Cancer cells are under intense evolutionary pressure and
Cancer 9(S)-HODE are under intense evolutionary pressure and therefore comprise a heterogeneous population. Furthermore, given that PD-L1 expression is induced by IFNγ in adaptive immune resistance , heterogeneity of PD-L1 expression is a critical and inevitable problem, particularly when assessed using small biopsy specimens. Because the current decision to use ICIs for the treatment of NSCLC patients is partly dependent on the tumor PD-L1 expression status, this is becoming a more relevant problem that requires urgent attention. Though previous studies have compared the PD-L1 expression status between biopsy samples and primary tumors, EBUS-TBNA-derived specimens were not included [15,17,46]. Only a few studies have reported the utility of biopsy specimens obtained by EBUS-TBNA for the assessment of PD-L1 expression. Sakakibara et al. first reported the concordance among EBUS-TBNA, TBB, and resected specimens . They demonstrated strong correlations between EBUS-TBNA-derived and corresponding specimens. However, the sample size was small, especially for EBUS-TBNA (n = 15). Recently, Sakata et al. compared EBUS-TBNA samples with resected primary tumors in a relatively larger cohort (n = 61) . The concordance rates were 87% with a cutoff of ≧ 1% and 82% with a cutoff of ≧ 50% PD-L1 positivity, respectively. Given that the cutoff of ≧ 50% for PD-L1 positivity resulted in a marked decrease in both sensitivity and positive predictive value of EBUS-TBNA-derived specimens, they concluded that the ≧ 1% PD-L1 positivity was the most appropriate cutoff. The present study corroborated that the concordance of PD-L1 positivity between EBUS-TBNA-derived specimens and corresponding tumor specimens obviously decreased based on increasing cutoff values. This should be largely attributed to the fact that small EBUS-TBNA cores had significantly lower PD-L1 TPSs than resected primary tumors, indicating that PD-L1 expression in EBUS-TBNA samples is apt to be underestimated when the cutoffs are set at higher values such as 5% and 50%. Indeed, Irie et al. reported a 47% discordance rate between 170 pairs of TBB and resected tumor specimens  and claimed that PD-L1 expression in TBB specimens was often underestimated. Nonetheless, PD-L1 expression in EBUS-TBNA specimens was concordant with the corresponding tumors in most cases in our study when the cutoff was defined as ≥ 1%. This result validated the utility of EBUS-TBNA cores for the assessment of tumor PD-L1 positivity when the ≥ 1% of cutoff is applied. Amplification of the chromosome band 9p.24.1, including the loci encoding the PD-L1 and PD-L2 genes, induces PD-L1 expression both as an innate [47,48] and adaptive mechanism . In Hodgkin’s lymphoma, tumors with 9p24.1 CNAs are frequently observed and associated with PD-L1 expression and JAK-STAT signaling . Clinical responses to nivolumab on malignant tumors with 9p24.1 amplification were also reported in several case reports [, , , ]. A recent study analyzed 118,187 tumor samples by next-generation sequencing and showed that PD-L1-amplified tumors tended to show better response to ICIs . Moreover, loss of PD-L1 amplification was reported as an acquired resistance mechanism to nivolumab in a patient with colon cancer . We previously reported the prevalence of PD-L1 copy number gains in NSCLC patients and showed that PD-L1 CNAs were relatively homogenous between resected primary tumors and resected metastatic lymph nodes than PD-L1 protein expression . Evaluation of CNAs using the FISH technique in EBUS-TBNA specimens has been validated for the evaluation of relevant oncogene status such as EGFR and ALK [29,51]. Accordingly, we here showed that EBUS-TBNA samples were well assessed by FISH analysis for the evaluation of PD-L1 copy number and that the concordance of PD-L1 CNAs between EBUS-TBNA-derived and corresponding specimens was comparable with that of PD-L1 expression. In addition, we also showed that PD-L1 CNAs were intratumorally more homogeneous than PD-L1 protein expression using whole tumor sections. It should be noted that there is currently no evidence established by prospective trials showing that PD-L1 CNAs have any potential to predict efficacy of ICIs in patients with NSCLC. However, our results indicate that PD-L1 CNAs could provide useful and reproducible complementary information to predict the efficacy of ICIs, especially in cases where only small biopsy specimens obtained by EBUS-TBNA are available.