A Possible De nition of Oligometastases in Pancreatic Cancers and Their Survival Outcomes

Masaya Yamanaka Nagoya University Graduate School of Medicine Masamichi Hayashi (  m-hayashi@med.nagoya-u.ac.jp ) Nagoya University Graduate School of Medicine Suguru Yamada Nagoya University Graduate School of Medicine Fuminori Sonohara Nagoya University Graduate School of Medicine Hideki Takami Nagoya University Graduate School of Medicine Yoshikuni Inokawa Nagoya University Graduate School of Medicine Dai Shimizu Nagoya University Graduate School of Medicine Norifumi Hattori Nagoya University Graduate School of Medicine Mitsuro Kanda Nagoya University Graduate School of Medicine Chie Tanaka Nagoya University Graduate School of Medicine Goro Nakayama Nagoya University Graduate School of Medicine Masahiko Koike Nagoya University Graduate School of Medicine Yasuhiro Kodera Nagoya University Graduate School of Medicine


Introduction
Pancreatic ductal adenocarcinoma (PDAC) is the fourth reason for estimated cancer deaths in the United States [1]. PDAC remains to be lethal disease in that the majority of patients present with distant metastases at the time of diagnosis [2]. These patients are usually not indicated for surgery and the prognosis is deemed dismal.
In 1995, Hellman et al. proposed the idea of oligometastasis of advanced malignancies, which is de ned as a state of limited metastases [3]. The basis of this concept is that a particular group of patients could bene t from long-term survival through surgical resection after precise imaging and appropriate multimodal treatment. For instance, hepatic resection of limited colorectal liver metastases demonstrated 5-year overall survival (OS) of 28-58% [4][5][6]. Local therapy was also reported to improve the OS of the oligometastatic non-small cell lung cancer patients [7]. Furthermore, Ozawa H et al. reported that the 5-year OS rate of the patients treated with curative resection of limited peritoneal metastases was signi cantly higher than patients with diffuse peritoneal metastases [8].
As for the oligometastasis in PDACs, there are no established criteria or consensus of diagnosis and treatment strategy. Damanakis AI et al. proposed a de nition of oligometastases in PDACs as the cases with single organ metastases, the presence of ≤ 4 metastases in liver or lung and CA19-9 baseline < 1000 U/ml. These patients survived signi cantly longer than other metastatic unresectable PDACs [9]. However, they failed to mention peritoneal metastases. The current study attempted to create a new de nition of oligometastasis which takes peritoneal metastases into consideration among other types of metastasis, using our metastatic PDAC patients cohort.

Patient cohort
To ensure accurate evaluation of peritoneal metastases, we retrieved 140 consecutive cases of unresectable metastatic PDAC patients who received staging laparotomy or gastrointestinal bypass operation at Nagoya University Hospital (Nagoya, Japan) from April 2001 to December 2019. Ten cases with multi-organ metastases were excluded, and the remaining 130 patients with single organ metastasis at the diagnosis were enrolled. Preoperative serum CA19-9 values had been recorded in all patients. The clinical background features were summarized in Table 1. Because there were only two metastases to lung cases, we excluded it from the de nition of OM.

Liver metastases
Information of liver metastases was collected from preoperative CT, MR images and operation records. Following the criteria of the Japanese Society of Cancer of the Colon and Rectum (JSCCR) [10], liver metastases were classi ed into four categories. H0: No liver metastasis, H1: 1-4 metastatic tumors, all of which are 5 cm or less in maximum diameter, H2: Other than H1 or H3, H3: Five or more metastatic tumors, at least one of which is more than 5cm in maximum diameter. We also investigated an association between the number of liver metastasis and survival outcomes. We then tried to nd which of the categories can be an ideal threshold of relatively favorable survival outcomes.

Peritoneal metastases
Information of peritoneal metastases was also collected from preoperative CT images and operation records. Peritoneal metastases were also classi ed into four categories following the criteria of JSCCR [10]. P0: no peritoneal metastasis, P1: metastasis localized to adjacent peritoneum, P2: limited metastasis to distant peritoneum, P3: diffuse metastases to distant peritoneum. We again tried to nd which of the categories is most appropriate as a threshold to identify patients with relatively favorable survival outcomes.
Statistics Continuous variables were analyzed by the Mann-Whitney U test as a non-parametric test and Student's ttest (two-tailed) as a parametric test. Categorical variables were analyzed by Fisher's exact test. OS was de ned as the time from non-curative surgery to the date of death of PDAC. The association of OM status and clinical factors with OS was evaluated using the log-rank test or Cox proportional hazards model. P < 0.05 was considered statistically signi cant for all statistics. Statistical analyses were carried out using JMP 15 software (SAS Institute, Cary, NC, USA).

Liver metastasis
Among 58 liver-metastatic cases, survival curves of three categories (H1, H2, H3) were compared in Figure 1. OS of H1 was signi cantly longer than others. The median survival time (MST) was 9.8 months (95% CI: 6.8-13 months) in H1 and 4.8 months (95% CI: 3.4-7.3 months) in H2-3 (P=0.001). On the other hand, no signi cant difference was detected between H1-2 (MST: 7.5 months, 95% CI 5.9-9.8 months) and H3 (MST: 4.8 months, 95% CI 1.3-14.4) (P=0.537). H1 cases solely seemed to have a potential of relatively favorable prognosis in liver-metastatic cases. Survival curves based on the number of liver metastases were shown in Figure 2a. Although the cases with solitary liver metastasis did not show signi cantly longer OS than others, the cases with two or fewer liver metastases demonstrated signi cantly longer OS than others (Figure 2b, P=0.001). The cases with four or fewer liver metastases also showed better OS compared to ve or more. (Figure 2c, P=0.001).

Tumor markers
As for serum tumor markers, we chose CA19-9 as the most frequently available and reliable tumor marker of PDACs in this study. Hartwig et al. reported CA19-9 predicts resectability, stage of disease, as well as survival in patients with pancreatic adenocarcinoma [11]. To detect an optimal cut-off of preoperative serum CA19-9 value, ROC curve analyses were performed for ve months survivors, ten months survivors, or 20 months survivers-speci c settings after non-curative operations. We chose 20 months survivorsspeci c ROC curve because of the most signi cant area under the curve (0.661) compared with others (0.576 in ve months survivors, 0.523 in 10 months survivors). The optimal cut-off value for predicting survival of ≥20 months was 2000 U/ml. The sensitivity and speci city of this setting were 100% and 40%, respectively (Figure 4).

De nition of oligometastatic cases
As a result, we de ned OM cases as single organ metastasis (H1 or P1-2) and low serum CA19-9 (< 2000 U/ml). All cases were classi ed according to the number of metastatic organs, metastatic sites and serum CA19-9 ( Figure 5). Finally, 130 patients with single organ metastasis were divided into OM cases (n=54) and non-OM cases (n=76). The survival curves according to the OM criteria were shown in Figure  6. OM cases had signi cantly longer OS than non-OM cases (MST: 13.0 months, 95% CI 8.8-18.4 months vs. 8.4 months, 95% CI 6.3-9.7 months, P=0.003). Clinical characteristics of both groups were compared in Table 2. No difference was found in age, gender, operation procedure, serum CEA and tumor location, while signi cantly greater proportion of OM patients underwent chemotherapy (P=0.015). Thus, we compared OM cases with non-OM cases in the cohort that underwent chemotherapy and found that the OM cases showed signi cantly better OS outcomes (Figure 7). On the contrary, there was no difference in survival between the two groups in the cohort that received no chemotherapy.

Discussion
In some malignancies, such as colorectal, kidney and lung cancers, there is growing body of evidence that a metastasectomy can improve survival outcomes of selected patients [4][5][6][7][12][13][14]. For colorectal cancers and kidney cancers, surgical metastasectomy is the treatment of choice in the National Comprehensive Cancer Network (NCCN) guidelines. In non-small-cell lung cancer, Gomez et al. showed local consolidative therapy with radiotherapy or surgery improved both PFS and OS of OM cases [7]. In colorectal cancer, Kobayashi et al. proposed synchronous resection of localized peritoneal metastasis improved survival outocomes [13].
In PDAC with distant metastases, Tachezy M et al. suggested a survival bene t for undergoing simultaneous pancreas and liver resection [15]. Several prior studies have revealed that the resection of lung and liver metastases prolong prognosis in PDAC [15][16][17]. Kandel [9]. They insist that patients with limited metastatic status have a chance to get a good prognosis by metastasectomy, even for PDACs.
Another standard requirement of oligometastases is a low level of serum tumor markers. Although we de ned the optimal cut-off value of preoperative serum CA19-9 as 2000 U/ml, it does not appropriately work for Lewis antigen-negative patients. Luo et al. proposed that CEA and CA125 can be applied as biomarkers in patients with no CA19-9 secretion from PDACs [19]. Wei et al. proposed that tumor makers' criteria was at least a 50% reduction of serum CA125 or CEA levels if the patient had a normal CA19-9 level before conversion chemotherapy [16]. Basically, we need other rules for theses patients.
Our analyses revealed that four or fewer liver metastases have a good prognosis. This is the same cut-off number as previous retrospective reports [9,15,20]. As for peritoneal metastases, some studies mentioned that only localized peritoneal metastasis can be included in oligometastases and can be treated [21,22]. It implies that peritoneal metastasis basically has a poor prognosis, and incomplete metastasectomy does not affect the survival outcomes. In colorectal cancers, synchronous resection of localized peritoneal metastasis improved survival outcomes, while the diffuse or larger size (> 20 mm) peritoneal metastases were independent poor prognostic factors [13]. Staging laparoscopy of PDAC cases sometimes reveals unsuspected peritoneal dissemination, as Karabicak et al. reported (19%) [23]. At that time, it is still unkown peritoneum metastases removal affects the patient's survival or not.
The period when we examined it is long, and there is the change of the standard treatment, too, and chemotherapy varies in this examination. With evolution of the chemotherapy, a recent case tends to have a good prognosis. However, as for the prognostic difference by OM, the developing front of the chemotherapy is more remarkable. It may be said that the situation of tumor determines OM than contents of the treatment intervention.
Treatment of OM cases of PDAC is still controversial. The current study suggests that patients with OM status could bene t from systematic chemotherapy. This may be an important nding that should however be veri ed with a larger cohort of patients. Non-OM patients with su cient performance status should not be denied the opportunity to receive chemotherapy at this time. Since the study population does not include patients who underwent metastasectomy, we could not make any recommendation on whether or not to consider metastasectomy for PDAC. This issue could be solved from the viewpoint of conversion surgery, for which only responders to the chemotherapy are usually considered eligible.
Our study has some limitations. First, patient selection bias can exist because of the retrospective nature of the study. Most of subjects in this cohort had relatively good performance status enough to received palliative or probe laparotomy without postoperative complications. Secondly, we did not evaluate metastases outside of the liver and peritoneal surface because of the small number of such cases. Thirdly, we did not evaluated the case that a bypass does not need with distant metastasis diagnosed without laparotomy. Fouthly, considerations for Lewis antigen-negative patients have not been made at this time. Unlike the case with other types of cancer, we have not discussed on relevance of metastasectomy. This is in part attributable to the particular poor prognosis of PDAC, but metastasectomy could still be an issue for future debate, pending improvements in systemic treatment.
In conclusion, PDAC OM cases can be identi ed by limited visible metastatic sites with moderately low serum CA19-9.

Declarations
Ethics approval and consent to participate All methods were carried out in accordance with relevant guidelines and regulations. The institutional review board of the nagoya University Graduate School of Medicine approved (registration no. 2020-0344) in Oct 2020. By opt-out, informed consent was obtained from all subjects, or from parents and / or legal guardians if subjects were under the age of 18. Survival curves of liver matastatic cases were shown. a: Survival curves of three categories (H1, H2, H3) were drawn (H1 vs. H2 (P<0.001), H2 vs. H3 (P=0.261), H1 vs. H3 (P=0.229)). b: Survival curves of H1 cases and H2-3 cases were compared. c: Survival curves of H1-2 cases and H3 were compared. The survival curve of the cases with two or fewer liver metastases was compared with others. c: The survival curve of the cases with four or fewer liver metastases was compared with others. Figure 4 a: ROC curve analyses were performed for 20 months survivers-speci c setting after non-curative operations. The optimal cut-off of CA19-9 was 2000 U/ml. b: Survival curves of the cases with CA19-9 ≥ 2000 U/ml and CA19-9 < 2000 were compared.

Figure 6
Survival curves of the OM group and the non-OM group were shown.