Long-term follow up of NSCLC patients treated with cetuximab plus avelumab in the CAVE-lung trial
At 18-month longer follow-up analysis (November 30, 2021 vs April 15, 2020) [5], 3/16 patients were alive with one of these patients still on treatment (see Supplementary Fig. 1 for details). Of interest, among patients with the longest survival, 3 patients had received a previous line of therapy with single agent anti-PD-1 mAb (one patient, pembrolizumab; two patients, nivolumab); thus, suggesting clinically significant anti-tumor activity of this experimental treatment following progression to single agent immune checkpoint inhibitor therapy. These 3 patients had initial clinical benefit to previous anti-PD-1 therapy, with partial response (PR) as best radiological response in one patient and stable disease (SD) in the other two patients. PFS was 6, 7 and 7 months, respectively (Fig. 1A). Baseline tumor PD-L1 protein expression was 60, 5% and unknown, respectively. Cetuximab plus avelumab treatment obtained SD as best radiologic response in these patients with PFS of 15, 19, and 34+ months, respectively.
We have previously reported that clinical benefit following cetuximab plus avelumab treatment was accompanied by NK cells activation in the CAVE-Lung responding patients, as compared to non responders group of patients [5]. We have extended the analysis of NK cell activation in the PBMC, as assessed by LDH release cytotoxicity assay, with serial measurements during treatment at 10, 12, and 15 months, respectively, for these patients (Fig. 1A). For the patient with the longest response (34 months, treatment ongoing), LDH levels were constantly higher as compared to baseline (increase of 20%, p < 0.05) while in the other two patients LDH levels were high until clinical benefit was maintained (increase of 12 and 10% respectively, p < 0.05) and started to decline at the time (15 months) of progression of disease (PD) or close to PD (Fig. 1B). To further evaluate the immune effects that could be determined by cetuximab plus avelumab treatment in these patients, we analyzed by FACS the modifications in immune cells population prevalence in the PBMC. As shown in Fig. 1C, we observed an increase of CD107A+ cells, a specific marker of NK cell degranulation, at the time of clinical response: in particular, in patient #09, the one with the longest PFS, CD107A+ cells increased of 32% (p < 0.01) at 15 weeks and in patient #02 of 16% (p < 0.05). In the other patient #11 at the last time point, that was close to the time of radiological PD, we detected a not significant change CD107A+ cells (3%,p not significant). Also, we explored changes in TIM3+ and PD-L1+ immune-suppressive cells (Fig. 1 D), and we found a significant decrease of these two sub-populations at the time of clinical response in patient #09 (decrease rate of 14 and 26%, p < 0.01). Conversely, TIM3+ and PD-L1+ cells raised up at the time of PD in the other two patients (patient #02, increase rate of 14%, p < 0.01 and 12%, p < 0.05; patient #11, increase rate of 6 and 7%, p not significant).
Mutation profiling in ctDNA, as assessed by liquid biopsy, detected somatic mutations in DDR genes, that have been associated with a phenotype of immune responsiveness and of innate immunity activation [11]. Specifically, one patient presented STK11 and TP53 mutations, one patient had CHK2, ATM and BRCA2 mutations, while the third patient showed MDM2 mutation (Fig. 1E).
Recent genetic studies suggest that the host’s genetic background contributes to cancer immunity and rare variants (Minor Allele Frequency, MAF < 1%) that are functionally deleterious have large effect size than common variants. To unveil the inherited casual variants which could influence anti-tumor immune response we performed Whole Exome Sequencing (WES) analysis in PBMCs from these 3 patients and focused on rare and predicted deleterious variants [17]. Interestingly, rare variant in POLE family genes was found in two patients, whereas it was observed in ATR gene in one patient. Moreover, rare variants in CDC27 gene were found in all three cases. Of note, POLE, ATR and CDC27 genes are implicated in DDR response and in DNA replication stress (Fig. 1E).
TCGA analysis
We have previously shown that activation of the STING pathway, as suggested by increased gene expression for CCL5 and CXCL10, two specific STING effector chemokines, identifies a subgroup of immune checkpoint inhibitor responsive NSCLC patients [11]. Here, we have investigated a potential connection between STING pathway and NK cell activation, by correlating gene expression of CCL5 and CXCL10 with the expression of NK related genes in the TCGA NSCLC dataset. In this respect, CCL5 gene expression highly correlated with PRF1, which encodes for perforin, a marker of NK cell activation (Spearman Rho = 0.76 and 0.828 in lung adeno-carcinoma and squamous, respectively; p < 0.001) and with NK cell receptor NKG7 (Spearman Rho = 0.908 and 0.859 in lung adeno-carcinoma and squamous, respectively; p < 0.001); a moderate correlation was detected for the NK receptor FCGR3A (Spearman Rho> 0.5; p < 0.001) (Fig. 2A). Similar results were obtained for CXCL10 (Spearman Rho> 0.5; p < 0.001, data not shown). The presence of DDR gene mutations have also been correlated to expression of specific immune-active transcriptomic signatures in multiple tumor types [11]. POLE mutations have been correlated with response to immune checkpoint inhibitors [18]. Preclinical studies have also shown that DDR inhibitors could mediate activation of innate immune pathways [10, 12]. Therefore, we next investigated the TCGA NSCLC dataset for associations between STING pathway genes and DDR gene mutations. As shown in Fig. 2B, differential gene expression between POLE mutant and POLE wild-type tumors was assessed. The STING activated genes, CCL5 and CXCL10, were significantly higher in POLE mutant tumors (FC = 1.72; P = 0.02 and FC = 1.89; P = 0.01, respectively); thus, suggesting that POLE mutation could induce intrinsic STING activation. To further extend this observation, we also compared the differential expression of STING/immune signature genes, previously [11] correlated to immune-responsiveness, between DDR-mutant and DDR wild-type NSCLC samples from the TCGA lung adenocarcinoma cohort. DDR-mutant NSCLC were defined according the presence of mutation in one of DDR genes of a curated list (for the DDR gene list, see Supplementary Table 1). Of interest, the presence of mutation in at least one DDR gene, including a variety of tumors with potential heterogenous genomic landscape, occurs in NSCLC displaying features of immune-responsiveness, as indicated by gene expression of immune-genes, as listed in Fig. 2C. (Fig. 2C) [11].
Anti-tumor activity and immune effects of cetuximab plus avelumab treatment in NSCLC patient-derived ex vivo 3D cultures
In vitro spheroids, that were obtained from the 3 NSCLC patients tumor samples, as previously described [15, 16], were treated with cetuximab plus avelumab for 7 days. As shown in Fig. 3A, both single agent cetuximab and avelumab determined approximately 15 to 30% growth inhibition in all cases. An additive anti-tumor activity (40 to 50% growth inhibition) was observed with the combined treatment (chi-square: 37,503, p < 0.05). The reduction in spheroid number and viability, that was caused by cetuximab plus avelumab treatment, was partially reverted by combined treatment with an anti-CD16 blocking mAb, that inhibited NK cell activation; thus, suggesting that NK cells are involved in the mechanism(s) of cytotoxicity of cetuximab plus avelumab. Figure 3B shows exemplificative qualitative images from each treatment point.
To further assess potential effects of treatment with cetuximab and/or avelumab on cytokines that could be involved in NK cell activation, their gene expression was measured by quantitative PCR of RNA, that was obtained from in vitro 3D spheroid cultures. As shown in Fig. 4A, following cetuximab plus avelumab treatment a significant 10- and 20-fold increase was detected for two STING downstream effector chemokines, CCL5 and CXCL10, respectively, as compared to untreated control samples. Moreover, a 20-fold increase in IFNβ gene expression was observed (p < 0.01). Similarly, a 10-fold increase in the gene expression for biomarkers of NK activation, such as of Granzyme B and Perforin, was induced by cetuximab plus avelumab treatment, as compared to untreated control samples (Fig. 4B) (p < 0.01). In parallel, decreased levels of PD-L1 and TIM-3 gene expression were found (up to 60% of decrease with combination, p < 0.01) suggesting also CD8+ T cell activation, as part of the anti-tumor immune response, that was induced by cetuximab plus avelumab treatment (Fig. 4C).