluable tumour specimens, only one showed tumour PD-L1 expression, excluding the opportunity for evaluations of predictive value 
Table 1. Studies concerning PD-L1 expression in SCLC.
||Pattern of PD-L1 Expression
|Phase II of maintenance Pembrolizumab. 
||ORR 37.5% in 8 patients with PD-L1 positive vs. 8.3% in 12 patients with PD-L1 negative.
||Median PFS 6.5 months in 8 patients with PD-L1 positive vs. 1.3 months in 12 patients with PD-L1 negative.
||Median OS 12.8 months in 8 patients with PD-L1 positive vs. 7.6 months in 12 patients with PD-L1 negative.
|Phase II of maintenance Pembrolizumab. 
||Median PFS 11 months among 3 patients with PD-L1 positive.
|Phase III Atezolizumab, carboplatin, and etoposide. 
||Tumour or immune cells
||Median OS 10.2 months in 28 patients with PD-L1 negative in the Atezolizumab arm vs. 8.3 months in 37 patients with PD-L1 negative in the Placebo arm.
|Phase Ib of Pembroliuzumab for only PD-L1 positive patients, Keynote-028. 
||CPS and stroma
||Median PFS 1.9 months in 24 patients with PD-L1 positive
||Median OS 9.7 months in 24 patients with PD-L1 positive
|Phase II of Pembrolizumab Keynote-158. 
||ORR 35.7% in 42 patients with PD-L1 positive vs. 6% in 50 patients with PD-L1 negative.
||Median PFS 2.1 months in 42 patients with PD-L1 positive vs. 1.9 months in 50 patents with PD-L1 negative.
||Median OS 14.6 months in 42 patients with PD-L1 positive vs. 7.7 months in 50 patients with PD-L1 negative.
|Phase II Study, patients with relapsed SCLC treated with pembrolizumab plus amrubicin. 
||ORR 58% in 19 patients with PD-L1 positive vs. 33% in 6 patients with PD-L1 negative or not assessable.
||Median PFS 4.4 months in 19 patients with PD-L1 positive vs. 3.0 months in 6 patients with PD-L1 negative or not assessable.
In accordance with the analysis of the IMPower133 trial, the phase III CASPIAN trial, evaluating the use of durvalumab in combination with etoposide plus either cisplatin or carboplatin, showed no significant impact of PD-L1 expression on the effect of treatment. About half of the tumour specimens (277/531) were evaluable, showing low levels of PD-L1 expression. In particular, 22% and 5% of patients had tumours with expression levels ≥1% in immune and tumour cells, respectively. Neither the PD-L1 expression in tumour cells nor in immune cells correlated with better outcomes in patients treated with chemotherapy plus durvalumab 
Similarly to the CASPIAN and IMPower133 trials, in the phase III study, Keynote 604, the addition of pembrolizumab to a platinum and etoposide regimen was shown to improve the PFS rate (12-month PFS, 13.6% versus 3.1%), and prolonged OS (24-month, 22.5% versus 11%), although the significance threshold was not reached for the OS rate. In this trial, the PFS and OS HRs were shown to be similar in participants with PD-L1-positive as well as PD-L1-negative malignancies 
In a single-arm, phase II study investigating first-line maintenance pembrolizumab in patients with SCLC, 66% of the tumour specimens (30/50) were evaluable. Only three patients showed PD-L1 expression levels ≥1%. Among these, one had no measurable disease at study entry and two responded to therapy with a median PFS of 11 months. PD-L1 expression was also detected at the tumour–stromal interface in 8 out of 20 patients who responded or who had stable disease after induction chemotherapy. PD-L1 expression at the tumour–stromal interface resulted in a better outcome, specifically, an improvement of ORRs (37.5% versus 8.3%), PFS (6.5 versus 1.3 months), and median OS (12.8 versus 7.6 months) 
The combined positive score (CPS), consisting in the number of PD-L1-positive cells (tumour and immune cells) divided by the total number of viable tumour cells and multiplied by 100, was evaluated in the phase 1b multicohort trial KEYNOTE-028, involving patients with relapsed SCLC who were treated with pembrolizumab. In this trial, a CPS of ≥1% was an inclusion criterion for treatment with pembrolizumab, showing an ORR of 33% (8 of 24) 
. More recently, KEYNOTE-158, a phase II trial investigating the efficacy of pembrolizumab monotherapy, compared patients with a CPS ≥1% (n
= 42) with patients characterized by a CPS <1% (n
= 50). A CPS ≥1% was associated with an improvement of ORRs (35.7% versus 6%) and median OS (14.6 months versus 7.7 months), and an OS rate of 1 year for 53.1% patients with positive CPSs. The ORR was 27% (4 out of 15) among patients with unknown PD-L1 status (Table 1
), whereas a pooled analysis of KEYNOTE-028 and KEYNOTE-158, both including patients with PD-L1-positive tumours, showed an ORR of 19.3% (16 out of 83). Concerning responsive patients, there were two complete responses; of these, one concerned a patient with a PD-L1-positive tumour. There were 14 partial responses, with almost all of these regarding PD-L1-positive tumours (13/14). Interestingly, 61% of responders experienced responses that lasted 18 months or longer 
In a recent single-arm phase II Study, patients with relapsed SCLC were treated with pembrolizumab plus amrubicin until they showed progression. In total, 76% (n
= 19) of the patients had a CPS ≥1%, showing a better outcome than patients with CPS <1% or those who were not assessable (n
= 6). In particular, the post hoc analyses demonstrated a higher ORR (58% versus 33%), and mPFS (4.4 months versus 3.0 months, hazard ratio (HR) = 0.73, 95% CI, 0.25 to 1.91). Similarly, patients with higher levels of tumour-infiltrating lymphocytes (TILs) (n
= 13) had a better ORR and mPFS 
Even though PD-L1 seems to be a potential predictive biomarker of response to pembrolizumab in the same population of patients, this was not shown in clinical trials using nivolumab. Concerning the CheckMate032 trial, no differences of outcomes were found between patients with positive PD-L1 expression and patients with PD-L1 <1% or those who were not assessable, both for patients who received nivolumab and for patients who received nivolumab plus ipilimumab 
. The researchers evaluated tumour PD-L1 expression with a qualitative immunohistochemical assay using Monoclonal Rabbit Anti-PD-L1, namely, 28-8 pharmDx antibody 
. The tumour specimens were obtained in the pretreatment phase, within 3 months of beginning ICIs therapy. In total, 148 tumour samples were evaluated, and those with ≥100 evaluable tumour cells were considered as acceptable samples. Among patients in the nivolumab monotherapy arm, the ORR was 38% (3 out of 8) if the PD-L1 level was ≥1%, 28% (12 out of 43) if the PD-L1 level was <1%, and 24% (6 out of 25) in those with non-evaluable expression. In the nivolumab (1 mg/kg) and ipilimumab (3 mg/kg) arm, ORRs were 33% (2 out of 6), 36% (8 out of 22), and 33% (6 out of 18), respectively. In comparison, ORRs were 60% (3 out of 5), 24% (7 out of 29), and 15% (2 out of 13), respectively, among these subgroups in the nivolumab (3 mg/kg) and ipilimumab (1 mg/kg) arm 
. Summarizing the data concerning PD-L1 expression in SCLC, this biomarker does not seem suitable for patients treated with chemotherapy plus ICIs. Additionally, considering the heterogeneity and plasticity of SCLC, future studies should evaluate different biomarkers than PD-L1 expression.