Molecular signatures for inflammation vary across cancer types and correlate significantly with tumor stage, gender and vital status of patients

Cancer affects millions of individuals worldwide. One shortcoming of traditional cancer classification systems is that, even for tumors affecting a single organ, there is significant molecular heterogeneity. Precise molecular classification of tumors could be beneficial in personalizing patients’ therapy and predicting prognosis. To this end, here we propose to use molecular signatures to further refine cancer classification. Molecular signatures are collections of genes characterizing particular cell types, tissues or disease. Signatures can be used to interpret expression profiles from heterogeneous samples. Large collections of gene signatures have previously been cataloged in the MSigDB database. We have developed a web-based Signature Visualization Tool (SaVanT) to display signature scores in user-generated expression data. Here we have undertaken a systematic analysis of correlations between inflammatory signatures and cancer samples, to test whether inflammation can differentiate cancer types. Inflammatory response signatures were obtained from MsigDB and SaVanT and a signature score was computed for samples associated with 7 different cancer types. We first identified types of cancers that had high inflammation levels as measured by these signatures. The correlation between signature scores and metadata of these patients (gender, age at initial cancer diagnosis, cancer stage, and vital status) was then computed. We sought to evaluate correlations between inflammation with other clinical parameters and identified four cancer types that had statistically significant association (p-value < 0.05) with at least one clinical characteristic: pancreas adenocarcinoma (PAAD), cholangiocarcinoma (CHOL), kidney chromophobe (KICH), and uveal melanoma (UVM). These results may allow future studies to use these approaches to further refine cancer subtyping and ultimately treatment.

38 studies to use these approaches to further refine cancer subtyping and ultimately 39 treatment.

41 Introduction
42 Cancer is a major public health problem with high mortality rates in the United States and 43 worldwide and poses an enormous burden to individuals and society. Over 1.7 million 44 newly diagnosed cancer cases and over 600,000 cancer deaths were estimated in the 45 United States in 2018 (1). Screening for some cancers can lead to early detection 46 (mammography for breast cancer and colonoscopy for colon cancer, for example), when 47 local resection or definitive treatment may still be feasible (2). However, many cancers 48 are found when there is already local invasion or even distant metastatic disease. In those 49 cases, common treatment options include chemotherapy, locoregional therapies and 50 radiation treatment (3). Among the issues complicating treatment options are the fact that 51 there are many tumor types, whose response to therapy may differ depending on site of 52 origin and cellular composition (4). Furthermore, even within the same organ, there are 53 heterogeneous tumor types with different responses to therapies.

54
55 As a result, precise tumor classification is crucial; depending on the categorization of a 56 tumor, the clinical course, prognosis, and treatment can vary dramatically (5). In general, 57 there are two ways to classify cancer: the traditional histology-based method and 58 molecular methods. The traditional method is based on observing the site of origin, 59 degree of spread and cellular morphology, while the molecular method identifies gene 60 expression and genetic profiles (6)(7)(8) 166 Results

168
Tumor Types

169
Hierarchical clustering was performed to group cancer subtypes by inflammatory 170 signature scores, and the three subgroups were determined by the dendrogram structure 171 resulting from the hierarchical clustering (Figure 1). Of the tumor types evaluated in our 172 study, we found that tumors in areas exposed to airways or gastrointestinal tracts,

173
including pancreatic, lung and esophageal cancer, tended to be more inflammatory.

174
Based on these results we selected seven tumor types with varying levels of inflammation

244
Once we compared the patient data to the inflammatory signatures, we found three 245 distinct groups of cancers: (1) those with high inflammation, (2) those with low 246 inflammation, and (3) those with both high and low levels of inflammation. We found 247 PAAD to be a member of the high inflammation group. This grouping is supported by 248 multiple studies associating pancreatic inflammation (pancreatitis) with the development 249 of pancreatic cancer (25, 32). One of the cancers we found to be in the low inflammation 250 group was UVM. Melanomas are associated with environmental insults, such as 251 exposure to ultraviolet light. As such, we expect that inflammation is not necessarily 252 involved in the mechanism responsible for the development of skin cancer.

254
We believe that the gene expression data for these tumor types is heterogeneous across 255 individuals, with multiple subgroups of patients per type. As such, the inflammatory 256 signature presented in this first analysis is averaged across individuals, and that within these broad categories there may be subgroups of patients with high inflammation and 258 others with low inflammation. This suggests that patients with these cancers could 259 potentially benefit from further molecular subclassification.

261
In addition to correlating levels of inflammation with specific cancer types, we utilized

314
In summary, our study evaluated the association between inflammation signatures for 315 different tumor types. We found associations between levels of inflammation and tumor 316 types, and also found statistically significant relationships between patient metadata and 317 inflammation for four tumor types. We believe our results demonstrate potential clinical