Patients with recurrent metastatic disease represent a significant proportion of the total number of pancreatic cancer patients presenting to oncology clinics. Despite this fact, there is limited data to help guide the management of these patients in the palliative setting. In this study, we examined real-world outcomes of patients with recurrent PDAC based on an analysis of a large US database. As would be expected in a real-world setting, our data demonstrate approximately one-fifth of patients receiving treatment for advanced disease are classified as having recurrent disease following curative-intent resection. When comparing outcomes of patients with recurrent PDAC to those with de novo metastatic PDAC, our survival analysis suggests that patients with recurrent PDAC who receive palliative systemic therapy have a significantly greater OS, with a clinically significant absolute difference of approximately three months. Additionally, we did not detect any difference in overall survival in the recurrent patient population based on a recurrence within six months compared to after six months from surgery.
There are multiple potential explanations for the increased overall survival in the recurrent population. It is likely that at least some of the difference in survival between groups in this study is the result of biases inherent to retrospective studies that could not be completely controlled. Specifically, a selection bias may exist in terms of the patients who are selected for surgery. One can hypothesize that patients with better ECOG performance status are selected for surgery, as evidenced by the greater percentage of patients with an ECOG less than 2 in the recurrent group compared to the de novo group (65% versus 51%). Additionally, a poor performance status following resection may result in a patient not proceeding with or being offered systemic therapy following a diagnosis of recurrent disease while patients with an equally poor performance status who are treatment naïve may still elect to pursue systemic therapy. It is also possible that our findings are the result of lead-time bias with close surveillance following surgery leading to detection of recurrent disease in patients at an earlier time when they have a smaller overall tumor burden than those diagnosed with de novo advanced disease who may have a large intact primary tumor or more extensive metastatic disease. The significant difference in CA 19–9 at time of advanced diagnosis (92.8 U/ml in the recurrent group vs. 617 U/ml in the de novo group) supports this hypothesis. Although the data set we utilized did not have data regarding surveillance frequency following surgery, future studies examining the correlation between post-surgical surveillance, CA19-9 levels and overall survival may help guide treatment in a recurrent PDAC population.
Despite these potential biases, it is also important to consider whether our findings could be related to actual differences in the underlying biology of disease in patients with recurrent versus de novo disease. When recurrent PDAC tumor samples were sequenced, the genomic landscape shows a heterogeneity of driver mutations between the primary tumor sample and metastasis. Namely, in patients who have undergone adjuvant treatment, samples from their metastatic sites showed an increased tumor mutational burden, as well as enrichment in genetic aberrations in the MAPK or PI3K-AKT pathways. This is in direct contrast to de novo stage IV PDAC tumors which showed a homogenous driver mutation pattern between primary and metastatic samples consistent with a clonal population [13]. Peri-operative chemotherapy likely plays a large role in selecting for a certain subpopulation of cells, however tumor microenvironment, SMAD4status and immunophenotypes have all been explored as factors contributing to disease recurrence and overall outcome in initially resectable disease [21,22,23]. This may suggest that patients with recurrent stage IV PDAC should perhaps be thought of differently than de novo patients when choosing treatment.
Regardless of the explanation for differences in survival discussed above, what seems clear from our data is that patients with recurrent PDAC appear to do at least as well or better than patients with de novo advanced PDAC, even if their disease recurs early after surgery. Thus, we argue that there is little justification for the exclusion of patients with recurrent PDAC in future clinical trials evaluating palliative systemic therapy.
Increasing enrollment in these trials is imperative to improving the outcomes of patients with this disease. While the percentage of patients enrolled in clinical trials for pancreatic cancer has been increasing modestly over the past decade, up from 3.85% in 2011 to 4.15% in 2014, there is still a large imbalance between patient need and number being enrolled [24]. This imbalance is unlikely to be due to difficulty enrolling patients who qualify for trials. A single-center retrospective study in 2014 found that 71% of patients eligible for clinical trials were enrolled [25]. More likely, the low number of patients being enrolled is due to overly restrictive inclusion and exclusion criteria, such as limits on prior therapy or surgery. Indeed, of the patients that called the Pancreatic Cancer Action Network with interest in enrolling in a clinical trial, two-thirds were ineligible due to prior therapy [24, 26]. In the context of a disease where patients can decline rapidly in the absence of systemic therapy, it would be reasonable to allow patients who have received one or two doses of chemotherapy to still be enrolled in first line trials.
From 2011 to 2014 there was an increase in clinical trials for refractory and previously treated disease to 38%. This had the effect of decreasing the time it would take for trials studying this population to complete enrollment from 7.1 years to 6.0 years [24]. However, this is still extremely slow for a disease in which survival is generally on the order of months. Our work shows no difference in overall survival based on when patients recurred after surgery. Future trials evaluating palliative therapies in the advanced disease setting should consider inclusion of patients with recurrent PDAC, even if their recurrence occurs early after surgery. Given potential differences in survival between these populations, trials should stratify any randomization by disease presentation.
This study was based on a well-validated, nationwide real-world database; however, several limitations should be noted. Despite attempts to reduce confounding utilizing a multivariable analysis, it is possible that residual confounding may have contributed to our results. Most notably, data regarding receipt of neoadjuvant or adjuvant treatment was not available for inclusion in our statistical models. Additionally, some incomplete data was present for several variables included in our analysis, such as ECOG status. It should also be noted that in order to compare overall survival between our de novo metastatic population and recurrent metastatic population we chose to define overall survival as time from metastatic diagnosis to death. Despite these limitations, we believe this study provides valuable information that can be used to guide discussion of treatment options with patient in the clinic and to help inform the development of more inclusive clinical trials in this devastating disease.