**Lead Investigator:** Susan Bates, MD, Columbia University Medical Center

**Title of Proposal Research:** Assessing the tumor growth and decay rates in patients with pancreatic cancer treated with Gemcitabine/Abraxane: Comparing real world data with clinical trial data

**Vivli Data Request:** 5163

**Funding Source:** None.

**Potential Conflicts of Interest:** None.

**Summary of the Proposed Research:**

Pancreatic cancer remains one of the most difficult-to-treat cancers, in part due to late diagnosis and in part due to chemotherapy refractoriness. In 2017, there were an estimated 53,670 new cases of pancreatic cancer. It is the 3rd leading cause of cancer deaths, with a projected 43,090 deaths in 2017. Most pancreatic cancers are diagnosed beyond the age of 60 years. The incidence rate is greater in blacks than in whites and greater in males than in females. Tobacco use and exposure are associated with an increased risk of pancreatic cancer, and tobacco use has been prevalent in the VA system. While there are two regimens with proven efficacy for pancreatic cancer available to clinicians, patients in the VHA system have co-morbidities particularly related to tobacco and alcohol use, and also often have cardiac, renal and hepatic dysfunction limiting care. They are thus more often treated with gemcitabine/Abraxane due to its greater tolerability. However, there has been no analysis of PDAC outcomes in the VHA system.

Pancreatic cancer remains one of the most difficult-to-treat cancers, in part due to late diagnosis and in part due to resistance to chemotherapy. In 2017, there were an estimated 53,670 new cases of pancreatic cancer. It is the 3rd leading cause of cancer deaths, with a projected 43,090 deaths in 2017. Most pancreatic cancers are diagnosed beyond the age of 60 years. The incidence rate is greater in blacks than in whites and greater in males than in females. Only 8% of people with pancreatic cancers are expected to live 5 years.

A complete surgical resection offers the only prospect of cure, and only 20% of patients have disease that is considered amenable to resection. For the small fraction considered immediately operable, surgical resection followed by adjuvant chemotherapy is demonstrated in randomized clinical trials to increase survival. Other patients have tumors that may become resectable with a good response to chemotherapy, considered ‘neoadjuvant’ therapy. Observational studies have confirmed a beneficial effect on survival and the best therapy in the neoadjuvant setting is unknown. Regimens used in these settings are derived from data obtained in the metastatic setting, and the choice of regimen often depends on the clinician’s impression of the patient’s ability to tolerate that regimen. Thus, demonstration of efficacy in the metastatic setting has implications for treatment across all disease stages. Almost 14,000 patients were enrolled in over 40 randomized trials to prove that these therapies exceeded gemcitabine alone. Ultimately, two regimens were successful – FOLFIRINOX, with an 11-month overall survival in metastatic disease, and gemcitabine/nab-paclitaxel with an 8.6-month overall survival. FOLFIRINOX is generally considered a more difficult regimen to tolerate, particularly in older patients with more comorbidities. As a result the field has seen increasing use of gemcitabine/Abraxane in many settings, including among Veterans with pancreatic cancer.

Our intent is to evaluate gemcitabine/Abraxane in the VHA system. The Veterans Health Administration (VHA) is a nationwide healthcare system that provides care for over 9 million US Veterans enrolled into the VA health care program. U.S. Veterans of military service generally have high rates of co-morbidities, including those due to tobacco and alcohol use. The VHA has the world’s largest common electronic health care record system, with research access hosted on the VA Informatics and Computing Infrastructure (VINCI). We plan to use VINCI data to examine the outcomes of patients with pancreatic cancer in the VHA system. Gemcitabine/Abraxane is widely used in the VHA system and one question that can be asked is whether outcomes in this setting are comparable to those observed in the landmark clinical trial that established Gem/Abraxane as effective in pancreatic cancer, particularly given the increased rates of co-morbidities seen in the VHA patient population. We will perform a retrospective analysis of Veterans treated for pancreatic cancer within the VA system. With an estimated 300 new diagnoses each year in the VHA, we will be able track outcomes for over a decade. We plan to use the CA046 data as a benchmark to validate the VHA Gem/Abraxane data. We will be able to gather co-morbidities, dosing, survival, and CA 19-9 values as a marker of efficacy. We will determine whether Veterans treated with Gem/Abraxane in the VHA systems have comparable outcomes as patients treated in the Celgene CA046 clinical study. We will initially focus on using the CA19-9 values to determine tumor growth rates during treatment, and gather duration of treatment and survival data as well.

**Statistical Analysis Plan:**

Model process

The regression-growth models used to generate growth rates are based on the assumption that change in tumor quantity during therapy results from two independent component processes: an exponential decrease or regression, d, and an exponential growth or regrowth of the tumor, g. The model for this is displayed below (labeled as gd) where f(t) is the tumor quantity at time t in days, normalized to the tumor quantity at time 0, dis the rate of decay, and g is the rate of growth.

f(t)=e−dt +egt −1

For data showing continuous decrease from the start of treatment, g is eliminated as shown below (labeled as dx).

f(t) = e−dt

Similarly, d is eliminated when data show a continuous growth from the start of treatment as shown below (labeled as gx).

f(t) = egt

The fourth model (below and labeled as gdphi) contains an additional parameter, phi (Ø), which represents the proportion of tumor cells that undergo cell death due to therapy. In this model, d again is rate of regression or decay of the fraction of the tumor that is sensitive to the therapy (Ø), while g is rate of growth of the therapy- resistant tumor fraction (1 – Ø):

f(t) = (Ø)e−dt + (1 − Ø)egt

The Levenberg-Marquardt algorithm was used to solve these 4 non-linear least squares problems and among models where all parameters were significant predictors (given user supplied cutoff that defaults to 0.10), the model that minimized the Akaike Information Criterion (AIC) was the selected model for a given patient from which tumor growth and regression rates were obtained. The basis of AIC is information theory and provides a “relative estimate” of how much information is lost when a given model is used to represent data – it looks to minimize this loss of information. It does this by dealing with the trade-off between the complexity of the model and the “goodness” of the fit. We used it to discriminate amongst the models in deciding which allowed for best fit of the data. The selected model minimizes AIC with AIC = [20 parameters in model)]. Comparisons of growth rate distributions were done by Wilcoxon two-sided tests (where groups analyzed = 2) or by Kruskal Wallis tests (where groups analyzed >2) followed by a Dunn’s test for pairwise difference where there was an overall difference.

Patients with insufficient or missing data, or patients with sufficient data where no model converged were excluded and noted individually in results and summarized in models output with one of the following explanations: no data (cases with all missing data), only 1 or 2 data points (where the latter has less than 20 percent difference in tumor measurements), error data (where only one unique measurement value for a patient that is repeated 3 or more times, and/or where both the initial and final measurement value is zero), or not fit. Patient data that did not fall into one of the categories listed above are labeled as included. Plots were generated for all included cases (cases fit by models), where the observed and predicted values from the selected model (labeled in plot legend) are depicted. Note that because 2 data points with less than 20 percent difference in tumor measurements would not be scored by RECIST as either progression or response and the reliability of differences of this magnitude can be questioned, we do not calculate rates of growth or regression.

We have previously validated these equations and observed that most patient data are fit by one of these four equations. We developed an R package, designated tumgr that allows one to obtain tumor growth rates using the models above.

Statistical analyses:

All analyses and graphical output will be done in R 3.3.3. Comparisons of model estimate distributions will be performed using the two-sided Wilcoxon test of location (Kruskal Wallis tests (where N groups >2), and post hoc analysis of any overall difference detected (# groups > 2) will be tested using Dunn’s test with Bonferroni adjustment for multiple comparisons.

**Requested Studies:**

Phase III Study of ABI-007(Albumin-bound Paclitaxel) Plus Gemcitabine Versus Gemcitabine in Metastatic Adenocarcinoma of the Pancreas

Sponsor: Celgene Corporation

Study ID: NCT00844649

Sponsor ID: NCT00844649

**Public Disclosures:**

- Keith Sigel, Mengxi Zhou, Yeun-Hee Anna Park, Tinaye Mutetwa, Girish Nadkarni, Celine Yeh, Paz Polak, Carlie Sigel, Thierry Conroy, Béata Juzyna, Mark Ychou, Tito Fojo, Juan P Wisnivesky, Susan E. Bates, Gemcitabine plus nab-paclitaxel versus FOLFIRINOX for unresected pancreatic cancer: Comparative effectiveness and evaluation of tumor growth in veterans, Seminars in Oncology, 2021, ISSN 0093-7754, doi: 10.1053/j.seminoncol.2021.02.001.
- Celine Yeh, Mengxi Zhou, Keith Sigel, Gayle Jameson, Ruth White, Rachael Safyan, Yvonne Saenger, Elizabeth Hecht, John Chabot, Stephen Schreibman, Béata Juzyna, Marc Ychou, Thierry Conroy, Tito Fojo, Gulam A Manji, Daniel Von Hoff, Susan E Bates. Tumor Growth Rate Informs Treatment Efficacy in Metastatic Pancreatic Adenocarcinoma: Application of a Growth and Regression Model to Pivotal Trial and Real-World Data. The Oncologist, 2022. doi: 10.1093/oncolo/oyac217