Patients with Lung Cancer of Different Racial Backgrounds Harbor Distinct Immune Cell Profiles

Tumors accumulated with infiltrated immune cells (hot tumors) have a higher response rate to immune checkpoint blockade, when compared with those with minimal T-cell infiltration (cold tumors). We report here that patients with lung cancer with different racial backgrounds harbored distinct immune cell profiles in the tumor microenvironment. Compared with African Americans (AA), Caucasian Americans (CA) exhibited increased immune cell infiltration and vasculature, and increased survival. Changes of survival and immune profile were most pronounced among active smokers and nonsmokers, compared with former smokers and total patients. Neighborhood analysis showed that immune cells accumulated around cancer cells in CAs but not AAs. Our findings reveal intrinsic biological differences between AA and CA patients with lung cancer, suggesting that treatment plans should be tailored for patients with different racial backgrounds. Significance: We report biological racial differences among patients with lung cancer where Caucasians present a hot tumor microenvironment compared with cold tumor in AAs. Treatment plans should be customized to maximize therapeutic outcomes.


Introduction
Lung cancer, the leading cause of cancer mortality in the United States accounts for more deaths each year than colon, breast, prostate, and pancreatic can-development in patients across different racial backgrounds, disease stages or inflammatory status.
In this study, we identified intrinsic differences between AA and CA patients with lung cancer. CA patients had better survival compared with AA patients, which may result from increased tissue inflammation and elevated immune populations infiltrated into the tumor in CA patients. More specifically, CA patients had significantly greater numbers of CD4 and CD8 T cells, as well as increased vasculature within the tumor, indicating a more active immune response against tumor progression. Interestingly, we found a striking difference in survival between AA and CA patients, for both active cigarette users and nonsmokers. This observation was consistent with their immune profiles where greater numbers of CD4 and CD8 T cells were detected among active smokers or nonsmokers CAs compared with AAs. Moreover, neighborhood analysis of the tumor microenvironment revealed that cancer cells in CA patients were surrounded by immune cells; while in AA patients, these immune cell populations avoided the cancer cells. These data demonstrate a "hot tumor" microenvironment in CA patients with lung cancer and a "cold tumor" microenvironment in AA patients with lung cancer, which may define different clinical trajectories for these patient populations.

Methods and Materials Patient Cohort
A total of 157 patients with lung cancer were enrolled in Houston Methodist hospital since 2015. All patients provided written informed consent for using their tissues for Institutional Review Board (IRB)-approved research. The study was conducted in accordance with the Declaration of Helsinki and the International Conference on Harmonization Guidelines for Good Clinical Practice. Records of patient information were obtained under approval of IRBs. Among these patients, tissue from 55 patients with paired tumor and adjacent normal tissue and 21 patients with only tumor tissue was stored at the Houston Methodist Biorepository. Patient demographics are shown in Table 1.

RNA Sequencing
Lung adenocarcinomas and lung squamous cell carcinomas and paired adjacent normal frozen tissue samples (15 each subtype) were obtained from the Houston Methodist biorepository under IRB-approved protocol. These included 12 female and 18 male samples with an age range of 54-80 years. Tumor versus normal adjacent tissue designations in hematoxylin and eosin slides were made by a trained pathologist. RNA extracted from fresh frozen tissue by laser capture microdissection was sent to the Sequencing and Gene Editing Core at the University of Houston for: RNA quality analysis using the Agilent 2100 Bioanalyzer; sequencing library preparation using the Illumina Truseq stranded total RNA kit, with 1,000 ng of RNA per sample with a RNA integrity number (RIN) value >8.4; and sequencing using the NextSeq 500 (Illumina). RNA sequencing (RNA-seq) data were mapped onto the human genome UCSC hg19 using hisat2 (18), and gene expression quantified using StringTie and GENCODE model. Differentially expressed genes were detected using unequal variance two-sided Student t test with the Benjamini-Hochberg correction, with significance achieved at FDR < 0.05 and fold change exceeding 1.25 ×. Enriched pathways were inferred using the hypergeometric test, with significance achieved for FDR < 0.05.

Imaging Mass Cytometry
Methods of conjugating metal-labeled antibodies, tissue staining, and data analysis were described previously (19,20). Each antibody was individually  (colored in blue) by cluster Y in Y axis. Change of cluster relationship was identified between AA and CA patients.

Statistical Analysis
Statistical analysis was performed in Prism (Graphpad) except the correlation analysis. P values were calculated by unpaired t test unless stated otherwise in figure legends. For correlation analysis, the trend of increase or decrease from CA and AA patients was calculated, and the P values were calculated in R scripts to determine significance of difference between these trends.

Data Availability Statement
The RNA-seq data underlying this article are available in the NCBI Gene Expression Omnibus at http://www.ncbi.nlm.nih.gov/geo/ and can be accessed with accession number GSE159857.

Differential Gene Expression Between AA and CA Patients with Non-small Cell Lung Cancer
To elucidate differences in tumor biology between AAs and CAs, we performed RNA-seq analysis on tumor and adjacent normal tissues from 30 patients with non-small cell lung cancer (NSCLC) (15 adenocarcinoma and 15 squamous cell carcinoma). After removing samples with low-quality RNA-seq data, the final analysis included 11 tumor tissues for adenocarcinoma and 12 tumor tissues for squamous cell carcinoma. In terms of race distribution, the patients included 5 AAs (2 adenocarcinoma and 3 squamous cell carcinoma) and 18 CAs (9 adenocarcinoma and 9 squamous cell carcinoma).
Significant differences in gene expression profiles between AAs and CAs were identified (

Increased Patient Survival and Immune Cells in CA Patients with NSCLC
To further validate our RNA-seq findings, we obtained information and tissues from a cohort of 157 patients with lung cancer from the Houston Methodist Biorepository. Tissues were obtained from 24 AA and 52 CA patients with lung cancer treated at Houston Methodist Hospitals. AA patients showed poorer overall survival when compared with CA patients (53.97% vs. 64.89%, respectively; Fig. 2A). To determine whether the increased survival in CA patients is associated with increased immune infiltration, as suggested by RNA-seq, we sampled a subgroup of 26 patients (12 AA patients and 14 CA patients at different stages of lung cancer) and performed imaging mass cytometry (IMC) to assess their immune profiles ( Fig. 2B and C). The unsupervised clustering algorithm generated a phenograph heatmap based on marker profiles (Fig. 2C) and identified 38 cell clusters as shown in the tSNE map ( Fig. 2B; Table 2).
An elevated level of total CD45 + cells was identified in CAs (Fig. 2D), denoting increased numbers of immune cells in the tumor microenvironment. This observation is found within patient tumor tissue, but not in the normal adjacent tissue, which has relatively few immune cells ( Supplementary Fig. S1).
suggesting a stronger CD4 T-cell response and a higher potential of incoming CD4 T-cell response in CA patients. Interestingly, there were more immunesuppressive Treg and M2-like macrophages (CD163 + CD68 + ) in CAs, possibly to counter balance the immune response associated with increased CD4 and CD8 T cells. Together, these observations indicate a "hot tumor" microenvironment in CA patients, compared with a "cold tumor" microenvironment in AA patients. In keeping with this interpretation, CA patients exhibited higher numbers of CD31 + endothelial cells, indicating increased vasculature in the "hot tumor" microenvironment. In summary, CA patients with lung cancer had increased immune cells and vasculature present within their tumor microenvironment. This pattern is consistent with the increased survival seen among CAs.

Smoking Contributes to Differences in Survival and Immune Profile Between AA and CA Patients with NSCLC
Cigarette smoking is the major cause of lung cancer. Between 50% and 80% of patients with lung cancer have preexisting COPD, compared with a 15% to 20% prevalence of COPD in the general smoking population (4,21,22), and studies have shown that smokers with COPD are at increased risk for developing lung cancer (2,23). We further stratified the survival data of the 157 patients with lung cancer based on their smoking history (Fig. 3A) Among both current smokers and nonsmokers, we also observed significantly higher numbers of CD45 + cells in CA patients ( Fig. 3B and C). Similarly, total CD4 + T cells, CD8 + T cell, and CD31 + endothelial cells were increased in CA patients who were current smokers and nonsmokers, but not in former   (Fig. 3B). Although not statistically significant, CA patients showed higher abundance of CD14 + monocytes and M2-like macrophages (CD163 + CD68 + ) among current smokers. However, this increase was not observed in former smokers, that is, CA patients had similar numbers of immune cells compared with the AA patients, consistent with similar survival rates. These findings are consistent with survival outcomes in these patient subpopulations ( Supplementary Fig. S2).
Among patients of same race, comparisons between nonsmokers, former smokers and current smokers were also performed. Significant difference was only found in CD45 + cells and CD4 + cells between former smokers and current smokers of CA patients (Fig. 3C). We also investigated other factors that might contribute to differences of survival and immune profile between CA and AA patients, such as sex, tumor type, survival status, and disease stage. However, these factors had minimal to no effect on the immune cell profile (Supplementary Figs. S2-S7). To sum up, CA patients with lung cancer who were active smokers or nonsmokers had better survival and increased intratumor immune cell infiltration compared with AA patients.

Spatial Relationship Between Immune Cells and Cancer Cells Reveals Distinct Tumor Microenvironments Between AA and CA Patients with NSCLC
We next performed neighborhood analysis of the tumor microenvironment to assess the spatial relationship between the cell populations identified (Fig. 4).
Each block represents clusters in the X axis that are surrounded (colored in red) or avoided (colored in blue) by clusters in the Y axis. We divided the data into different groups based on smoking history and compared AA and CA patients. Considering that the white blocks (neither surrounding nor avoidance) could result from low sample numbers in some groups, we looked for changes in immune cell populations between AA and CA patients with opposite relationships. Interestingly, we found that cancer cells (cluster 4, PanCK + Ecadherin + VCAM1 + pNFκB + ) from current smoker AA patients ( . This establishes a "hot tumor" environment within the CA patients and further explains survival differences (Fig. 3A). We did not observe a different relationship between these populations within former smokers (Fig. 3C), a finding consistent with their similarity in survival (Fig. 3A). However, we also did not observe a different relationship between these populations within nonsmokers (Fig. 3C), despite survival differences between AA and CA patients (Fig. 3A), which may be the result of limited sample numbers from nonsmoker AA patients. Nevertheless, our results demonstrate that in sharp contrast to CA patients, immune cells and endothelial cells are not surrounding the tumor cells in AA patients, suggesting a "cold tumor" microenvironment.

Minimal Differences in Tumor Microenvironment Between AA and CA Patients with NSCLC at Early Disease Stages, that Become Significant as Disease Progresses
Next, we asked whether the cell clusters identified correlated with disease stages in a different pattern between AA and CA patients. We divided these 26 patients based on disease stages and plotted the cell density at different disease stages on a log scale. Linear regression fitting each cluster found distinct patterns of correlation between AA and CA patients (Fig. 5). Ki67 + cells, collagen + E-cadherin + cells, CD31 + endothelial cells, CD163 + macrophage, and pSTAT1 + cells were significantly increased in CA patients with advanced disease stages. While Ki67 stains for proliferative cells, with the majority of Ki67 + cells expected to be tumor cells, this relationship only held true for CAs and not AAs (Fig. 5). This difference may be partly accounted for by differences between the hot versus AA patients showed a significant decrease in these populations with increased disease severity. Conversely, CA patients displayed a significantly decreased correlation between pERK + cells and disease severity, while AA patients had an increased correlation between these cells. Consistent with earlier findings establishing a "hot tumor" environment in CA patients and a "cold tumor" environment in AA patients, we found increasing differences of endothelial cells and M2-like macrophages as the disease progressed to late stages (Fig. 5). More interestingly, most of these clusters (5/6) showed similar numbers between AA and CA patients at the earliest stage of disease, with differences becoming more significant with disease progression (Fig. 5). This indicates that AA and CA patients have minimal differences in their tumor environment at early stages.
As the disease progresses, differences were amplified and potentially impact patient survival. In summary, AA and CA patients with lung cancer display distinct correlation patterns between cell populations and disease severity, especially during late stages of disease, which could affect patient survival and treatment outcomes.

Long Interspersed Element-1 Expression Among Patients with NSCLC
During tumorigenesis, immune cells can be recruited and altered by cancer cells to create an immune-suppressive environment. This process is influenced by long interspersed element-1 (LINE-1) for its ability to regulate cell metabolism, tissue inflammation, and extracellular matrix deposition (26)(27)(28)(29)(30)(31). Moreover, LINE-1 is associated with TGFβ1 levels (32), which skews the immune cells toward a suppressive phenotype (33)(34)(35). LINE-1 is the only retrotransposon that remains active in the human genome (36). Retrotransposition is a highly mutagenic process associated with genomic instability, aberrant regulation of DNA repair, altered extracellular matrix deposition, inflammation, and changes in cellular metabolism (37,38). As we identified differences in immune profiles between CA and AA patients, we examined the expression level of LINE-1 ORF1p AACRJournals.org Cancer Res Commun; 2(8) August 2022   by IMC (Fig. 6A). We found more LINE-1 ORF1p + cells in patients with adenocarcinoma compared with patients with squamous carcinoma (Fig. 6B). Within patients with adenocarcinoma, more LINE-1 ORF1p + cells were observed in AA patients compared with CA patients (Fig. 6B), which may potentially contribute the lower survival rate ( Fig. 2A). Interestingly, by dividing these patients based on tumor types, we found significantly more LINE-1 ORF1p + cells in CA patients with squamous carcinomas, but not in CA patients with adenocarcinomas (Fig. 6B). In view of these differences, we conclude that LINE-1 was negatively associated with patient survival, primarily in AA patients with adenocarcinoma.

Discussion
In this study, we established an association between patients' racial background and immune cell profiles in lung cancer, with marked differences in immune cell infiltration and survival found between AA and CA patients with lung cancer.
CA patients, but not AA patients, showed increased immune cell accumulation into the tumor microenvironment. These immune cells surrounded cancer cells only in CA patients suggesting that a "hot" versus "cold" tumor microenvironment contributes to clinical outcome differences between CA and AA patients. Our findings are consistent with several reviews examining racial differences. Nazha and colleagues (39) addressed the underrepresentation of minority ethnic groups in clinical trials and showed that AAs represent less than 4% of all patients enrolled. Kalyanaraman and colleagues (40) proposed that altered mitochondrial oxidative metabolism and differences in the tumor immune microenvironment are critical determinants of poor clinical outcomes among AA patients. Finally, Kakarla and colleagues (41) summarized the impact of racial disparities for different cancer types (prostate, breast, ovarian and uterine, digestive, urinary, and respiratory) and concluded that aggressive forms of disease and overall mortality are higher in AAs compared with other races.
Smoking is known as the leading contributor to lung cancer survival (1,24,25). We found significant differences of survival immune profile between AA and CA patients with different smoking histories (Fig. 3). Interestingly, among patients of the same race, comparisons between nonsmokers, former smokers, and current smokers found significant differences in CD45 + and CD4 + cells between former and current smoker CA patients, but not in AA patients.
This suggests smoking status influences the immune response of CA patients more than AA patients, which is consistent with our findings that CA patients harbor a "hot tumor" environment while AA patients present a "cold tumor" environment. Moreover, within the nonsmoker subgroup, CA patients showed higher survival compared with AA patients (Fig. 3A; Supplementary Fig. S2), indicating there are other factors contributing to disease aggressiveness aside from smoking. For example, the immune profiles are distinct between CA and AA patients within the nonsmoker subgroup (Fig. 3C). Future studies will further examine other contributors to health disparities between CA and AA patients with lung cancer. 2). In our dataset, female AA patients had comparable death rates to female CA patients (37.5% vs. 30.9%), while male AA patients had a higher death rate compared with male CA patients (55.8% vs. 41.0%). These findings are consistent with a previous report (2) and indicate that subgroup differences must be considered as critical elements in the analyses of racial and sex disparities. In this study, we revealed the association between patients' racial backgrounds and the immune cell profiles within lung cancer. However, causal relationships are difficult to establish in human studies. Future investigations of immune cell profiles in patients with the same racial background at varying stages of disease may reveal how immune responses contribute to overall patient survival. Moreover, patients' immunity is likely one of many factors affecting tumor progression. Other biological differences between AA and CA patients may be identified in the RNA-seq data, which can be studied more extensively in preclinical models.
Our study had two limitations that must be considered. First, larger sample sizes, especially for AA patients, must be examined in future studies. This is exemplified by the high variability of RNA-seq expression profiles in CAs compared with AAs and the lack of statistical significance noted when complex IMC profiles were stratified for race, sex, and disease stage. Second, the self-designations of race used in our study were not confirmed with ancestry admixture markers for a more accurate estimation of ancestry proportions in admixed U.S. populations.

AACRJournals.org
Cancer Res Commun; 2(8) August 2022 In summary, our findings highlight key differences in immune cell profiles associated with the patients' racial background in response to lung cancer. Therefore, future treatment plans may need to be customized to maximize the therapeutic outcome among different racial and ethnic groups.