NEW YORK – Researchers have uncovered three subtypes of squamous cell lung cancer with varying proteomic, genomic, and other alterations that point to therapeutic pathways to investigate.
Previous subtyping approaches for squamous cell lung cancer have relied on genomic or transcriptomic alterations. But as the Moffitt Cancer Center's Eric Haura noted, there can be a disconnect between the genomes or transcriptomes of cancers and their expressed proteomes. By relying on proteomic as well as genomic data, Haura said he and his colleagues could get a more expansive view of the tumors.
By incorporating shotgun proteomic data into their analyses, researchers led by Haura identified three subtypes of squamous cell lung cancer that they dubbed inflamed, redox, and mixed, as they reported in Nature Communications last week. These subtypes, they found, also hinted that, for instance, the inflamed subtype might respond to therapy targeting CD33+ myeloid cells combined with immune checkpoint therapy, while targeting metabolic pathways in the redox subtype might be a potential therapeutic approach.
"This is, right now, the largest proteogenomic study of lung cancer," Haura, the director of the Lung Cancer Center of Excellence at Moffitt, said.
He and his colleagues characterized 108 squamous cell lung cancer samples, using a combination of copy number alteration, DNA mutation, RNA, and protein expression analysis. Their proteomic analysis — using LC-MS/MS with tandem mass tag labeling — of the samples uncovered five clusters with differing protein expression patterns.
Two clusters exhibited immune biology-related protein expression and were combined into an "inflamed" subtype that encompasses 40 percent of the overall tumors, while two other clusters both showed oxidation-reduction biology-related protein expression and were also combined into a "redox" subtype that includes 47 percent of the samples. The last cluster, which exhibited biology related to Wnt/stromal biology, was dubbed "mixed" and includes 13 percent of the samples.
Both clusters making up the inflamed subtype had significantly higher expression of neutrophil-associated proteins. In particular, inflamed A had high expression of MPO, DEFA1, and DEFA3, among others. While inflamed B had similarly high expression of neutrophil-associated proteins, it also had highly inflamed inflammatory response proteins, including 11 MHC Class II proteins and nine cathepsins. This, Haura noted, is consistent with antigen presentation biology.
Meanwhile, the two clusters making up the redox subtype both had higher expression levels of aldo-keto reductase and alcohol dehydrogenase family members.
He and his colleagues still found a degree of overlap between the subtypes they teased out from proteomic data and previously described classification schemes. For instance, one approach from Wilkerson and colleagues, which was reported in Clinical Cancer Research in 2010, divides squamous cell lung cancer into four groups — primitive, classical, secretory, and basal — based upon their gene expression signatures. When Haura and his colleagues classified their samples based upon this scheme, 44 percent belong to the classical group, 28 percent to secretory, 19 percent to basal, and 9 percent to primitive. Their inflamed subgroup was a mix of secretory and basal subtypes from this classification scheme, while their redox subgroup was mostly the classical group, and their mixed subgroup was, again, a mix of basal, secretory, and primitive tumors.
But these new proteomic-based subtypes also offered not only new insights into tumor biology, they also highlighted two potential treatment avenues to be explored.
For the redox subtype samples, the researchers compiled a list of proteins whose expression was also elevated at the transcriptomic level or copy number level to identify potentially overexpressed genes or amplified targets. They intersected that list with RNA interference and CRISPR screen databases. Overlapping between the two sets were proteins — PSAT1, TP63, and TFRC — involved in metabolic signaling, indicating that metabolic signaling could be a vulnerability to be targeted for treatment in this tumor subtype. He added that previous work using high-throughput drug screening has also suggested there might be metabolic vulnerabilities in squamous cell lung cancer.
"I think that this should push the field to be thinking about metabolic vulnerabilities in squamous cell," he said.
At the same time, the researchers found the inflamed subtype samples were enriched for infiltrating immune cells. Using RNA-seq data, they reported that the inflamed subtype samples had a higher proportion of memory B-cells, monocytes, and neutrophils than the other subtypes.
This, Haura noted, represents a slight divergence from the previous classification scheme. Previously, he noted that a cell adhesion subtype had been uncovered based on transcriptomic analysis. But here, they found that that the cell adhesion subtype might not reflect tumor cell adhesion, but rather might reflect the adhesion proteins and properties of neutrophils that are highly infiltrating in these tumors. "The proteomics data made us clearly see this subtype," he said.
These infiltrating neutrophils and other immune cells also represent potential treatment targets. The infiltration of these myeloid cells is likely to be a problem for either checkpoint therapies or other therapies to try to enhance T cell function, Haura said. But as CD33 is a marker of immature myeloid cells, Haura noted that eliminating cells with that marker could then enhance such a therapeutic approach, adding that he and his colleagues are exploring CD33+ targeting strategies.
When the researchers dug deeper into infiltrating subtype and reexamined those tumors under the microscope, they noticed they tended to harbor tertiary lymph node structures, which have been identified in other tumor types. While none of the three subtypes the researchers uncovered were associated with differences in progression-free or overall survival, the researchers did find that these B-cell-rich lymph node structures were associated with improved prognosis in these patients.
Since B cells make antibodies, Haura said they are interested in determining what antibodies these B cells are making.
"We're in the early days of trying to understand what these B cells are doing in these tumors," he said. "We're thinking about how to target these immature myeloid cells and these tumors for therapeutic intent."