Longitudinal Metabolomics Study on Pancreatic Cancer

Li Jiao

We plan to submit an application for U01 The Pancreatic Cancer Detection grant in response to PA-15-289 on May 26, 2017. We will also consider submit the grant to Pancreatic Cancer Action Network Translational Research Grant and Early Detection Targeted Grant. _x000D_Other choices: Investigator-initiated NCI-U01 or CPRIT grant. _x000D__x000D_We will budget a yearly 5% of co-investigators effort for the entire study period and purchase the sample at the rate of $50 per sample. In the first year of the research, we will budget 5% effort for a programmer for sample acquisition. In summary, we plan to submit a grant by July 1, 2017 and will follow the budget guideline provided by the participating cohorts.

1. No early detection tool for pancreatic cancer. Pancreatic cancer has the lowest five-year all-cause survival rate among all malignancies. Secondary prevention, i.e., early diagnosis of pancreatic cancer, holds promise for improving the prognosis of this disease. Patients have a 5-year survival rate of 70~100% when tumor is < 1 cm and operable, compared to 14% when the tumor is > 2 cm. However, most patients are diagnosed with advanced disease. Carbohydrate antigen 19-9, the only US-FDA approved biomarker, is a poor screening tool. The sensitivity and specificity of other modalities, such as MRI, CT and ERCP, are suboptimal to identify a tumor with a diameter of < 2 cm. There has been a consistent need for non-invasive early diagnosis markers for pancreatic cancer. With the alarming increased trend of pancreatic cancer, the need becomes urgent._x000D_ _x000D_2. Metabolomics approach may facilitate discovery of novel biomarkers for early diagnosis of pancreatic cancer. Metabolomics in combination with advanced biostatistics a growing and powerful non-invasive technology capable of detecting low-molecular-weight metabolites in cells, tissues and biofluids. Comparative analysis of the metabolome can aid in the discovery of signals influenced by gene, environment, or both. Metabolites are by-products of cellular metabolism and are the closest point to a phenotypic endpoint. Cancer, including pancreatic cancer, is a disease that is known to alter cellular metabolism. The premise of using metabolomics as an early detection tool is that the subtle chemical change due to perturbed metabolism in cancer development can be detected by sensitive mass spectrometry (MS) before the anatomic change can be captured. Previous studies have identified pancreatic cancer-specific metabolites using tissues, urine, serum, and plasma. Although internal and/or external validation of candidate markers are mostly lacking, studies concluded that the metabolites involved in metabolism of lipid, glucose, amino acid, DNA synthesis, or muscle protein breakdown had superior sensitivity and specificity in discriminating pancreatic cancer from non-cancer controls or chronic pancreatitis than existing diagnostic tools. _x000D__x000D_3. The diagnostic value of metabolite biomarkers in pre-diagnostic sample is unknown. The published studies were mostly on patients with late-stage pancreatic cancer. We propose to examine the metabolites using pre-diagnostic blood sample where cancer-specific metabolic changes are likely to be captured using sensitive LC/MS detection. Our ultimate goal is to identify highly sensitive and specific biomarkers for early diagnosis of pancreatic cancer among high-risk populations including those with genetic heritability, long-term or recent onset of diabetes, and those with chronic pancreatitis. Novel diagnostic markers in combination with evolving personalized treatment of pancreatic cancer are likely to have immediate impact on mortality of this disease. _x000D__x000D_4.Using the blood samples collected within one year before pancreatic cancer diagnosis, our previous feasibility study within the Women's Health Initiative Study found that serum levels of methyl-tryptophan were significantly lower in women with pancreatic cancer compared with controls. The area under the curve for combined markers of tryptophan, aspartic acid, and acadesine for differentiating pancreatic cancer from controls was 0.83. In addition, this research has also provided novel insight into etiological molecular pathway beyond those have been identified by other omic approaches.

The proposed research has two broad goals._x000D_To apply targeted metabolomics and untargeted lipidomics approach for identifying metabolite biomarkers of early subclinical pancreatic cancer within 10-year latent period window. Specifically, we will conduct metabolomics and lipidomics research on pancreatic cancer cases and controls using blood samples collected upto 10 years before pancreatic cancer diagnosis (or corresponding date in controls). _x000D_To determine the diagnostic value of the metabolites and identify metabolism pathway that is altered in subclinical pancreatic cancer.

To identify in a two-stage case-control study the dynamic change of metabolite levels 1 year, 3 years, 6 years or 10 years before pancreatic cancer using blood samples collected at baseline in prospective cohorts. _x000D_To evaluate the diagnostic value of specific metabolites in differentiating pancreatic cancer cases from controls overall or at different latent period of pancreatic cancer. _x000D_To identify novel metabolism pathway that is implicated in subclinical pancreatic cancer.

We propose a two-stage individually-matched case-control metabolomics study using a targeted metabolites and untargeted lipidomics approach in pre-diagnostic blood specimens. _x000D_ We will identify cases with pancreatic cancer that arose within 10 years after baseline blood collection. Each control will be individually matched to each case (1:1) according to age (± 1 year), race/ethnicity (exact), study cohort (exact), month of blood draw (± 3 months), season of blood draw, time of blood draw (± 3 hours), diabetes status (7 cases had self-reported type 2 diabetes), smoking status (current, former and never), and BMI (± 5 kg/m2). Matched cases and controls will be randomly assigned to the discovery set or the validation set with a 3:2 ratio. Both cases and controls should have no cancer history. All controls should be cancer-free during the follow-up. We aim to include at least 50 case-control pairs for each latent period (1, 3, 6 or 10 years before diagnosis)._x000D_ In the discovery stage, we propose to quantify ~500 metabolites (amino acid, sugars, nucleotide, carnitines, CoA & vitamin, prostagiandins, bile acids, and xenobiotics, as well as TCA and urea cycle) and ~600 lipids with the aim to identify 40 metabolites to be included in the validation stage (metabolites differentially present in cases and controls using P < 0.1 and fold change (FC) > 2 criteria). _x000D_ In the validation stage, we will test the identified metabolites and lipids in additional case-control pairs. We aim to identify at least five metabolites that are differentially present in cases and controls using FDR adjusted P value < 0.1 and FC > 2 criteria. _x000D_ _x000D_ Finally, we will identify a risk prediction model using multivariate logistic regression models, also incorporating traditional risk factors for pancreatic cancer, and evaluate the diagnostic value of metabolites using ROC. We will also conduct a pathway analysis using metabolomics data.

Profiling blood metabolites and lipids in subclinical pancreatic cancer may be valuable for early detection. We hypothesize that the metabolite profile varies during pancreatic cancer development and certain change will preserve. We propose to use blood collected 1 year, 3 years, 6 years and 10 years before diagnosis. A consortium is necessary to fulfill sample size requirement for each latent period. Larger sample size will also allow us to conduct stratified analysis by sex, ethnicity, BMI, and tumor stage.

The primary aim of the proposed metabolomics study is to identify specific metabolites that can differentiate future pancreatic cancer cases from non-cancer controls using blood specimens collected up to 10 years before the clinical diagnosis of pancreatic cancer. For each case, we will identify individually matched control from the same cohort. _x000D__x000D_In the consortium, we aim to have at least 50 cases arising for each latency period. _x000D__x000D_We will require 150 ul serum from each participant._x000D__x000D_Exclusion criterion: history of cancer; participants develop cancer during cohort follow-up.

Date of diagnosis or ascertainment of pancreatic cancer._x000D_Date of blood collection._x000D_Tumor stage (optional)_x000D_Pathological code (optional)

The major exposure is metabolites in this study. Other exposure data needed include age; sex; ethnicity; location, month, season, day, time of blood draw; freeze-thaw cycle; _x000D_history of cancer, type 2 diabetes, chronic pancreatitis, smoking status, body mass index, and diet data. If non-fasting blood was collected, the information on time since last meal will be needed.

Physical activity, history of hypertension or other chronic diseases, medication use, NSAIDs use, hormone replacement therapy for women._x000D_Some cohorts may have had already done non-targeted metabolomics research on pancreatic cancer. In this case, we will likely not pursue the use of blood sample. We propose to use the existing data depending on the study design of the original study.

Yes

Preferred: fasting blood; no freeze-thaw cycle, serum._x000D_Time since last meal time is known for non-fasting samples.