Prognostic Analysis of Different Metastatic Patterns in Invasive Intraductal Papillary Mucinous Neoplasm: A Surveillance, Epidemiology, and End Results Database Analysis

Objective To evaluate the impacts of different metastatic patterns on the prognosis of patients with invasive intraductal papillary mucinous neoplasm (IPMN). Materials and Methods All patients who were diagnosed with invasive IPMN in the Surveillance, Epidemiology, and End Results SEER database (2010–2015) were included in this study. They were grouped according to different metastatic patterns. Kaplan–Meier analysis and log-rank test were used for the comparison of their survival rates. The hazard ratio (HR) with 95% confidence interval (CI) was analyzed using the Cox proportional-hazards model. Results A total of 2264 cases were included in this study. The most common metastatic site was the liver. The patients with the nonorgan metastasis demonstrated the best survival outcomes, while those with multiple metastases showed the worst survival outcomes. As compared to the patients with isolated liver metastasis, those with isolated lung and other organ metastases showed better overall survival rates and tumor-specific survival rates. The patients with liver, lung, multiple, and other organ metastases or of age >60 years were the independent predictors of poor prognosis. Conclusions The patients with isolated lung and other organ metastases demonstrated better survival outcomes as compared to those with isolated liver metastasis. The patients with nonorgan metastasis demonstrated the best survival outcomes, while those with multiple metastases showed the worst survival outcomes. Further studies are needed to determine a highly selected subset of patients, who might benefit from surgery or chemotherapy.


Introduction
Ohashi et al. first described the intraductal papillary mucinous neoplasm (IPMN) about 30 years ago [1]. Due to the innovations in imaging technology and its expanded use, pancreatic cysts are easily detected and the incidences of IPMNs are also increasing [2]. Recently, IPMNs have become one of the most dramatic pancreatic tumors [3]. Currently, IPMNs represent 25% of all the cystic neoplasms of the pancreas, with an assumed incidence of 0.8 per 100,000 [4]. IPMNs are the most common of all the cystic tumors of the pancreas; branching IPMNs develop cancer in only 30% of the cases, but the main duct and mixed IPMNs have a 70% risk of becoming malignancies [5]. Invasive carcinomas, arising in or accompanying IPMN, can be of various types [6]. e primary treatment for the invasive IPMNs is surgical resection [7]. For patients with metastatic cancers, the primary treatment is antitumor therapy and palliative care. Despite the availability of various types of comprehensive therapies, the 5-year survival rates of the patients with invasive IPMNs are still poor [8,9].
As compared to the common pancreatic ductal adenocarcinoma (PDAC), the invasive IPMNs demonstrate a more favorable prognosis [10,11] and appear to be more indolent than the conventional PDAC [12]. For the invasive IPMNs, the American Joint Committee on Cancer (AJCC) staging classification is appropriate, where the 7th edition of staging classification is more applicable than the 8th edition [4]. However, the biological behaviors of invasive IPMNs are different from those of the PDAC [12,13]. Previous studies have analyzed the prognostic impacts of different metastatic patterns on patients with PDAC and pancreatic neuroendocrine tumors (pNETs) [14,15]. However, the studies focusing on the metastatic patterns of invasive IPMNs are limited. erefore, in this study, the prognostic potential of different metastatic patterns of the patients with invasive IPMN based on the SEER database was evaluated.

Data.
e data from the SEER database is a welldesigned electronic medical record database for cancer research. Use SEER * Stat software (version 8.3.4; National Cancer Institute, Bethesda, MD, USA) to obtain patient demographic data, clinical tumor characteristics, the first course of treatment, and follow-up data of life status from the SEER database. Because this study was based on a publicly available database, it was exempted from IRB approval [16].
We collected patients diagnosed with invasive IPMN that were reported to the SEER database from 2010 to 2015.

Variable Definition.
e variable included gender, age at diagnosis, race, marital status, tumor site, surgery, chemotherapy, radiation, grade, cancer-specific death, survival months, and vital status. e cancer-specific survival rate (CCS) is calculated based on the date of death associated with IPMN. Deaths attributed to other causes are considered to be censored observations; we divide the cases into the following groups according to the location of tumor metastasis. Since there is only 1 patient with isolated brain metastasis, this group was excluded, and we divided these cases into 6 groups.

Statistical Analysis.
Continuous variables were compared using Student's t-test, whereas categorical variables were compared using the chi-square test. We used Kaplan-Meier analysis and log-rank test to compare the survival rates. e Cox proportional model was employed to calculate the hazard ratio (HR) with 95% CI. P < 0.05 was considered statistically significant. e statistical analysis was performed using the software SPSS 24.0 (IBM, NY, United States). Proportional-hazards assumption was performed using the software STATA 16.0.

Survival Outcomes.
e overall survival (OS) and cancer-specific survival (CSS) of the different metastatic sites are shown in Figure 2. e patients with multiple organs metastases demonstrated the worst survival outcomes, while those with nonorgan metastasis had the best survival outcomes. Moreover, as compared to the patients with isolated liver metastasis, those with isolated lung metastasis and other organ metastasis showed better OS and CSS (for OS, lung vs. liver metastasis P < 0.0001 and others organ vs. liver metastasis P � 0.001 and for CSS, lung vs. liver metastasis P < 0.0001 and others organ vs. liver metastasis P � 0.002).
e multivariate analysis revealed that the age of <60 years, yellow race, surgery, and chemotherapy were associated with better OS and CSS (Table 2). Furthermore, as compared to the patients with isolated liver metastasis, those with isolated lung metastasis and other organ metastasis showed better OS and CSS (  values are given in Table 2. P values for the PH results of the integrity of two Cox regression models were not significant, which could not be considered as a violation of the PH assumption test. Besides, the PH assumption test for OS and CCS based on the variables of metastasis ( Figure 3) showed that the curve of the ln (−ln (survival probability)) tended to be parallel in the different metastatic groups. Consequently, the PH hypothesis of this variable was considered valid. Moreover, the median survival time (MST) of the isolated lung metastasis was 6 months, followed by the other organ, isolated liver, and isolated bone metastases (4 months each). e MST of the multiple organs metastases was only 2 months.

Treatment Modality.
e treatment modalities are given in Table 1. A total of 728 (32.2%) patients underwent surgery, while 1536 (67.8%) patients did not undergo surgery. e surgery for the patients with distant metastasis was mainly incisional, needle, or aspiration biopsy of the primary site and/or other than primary site or palliative surgery. About half of the patients (1068, 47.2%) received chemotherapy and a few (292, 12.9%) patients received radiotherapy. e patients who underwent surgery or chemotherapy showed significant improvement in OS and CSS for those having nonorgan, isolated liver, and isolated lung metastases (Figures 4 and 5). However, the radiotherapy could not improve the 5-year OS or CSS of the  9 * Multiple mean metastases in at least 2 of the four organs (liver, lung, bone, and brain). * Other organ metastasis mean metastases organs other than the liver, lungs, bones, and brain. * MST mean median survival time. For OS (a): nonorgan metastasis vs. other groups, P < 0.001; isolated liver vs. isolated lung metastasis, P < 0.001; isolated liver vs. isolated bone metastasis, P � 0.048; isolated liver vs. multiple metastases, P � 0.007; isolated liver vs. other organ metastasis, P � 0.001; isolated lung vs. isolated bone metastasis, P � 0.739; isolated lung vs. multiple metastases, P < 0.001; isolated lung vs. other organ metastasis, P � 0.151; isolated bone vs. multiple metastases, P � 0.019; isolated bone vs. other organ metastasis, P � 0.486; multiple vs. other organ metastasis, P < 0.001. For CSS (b): nonorgan metastasis vs. other groups, P < 0.001; isolated liver vs. isolated lung metastasis, P < 0.001; isolated liver vs. isolated bone metastasis, P � 0.078; isolated liver vs. multiple metastases, P � 0.011; isolated liver vs. other organ metastasis, P � 0.002; isolated lung vs. isolated bone metastasis, P � 0.824; isolated lung vs. multiple metastases, P < 0.001; isolated lung vs. other organ metastasis, P � 0.143; isolated bone vs. multiple metastases, P � 0.019; isolated bone vs. other organ metastasis, P � 0.487; multiple vs. other organ metastasis, P < 0.001.     patients with isolated liver and isolated lung metastases ( Figure 6).

Discussion
In this study, the prognostic impacts of different metastatic patterns on the invasive IPMNs were investigated. e results showed that the patients with isolated lung metastasis or other organ metastasis demonstrated better survival outcomes than those having isolated liver metastasis, while those with multiple organ metastases showed the worst survival outcomes. ese results were consistent with those in PDAC [17]. A previous study showed that adjuvant therapy for the invasive IPMNs could not improve the survival outcomes in patients with early stage metastasis [12]. However, in this study, the surgery or chemotherapy could improve the survival outcomes in metastatic invasive IPMNs (Figure 2). For the stage IV PDAC, chemotherapy was the primary choice when the performance status allowed. In clinical practice, therapy principles for the invasive IPMNs are considered as PDAC. However, the biological behaviors of invasive IPMNs are different from those of PDAC [12,13], and fewer studies have focused on the efficacy of adjuvant therapy in stage IV invasive IPMN.
Furthermore, the prognosis of the patients with isolated lung metastasis was better than those with other organ metastasis, which might be due to the more effectiveness of chemotherapy or surgery for the isolated lung metastasis [18] or due to the fewer complications in isolated lung metastasis. Moreover, the isolated lung metastasis had better OS and CSS as compared to the isolated liver metastasis, which was consistent with the results of a previous study on PDAC [15]. is was the first study focusing on the effects of metastatic patterns on the invasive IPMNs. e mechanisms of tumor metastasis mainly include direct invasion, lymphatic metastasis, and blood metastasis. Specifically, the metastatic process is closely related to the cross-talk between cancer and the vascular and/or lymphatic system. e normalization of tumor blood vessels can improve the infiltration of T cells, enhance the immune response and immune reaction, and halt the immunesuppressing environment to a more immune-activating phenotype and work together with cancer immunotherapy [19]. Antivascular endothelial growth factor receptor (anti-VEGFR) was used to normalize the tumor vascular system and restore its function and fostered further investigations, aiming at the formation of intratumoral immune cell phenotypes parallel to the normalization of blood vessels, as indicated by the reduction of tissue perfusion and intratumoral hypoxia [19,20]. e anti-VEGFR polarizes the macrophages in the M1 macrophage by altering the gene expressions at the same time parallel to an increase in the adaptive immune cells' infiltration in the setting of this antiangiogenic treatment. Vascular endothelial growth factor (VEGF) and inflammatory molecules are not only the key proangiogenic elements but also act as immunomodulators. ey promote the formation of blood vessels in most of the fatal malignant tumors and collaboratively create a permissive environment, resulting in the poor efficacy and survival of the patients. Cancer cells grow and progress by continuously interfering with the neighboring environment during their growth and progression. e combination of strategies, such as antiangiogenesis and immune-directed therapy, might shape the tumor ecosystem and improve the therapeutic effect [19]. In this study, there are some limitations too. First, the results should be interpreted with caution, given the inherent bias of a retrospective study. Second, the missing information on related comorbidities as well as the absence of therapy might have affected the results. Furthermore, the surgeries and chemotherapies were performed in the patients with longer life expectancies. However, this was the first study with the largest sample size to clarify the prognosis of patients with the metastatic invasive IPMN.

Conclusion
e patients with isolated lung and other organ metastases demonstrated better survival outcomes as compared to those with isolated liver metastasis, while the patients with nonorgan metastasis demonstrated the best survival outcomes and those with multiple metastases showed the worst survival outcomes. Further studies are needed to determine a highly selected subset of patients who might benefit from surgery or chemotherapy.

Data Availability
e data analyzed during the current study are available in the Surveillance, Epidemiology, and End Results (SEER) database of the National Cancer Institute (https://seer. cancer.gov/data/).

Conflicts of Interest
e authors declare that they have no conflicts of interest.