Assessment of Radiation Induced Therapeutic Effect and Cytotoxicity in Cancer Patients Based on Transcriptomic Profiling

Toxicity induced by radiation therapy is a curse for cancer patients undergoing treatment. It is imperative to understand and define an ideal condition where the positive effects notably outweigh the negative. We used a microarray meta-analysis approach to measure global gene-expression before and after radiation exposure. Bioinformatic tools were used for pathways, network, gene ontology and toxicity related studies. We found 429 differentially expressed genes at fold change >2 and p-value <0.05. The most significantly upregulated genes were synuclein alpha (SNCA), carbonic anhydrase I (CA1), X-linked Kx blood group (XK), glycophorin A and B (GYPA and GYPB), and hemogen (HEMGN), while downregulated ones were membrane-spanning 4-domains, subfamily A member 1 (MS4A1), immunoglobulin heavy constant mu (IGHM), chemokine (C-C motif) receptor 7 (CCR7), BTB and CNC homology 1 transcription factor 2 (BACH2), and B-cell CLL/lymphoma 11B (BCL11B). Pathway analysis revealed calcium-induced T lymphocyte apoptosis and the role of nuclear factor of activated T-cells (NFAT) in regulation of the immune response as the most inhibited pathways, while apoptosis signaling was significantly activated. Most of the normal biofunctions were significantly decreased while cell death and survival process were activated. Gene ontology enrichment analysis revealed the immune system process as the most overrepresented group under the biological process category. Toxicity function analysis identified liver, kidney and heart to be the most affected organs during and after radiation therapy. The identified biomarkers and alterations in molecular pathways induced by radiation therapy should be further investigated to reduce the cytotoxicity and development of fatigue.


Introduction
Regardless of its potential hazards, curative benefits of radiation have been reported in medicine, oncology in particular. It is routinely used in cancer treatment prior to surgery to help shrink the tumor, as a palliative therapy to relieve pain, pressure and other neurologic or obstructive symptoms, and post-surgery to kill any remaining cancer cells and to prevent tumor recurrence or in synergy

Pathways and Networks Underlying Immune Dysfunction
Molecular pathway analysis of radiation treatment associated genes has predicted biofunctions and molecular networks which may be involved in the fatigue and cytotoxicity. Biofunctions including cell-mediated immune response, cell signaling, mineral metabolism, cellular function and maintenance were predicted to be significantly decreased while cell death and survival process were significantly activated (p-value = 1.76ˆ10´7). Transcriptomic signatures displayed noteworthy disturbances in signaling pathways like calcium-induced T lymphocyte apoptosis (z-score =´3.31, Figure 2), and role of NFAT in regulation of the immune response (z-score =´3.77) were inhibited while apoptosis signaling (z-score = 2.23) and cytotoxic T lymphocyte-mediated apoptosis of target cells (z-score = 1.342) were predicted to be activated ( Table 2). Cell death and survival process was predicted to be significantly increased/activated (cell death of immune cells, p-value = 1.76ˆ10´7, z-score = 2.209; cell death of lymphocytes, p-value = 3.84ˆ10´5, z-score = 2.017; apoptosis of leukocytes, p-value = 5.72ˆ10´4, z-score = 2.424 and apoptosis of hematopoietic cell lines, p-value = 3.16ˆ10´3, z-score = 2.489) ( Figure 3). However, most of biofunctions were predicted to be decreased like cell-mediated immune response (T cell development, -homeostasis and -migration, p-value = 8.45ˆ10´1 1 , z-score =´2.867), cell signaling, and mineral metabolism (accumulation of Ca 2+ , mobilization of Ca 2+ , p-value = 5.72ˆ10´4, z-score =´2.764), cellular function and maintenance (homeostasis of leukocytes, cellular homeostasis, p-value = 3.88ˆ10´1 1 , z-score =´3.122) (Supplementary Table S2).  A Gene Ontology (GO) enrichment analysis was performed on those 429 differentially expressed probe sets that intersect on the basis of a p-value <0.05 and FC >2 between baseline and endpoint treatment groups. (Supplementary Table S3). The GO enrichment diagram illustrates functional groups that are significantly overrepresented in different categories ( Figure 4). The most significantly overrepresented groups in the categories cellular component, molecular function, and biological process were "cell part", "nucleic acid binding transcription factor activity", and "immune system process", respectively.     A Gene Ontology (GO) enrichment analysis was performed on those 429 differentially expressed probe sets that intersect on the basis of a p-value <0.05 and FC >2 between baseline and endpoint treatment groups. (Supplementary Table S3). The GO enrichment diagram illustrates functional groups that are significantly overrepresented in different categories (Figure 4). The most significantly overrepresented groups in the categories cellular component, molecular function, and biological process were "cell part", "nucleic acid binding transcription factor activity", and "immune system process", respectively.    . Gene Ontology of differentially expressed genes. Pie Chart obtained using Partek GS 6.6 software shows the change in the biological process of the immune system, biological adhesion, response to stimulus, multi-organism process, biological regulation and developmental process as a significantly affected process.

Discussion
Up to 90% of cancer patients treated with radiation experience cumulative fatigue that is pervasive and affects quality of life [23]. Cancer treatment elicits inflammatory and stress response, mitochondrial function impairment, endocrine dysfunction and immune dysregulation. This induces a cascade of biological changes which get translated into cancer-related fatigue (CRF) manifested with alteration in skeletal muscle function contributing to physical disability along with cognitive and behavioral symptoms. The above-mentioned fatigue has a multifactorial etiology and is reported to be associated with neuroendocrine, metabolic, immune/inflammatory, and genetic biomarkers whose systemic identification will help to understand its etiology and develop better therapies to lessen CRF burden [24]. Though, fatigue felt by the patients might also be due to conditions like anaemia, thyroid dysfunction, pain, depression and sleep deprivation [25].
In our results, α-synuclein (SNCA) (p-value = 2.70ˆ10´6 and fold change = 3.5) was one of the differentially expressed gene identified in the microarray meta-analysis, which was validated [5] and reported to be associated with neuroinflammation [26]. SNCA is primarily involved in diseases like Parkinson's and hereditary amyloidosis. During localized radiation therapy, its upregulation activates inflammatory pathways and signals the associated cancer-related fatigue [5]. We also found upregulation of carbonic anhydrase I (CA1) in our results which is in accordance with the previously published reports [5]. CA1 is a cytosolic protein and belongs to the large family of zinc metalloenzymes. It is found at the highest level in erythrocytes. It functions by catalyzing the reversible hydration of carbon dioxide and participates in a variety of biological processes like respiration, maintenance of acid-base balance, bone resorption and calcification [27].
One of the most down regulated genes in our study was membrane-spanning 4-domains, subfamily A, member 1 (MS4A1). It encodes for CD20 protein, a B1 cell-surface antigen of human B lymphocytes and plays a role in hematopoietic cell activation [28]. Lower MS4A1 gene expression levels will possibly contribute to the lower CD20 protein expression [29]. Association of MS4A1 down regulation and fatigue intensification has been recently reported by Hsiao et al., suggesting that fatigue during RT may be related to impairment in B-cell immune response [30]. MS4A1 was included in molecular signatures that reflects a response to radiation in mice and humans [31]. Reportedly, MS4A1 has indirect interaction with CA1 and SNCA genes associated with cellular movement and immune cell trafficking [30]. Our gene ontology enrichment analysis revealed immune system process as the most significantly overrepresented group under biological process category that has been reported to be positively associated with CRF [32].
We found that apoptosis signaling and cytotoxic T lymphocyte-mediated apoptosis of target cells pathways were strongly associated with RT. Radiation induced stress activates p53-dependent apoptosis marked by increased formation of intracellular reactive oxygen species (ROS) and activation of stress responsive pathways like p38MAPK [33]. Radiation acts as an apoptotic stimuli and cause changes in the inner mitochondrial membrane permeability [34]. This leads to release of regulatory proteins such as cytochrome c which binds and activates Apaf-1, forming an "apoptosome" [35,36]. ATP activates apoptosome complex, which thereby activates procaspase-9. Smac/DIABLO and the serine protease, and HtrA2/Omi promote apoptosis by inhibiting IAP (inhibitors of apoptosis proteins) activity. During apoptosis, the mitochondria releases pro-apoptotic proteins like AIF (apoptosis-inducing factor), endonuclease G and CAD (caspase-activated DNase) [34].
Toxicity functional analysis revealed that radiation therapy does carry a huge price, i.e., at the cost of hepato-, cardio-, and nephrotoxicity. Cases of radiation-induced liver diseases are becoming frequent. Radioembolization may affect the normal liver parenchyma and produce pertinent toxic effects like cholecystitis, gastrointestinal ulceration, pneumonitis, and liver toxicity as a result of radiation of non-target organs [37]. Long-term radiation induced damage in different tissues possibly is a consequence of injury to microvascular endothelial cells causing their apoptosis. Irradiation causes oxidative damage to DNA (both mitochondrial and nuclear) and significantly depletes mitochondrial glutathione which further enhances in vitro as well as in vivo toxicity levels [38].
Accumulating evidences indicate that radiotherapy involving the heart can result in premature ischemic heart disease in cancer survivors especially in those who had concurrently received chemotherapy specifically with doxorubicin (adriamycin, a drug that can cause heart muscle damage). Cardiac complication and damage can manifest even years after high-dose radiation treatment [39]. Severe RT induced coronary artery disease complications include pericarditis, myocardial fibrosis (scarring), stenosis (narrowing), angina and extensive blockage, valvular injury and myocardial infarction [40,41]. The relative risk of death from a fatal myocardial infarction increased from 1.5 to 3.0 times in patients who have received mediastinal RT as compared to those who have not [42].
Radiation induced nephritis is a degenerative inflammatory disease affecting kidneys after exposure to radioactive substances or body irradiation. Administration of radiometal-labelled peptide conjugates or combined high dose chemotherapy and RT in stem cell transplantation reportedly increases nephrotoxicity [43]. Radiation exposure causes renal endothelial damage and other kidney diseases and so renal shielding during total body irradiation is perhaps protective [44].
There are some factors accounting for rare adverse radiation reactions. In few cases, radiation sensitivity can be credited to particular genetic mutations and includes autosomal recessive uncommon diseases like ataxia telangiectasia (AT) [45], AT-like disorder [46], Nijmegen breakage syndrome [47], and radiosensitivity with severe combined immunodeficiency [48]. Heterozygosity for mutations in ATM, the gene mutated in AT, may occur in 1% of individuals and has been reported to confer moderate sensitivity to radiation in tissue culture based studies [49]. In our differentially expressed gene list, we found ATM to be down-regulated (fold change =´2.06843, p-value = 1.73ˆ10´7). A small number of adverse radiation reactions are linked with ATM mutations [50][51][52].
There are some scientific studies hinting towards safe use of alternative medicine like guarana, pineal hormone melatonin, curcumin, amifostine, etc. The extract of guarana, a highly caffeinated plant, was used to ease cancer related fatigue at a low cost [53,54]. Evidences had shown pineal hormone melatonin as radioprotective and can be used to reduce the oxidative injuries due to its free hydroxyl radical scavenging capacity [55][56][57]. Curcumin also has a radioprotection effect due to its capability to decrease oxidative stress and inflammatory responses, and also confers radiosensitization perhaps via the upregulation of genes responsible for apoptosis [58]. Multiple studies support the intrarectal application of amifostine (WR-2721/WR-1065), a phosphorothioate during external beam RT for prostate cancer for the prevention of radiation-induced rectal injury [59][60][61].
Radiations are an omnipresent stress to which all life forms are incessantly exposed in the environment. The more we turn towards nuclear power, the greater the concern for accidents, occupational risks or acts of radiological or nuclear terrorism [62]. Presently, the urban population's exposure to radiation via medical diagnostics and other applications goes beyond their exposure to natural background radiation. Radiation is a double-edged sword: irradiation-induced DNA damage can halt cancer cell proliferation, but collateral radiation damage to adjacent tissues is always a concern. Despite its potential dangers (can even induce tumors and burn skin), the utility of localized radiation in medicine has fueled research and studies focusing on safety. There have been modern advancements in precision technology which lessens the radiation exposure to the healthy tissue, and, therefore, short radiation sessions with escalating doses are feasible for curative local radiation surgery especially in the case of oligometastases [63]. We need drugs (radioprotectors and radiosensitisers) that can protect normal cells while leaving malignant ones susceptible, and ideally, even sensitized to radiation therapy.

Patients and Samples
We retrieved global expression dataset for radiation toxicity studies from NCBI's GEO database. Dataset were clearly divided into two groups: radiotherapy on cancer patients and radiation studies on healthy individuals or cell lines or mouse models. We focused on actual cancer cases over animal models or cell lines and provided sample information were used to classify case and control. To avoid variations owing to different tissue origins, we included single cancer type and focused on prostate cancer (GSE30174 [5]; GSE69961 [30]) having common platform (GPL570, HG-U133_Plus_2 array chip). To obtain a bigger cohort size for present meta-analysis study, we combined both the prostate cancer studies measuring the transcriptomics level of peripheral blood of patients receiving localized external beam radiation therapy at baseline, midpoints and endpoints.

Gene Expression Analysis
Transcription profiling of a total of 54,675 probe sets targeting~47,000 gene transcripts in the human genome was done as described previously [64,65]. Partek Genomics Suite version 6.6 (Partek Inc., St. Louis, MO, USA) was used to import affymetrix .CEL files which were normalized using robust multiarray average (RMA) algorithm. Analysis of variance (ANOVA) was applied on the grouped data set to analyze mean expression level on a gene-by-gene basis and list of differentially expressed genes were generated using a Benjamini Hochberg's false discovery rate (FDR), p-value <0.05 with fold change >2 as cut off. Disease and tissue type were two factors in ANOVA model and equal variance were assumed. Spearman's correlation similarity matrix was used for 2-dimensional unsupervised average linkage hierarchical clustering and classification. Principal component analysis was used to assess overall variance in gene expression and represents cohesive tendency of samples with similar features. Samples clustered tightly together were analyzed and outliers were removed from study to reduce unwanted variance. Venn diagrams were generated to display genes that intersect or non-intersect between groups of differentially expressed genes.

Functional and Pathway Analysis
The principal microarray data analysis was done to detect biological pathways to demonstrate the utilities of more robust biomarker discovery methods for radiation cytotoxicity. Ingenuity Pathways Analysis tool (IPA, build version 338830M, Ingenuity Systems, Redwood City, CA, USA) was used to define biological networks, interaction and functional analysis among the differentially regulated genes in localized external beam radiation treated patients. IPA workflow comprised core and functional analysis and the Ingenuity Knowledge Base was used as a reference data set. Both direct and indirect molecular relationships were included in the analysis settings and significance of relationships was indicated by z-score and Fisher's exact test p-values. Direction and ranking of pathways i.e., activated or inhibited, was determined by the number of uploaded molecules matching a canonical pathway. Network assembly to display significant biological functions was based on the interconnectivity of the uploaded molecules. We uploaded differentially expressed genes along with their p-values and fold changes, into the IPA tool for core analysis revealing associated genetic network, canonical pathway, and biofunctions.

Gene Set Enrichment Analysis
A broader understanding of global expression results is possible by grouping the genes of interest into biological processes, cellular component and molecular functions of the genes. Gene ontology (GO) enrichment study was done to functionally classify RT induced significant genes. The implication of this relationship amid transcriptomic data and canonical pathways were calculated by Fisher's exact test and a cutoff enrichment score >3 (p-value < 0.05) was used to identify major overexpression of functional categories.

Conclusions
Holistic simultaneous measurement of global gene expression using microarray technology transformed the field of cancer biology, and we used it to determine the impact of radiation on cells. We found complex molecular mechanisms including cell cycle arrest, DNA repair and apoptosis involved in regulating the cellular responses to radiation. The most prominent symptom experienced during radiotherapy was fatigue, probably mediated by SNCA and CA1 overexpression, and which may serve as a useful biomarker to understand the mechanisms and pathways related to its development. Hence, an additional impetus for research is the need to develop structure based inhibitors that may have potential co-therapeutic relevance. Also, it is vital to comprehend the radiation dose and mechanisms of response to define an ideal condition where the positive benefits notably outweigh the harmful effects. The cancer patients undergoing RT must be aware of the possible delayed cardiotoxic, hepatotoxic and nephrotoxic effects so that they can adopt a healthy lifestyle and have long term follow up with their health providers. Radiation oncologists should operate on the principle that there is no totally safe radiation dose especially for the heart, and they should avoid direct cardiac radiation and keep the dose as low as possible. Clinicians should look into the deeper aspect of recognizing and advocating radiation-induced organ toxicities and other adverse effects in young adult cancer survivors.