Functional immune characterization of HIV-associated non-small-cell lung cancer

Dear Editor,

In the combined anti-retroviral therapy (cART) era, non-small-cell lung cancer (NSCLC) is a highly incident cause of morbidity and mortality in people living with HIV (PLHIV) [1]. The immune-pathogenesis of NSCLC and HIV infection both rely on programmed-death 1 (PD-1) receptor–ligand interaction as a mechanism to induce T-cell exhaustion. To date, PLHIV have been excluded from clinical trials of immune-checkpoint inhibitors (ICPI), on the presumption that antitumour immunity might be compromised by HIV infection. To verify this, we evaluated the clinico-pathological significance of PD-ligands expression in a consecutive series of 221 archival NSCLC samples, 24 of which were HIV-associated (supplementary Table S1, available at Annals of Oncology online).

Most patients with HIV-associated NSCLC were active smokers (n = 18, 75%; median 40 packs/year, IQR 48), established on cART (n = 22, 92%) for a median duration of 9.7 years (IQR 7.1) with CD4 counts >250 cells/mm³ (n = 19, 79%) and suppressed HIV RNA (n = 20, 80%). Molecular profiling data (n = 12/24) revealed two EGFR mutation carriers. Tissue-microarray sections underwent PD-L1, PD-L2 immunostaining and multiplex immunohistochemistry (n = 21) for PD-1, CD4, CD8 and CD68 (Figure 1A–D, supplementary materials, available at Annals of Oncology online). Prevalence of PD-L1 positivity was 45% in tumour (n = 11/24) and 8% (n = 2/24) in tumour-infiltrating cells, whereas tumoural PD-L2 positivity was found in 33% (n = 8/24) (supplementary Figures S1, available at Annals of Oncology online). Following 2 : 1 case–control matching for age, gender, grade, stage and histotype (n = 66), PD-L1 (12/23 versus14/43, P = 0.12) and PD-L2 expression (9/23 versus 25/45, P = 0.20) were unrelated to the presence and severity of HIV-related immune dysfunction as indicated by nadir CD4 counts (P = 1.0, P = 0.8), HIV RNA levels (P = 0.14, P = 0.24), duration of HIV infection and cART (P > 0.05) suggesting independence between antiviral and antitumour immune-tolerogenesis. In keeping with this view, we documented a positive correlation between PD-L1 expression and density of tumour-infiltrating (TIL) (Figure 1A–F) but not peripherally circulating lymphocytes (Figure 1G and H), highlighting polarization of the tumour microenvironment to a type-1 response [2]. In type-1 or PD-L1⁺/TIL⁺ tumours, TILs are chemo-attracted to malignant cells and turned off by PD-L1 engagement. Type-1 tumours are generally sensitive to single-agent PD-1/PD-L1-targeted checkpoint blockade, due to the presence of an immune-reactive microenvironment [3].

Figure 1.

(A–D) Representative sections of HIV-associated NSCLC co-immunostained for CD8 (red chromogen), CD4 (brown chromogen), PD-1 (blue chromogen) and CD68 (green chromogen) illustrating CD4/CD8 enrichment in PD-L1-positive HIV-associated NSCLC (A–C) compared with PD-L1-negative counterparts (B–D). Histograms of multiplex immunohistochemistry data illustrating the relationship between the CD4- and CD8-positive immune infiltrate expressed as number of immunopositive cells per high power field (HPF) according to tumour cell PD-L1 (E) and PD-L2 expression status (F) in HIV-associated NSCLC (n = 21). Peripheral blood immunophenotyping illustrates the relationship between peripheral CD4, CD8, CD19 and CD56-positive cell counts categorized according to tumoural PD-L1 (G) and PD-L2 expression status (H) in HIV-associated NSCLC (n = 24). Medians and interquartile ranges are reported. (I) Volcano plot of differentially regulated genes identified by Nanostring analysis. The Benjamini–Hockberg P-values are correlated to fold-changes in transcripts identified in HIV-positive NSCLC (n = 5) versus HIV-negative controls (n = 3). Transcripts achieving statistical significance (FDR q-value of 5%) are highlighted by the presence of the corresponding gene name. (J) Heat map profile of the 12 transcripts that are differentially regulated in HIV-associated NSCLC compared with controls. Expression levels are shown as log transformed values of normalized counts (ln2).

In an exploratory targeted transcriptomic analysis of five HIV-positive cases and three stage/histotype-matched controls using the NanoString PanCancer-Immune panel (NanoString Technologies, Seattle), we demonstrated enrichment of transcripts involved in chemotaxis (CCL18), antigen presentation (HLA-A, HLA-DRA), cytotoxic T-cell (LAMP-1) and macrophage activation (SPP1, Figure 1I and J) in HIV-associated NSCLC, consistent with previous studies [4]. Whilst not exhaustive and limited by sample size, our findings challenge the conception that well-controlled HIV infection may negatively influence cancer-specific immune dysfunction by preliminarily demonstrating the presence of an immune-reactive microenvironment in HIV-associated NSCLC (supplementary Figures S2, S3, Tables S2 and S3, available at Annals of Oncology online). This concept should be further explored by comparing normal and neoplastic lung tissues in prospective studies.

As accumulating evidence shows PD-1/PD-L1-targeted ICPI to be safe and capable of enhancing HIV-specific immunity [5] without unexpected toxicity, taken together, our data suggest that HIV-associated NSCLC may be equally reactive to ICPI. Whilst preliminary in nature and warranting validation in larger patient cohorts treated with ICPI, our study supports an immunobiological rationale for the development of PD-1/PD-L1-targeted ICPI in HIV-associated NSCLC.