A retrospective cohort study on the differential overall survival rates between surgical intervention and chemotherapy in stage IV pancreatic cancer patients

Introduction

Pancreatic cancer has the highest mortality rate among cancers and ranks as the third leading cause of cancer-related death (1,2). According to the International Agency for Research on Cancer (IARC), global statistics reported that approximately 495,773 new cases and 466,003 deaths occurred in 2020 (3). Most pancreatic cancer cases are diagnosed at advanced stages, with metastatic disease accounting for approximately 60% of cases, which commonly spreads to the liver (4). For the treatment of pancreatic cancer with liver metastases (PCLM), Hackert et al. proposed that, with careful patient selection, resection of both the primary tumor and oligometastatic lesions offers superior outcomes compared with palliative care (5). In contrast, Tempero et al. reported that surgical intervention provides minimal survival benefit for such patients (6). Despite advancements in technology, both chemotherapy and radiotherapy remain largely ineffective, with median survival rates not exceeding 1 year (7-9). Surgery remains the only curative option for pancreatic cancer. Neoadjuvant therapy may create surgical opportunities, even for patients with PCLM (10). However, nearly 80% of patients do not reach the criteria for local resection following such treatment (11). While surgery is associated with a certain recurrence rate, the combination of surgery and chemotherapy substantially extends survival compared with chemotherapy alone (12).

Given these outcomes, the National Comprehensive Cancer Network (NCCN) and the American Society of Clinical Oncology (ASCO) do not generally recommend resection surgery (13,14). However, some experts argue that with rigorous patient selection (15), surgical treatment could yield favorable outcomes (16,17), and certain clinicians may choose to resect a solitary liver lesion that, although not detected during preoperative evaluation, is incidentally discovered during surgery (18). Thus, the role of surgery in PCLM remains contentious. Preliminary research by our team has indicated that for patients with liver metastasis, combining surgery with chemotherapy leads to improved overall survival (OS) (19). The NCCN suggests that, under strict evaluation, patients with oligometastasis in the liver (defined as three or fewer metastatic lesions) and resectable primary tumors might be candidates for synchronous or staged surgical resection (nccn.org/professionals/physician_gls/pdf/pancreatic.pdf). However, the prevailing recommendation remains focused on the use of chemotherapy, targeted therapies, and symptomatic relief as the primary treatment approaches. Owing to the violation of established guidelines, approval for clinical trials to study this issue is difficult to acquire. Therefore, there is no strong evidence to clarify this problem. To evaluate the potential benefits of surgical intervention for patients with stage IV PCLM, we conducted an analysis via the Surveillance, Epidemiology, and End Results (SEER) database.

SEER includes extensive data on the incidence, survival, and mortality of numerous histopathological cancer subtypes, with an expanding focus on molecular profiling. In this study, we analyzed data from the SEER 17 database spanning from 2010–2021, encompassing 24,802 pancreatic cancer patients aged 18 years and older with liver metastasis. Our investigation aimed to assess the relationship between treatment modalities and OS, specifically exploring whether a combined approach of surgery and chemotherapy is feasible and potentially superior to chemotherapy alone. By analyzing patient survival data collected from the SEER database, we evaluated the differences between surgical and nonsurgical treatment strategies, aiming to elucidate the relationship between therapeutic approaches and OS. In particular, our study sought to assess the feasibility of combining surgery with chemotherapy, investigating whether this dual approach offers superior outcomes compared with chemotherapy alone. Additionally, we aimed to identify patient subgroups most suitable for surgical intervention and those for whom chemotherapy may be a more appropriate treatment modality for sustained management. We present this article in accordance with the STROBE reporting checklist (available at https://gs.amegroups.com/article/view/10.21037/gs-2025-269/rc).

Methods

Patient

We extracted data from the SEER 17 Registry research database, one of the most recent sources available, which contains information from 2004–2021. The SEER 17 database encompasses cancer registries across the United States, including both major urban centers and rural areas. Cancer data collection is initiated by identifying patients diagnosed with or receiving cancer care through hospitals, outpatient facilities, radiology centers, physician offices, laboratories, surgical centers, and other providers involved in cancer diagnosis or treatment, such as pharmacists. Legislation across all 50 states mandates the reporting of newly diagnosed cancers to a central registry. When they do, registries extract relevant cancer information from patient medical records according to the North American Association of Central Cancer Registries’ (NAACCR) Data Standards External Web Site Policy.

The case list was generated via SEERStat version 8.4.0, a specialized software tool available at SEER.cancer.gov/seerstat. This dataset includes the following variables: age, sex, race, year of diagnosis, primary site, histology, tumor (T) stage, node (N) stage, radiotherapy (yes, no), bone metastases (yes, no), brain metastases (yes, no), lung metastases (yes, no), sequence number, marital status, and grade. We initially identified 235,116 pancreatic cancer patients and subsequently excluded 31,749 patients from 2004-2009. We further excluded 91,656 cases that were either unstaged or classified as stage I or II. An additional 190 patients were omitted because of age (<18 years) or missing OS data. We excluded 59,538 cases involving stage III patients, those without liver metastasis, or those with an unknown liver metastasis status. Another 27,181 patients were removed because they had not undergone chemotherapy. Ultimately, a total of 24,802 patients were included in our study cohort, which was then divided into two groups: those receiving surgery plus chemotherapy (N=686) and those receiving chemotherapy alone (N=24,116). Importantly, detailed information, such as surgical techniques, the timing of treatments, and whether the therapy paradigm was sequential or synchronous, was unavailable (Figure 1). Figure 1 comprehensively outlines our inclusion and exclusion criteria. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Figure 1 Flowchart used to screen the participants. OS, overall survival; TNM, tumor-node-metastasis.

Statistical analysis

We defined OS as the interval between the date of pancreatic cancer diagnosis and the date of death from any cause (Figure 2). Unadjusted univariate analysis was employed to identify covariates associated with OS, whereas stratified analysis (adjusted) was performed to ascertain the impact of these covariates on OS within each subgroup. Survival curves for stage IV pancreatic cancer patients aged 18 and older, comparing surgical and nonsurgical groups, were generated via the Kaplan-Meier method. Additionally, Kaplan-Meier all-cause survival curves and OS curves stratified by metastasis status and other variables were used to evaluate the effects of different treatment modalities on patient survival across various subgroups (Figure 3).

Figure 2 Kaplan-Meier curves before and after propensity score matching. (A) Overall survival stratified by treatment modality among patients ≥18 years old with PCLM. (B) Overall survival following propensity score matching. PCLM, pancreatic cancer with liver metastasis.

Figure 3 Overall survival stratified by treatment modality among patients ≥18 years old with PCLM in different stratifications. (A1,A2) Stratified by age; (B1,B2) stratified by bone metastases; (C1,C2) stratified by brain metastases; (D1,D2) stratified by lung metastases. PCLM, pancreatic cancer with liver metastasis.

Overall, our analysis encompasses a large, multicenter database of 24,802 cases, offering a robust dataset for prognostic assessment in PCLM. The study protocol received approval from the Ethics Committee of West China Hospital, Sichuan University, ensuring adherence to ethical principles and regulations. All analyses were conducted via Empower (R) (www.empowerstats.com, X&Y Solutions, Inc., Boston, MA, USA) and R version 3.6.3 (http://www.R-project.org). Empower Stats, a statistical software built on the R language, provides advanced data processing and analytical capabilities. Statistical significance was defined as P<0.05.

Propensity score matching (PSM)

To mitigate the selection bias inherent in this observational study, we performed PSM analysis. A logistic regression model was used to estimate each patient’s propensity score for receiving surgical treatment, incorporating the following covariates: age, sex, race, year of diagnosis, primary tumor site, histology, T stage, N stage, radiotherapy status, presence of bone/brain/lung metastases, sequence number, marital status, and tumor grade.

A 1:4 nearest neighbor matching algorithm was employed with a caliper width of 0.02 standard deviations of the logit of the propensity score, without replacement. The quality of matching was assessed via standardized mean differences (SMDs), with SMDs <0.1 indicating good balance between groups. Survival analyses, including Kaplan-Meier curves and Cox proportional hazards regression, were repeated in the matched cohort. The attenuation of the treatment effect after PSM was calculated as the relative reduction in survival compared with the unmatched analysis.

Results

Baseline characteristics of the study subjects by treatment modality

The characteristics of all patients are summarized in Table 1. Among the 24,802 pancreatic cancer patients, the subjects were stratified into two groups on the basis of treatment modality: chemotherapy alone (N=24,146, 97.2%) and chemotherapy combined with surgery (N=686, 2.7%).

The distributions of variables such as age, sex, race, year of diagnosis, primary site, histology, T stage, N stage, radiotherapy, bone metastases, brain metastases, lung metastases, sequence number, marital status, and grade were comparable between the groups. Tumor TNM staging was strongly associated with the likelihood of surgical intervention: 51.5% of patients with T2 tumors and 50.6% with N1 staging underwent surgery. Compared with patients receiving only chemotherapy, those who underwent surgery tended to be younger. For patients with bone or brain metastases, the P value was >0.05, likely due to the limited sample size for these subgroups. Our study revealed that among stage IV PCLM, the median OS for those receiving only chemotherapy was 6 months, whereas patients who underwent surgery had a median OS of 18 months.

Subgroup analysis of the association between treatment and OS

We conducted a stratified analysis and intergroup comparison based on variables including sex; race (black vs. white); age; year of diagnosis; tumor location; grade (I/II, III/IV); bone, brain, and pulmonary metastasis; pathological subtype [pancreatic ductal adenocarcinoma (PDAC) vs. other]; T stage (1/2/3/4); N stage (0/1/2); chemotherapy; and radiotherapy. Our analysis revealed that, for patients with bone metastases as well as those of African descent, surgical intervention does not confer superior outcomes compared with nonsurgical treatments [hazard ratio (HR): 1.0, 0.6]. Intriguingly, the mortality risk among Black patients is lower than that among White patients. With respect to the year of diagnosis, patients diagnosed prior to 2015 presented a greater mortality risk than those diagnosed thereafter. With respect to tumor location, proximity to the pancreatic head is associated with a significantly elevated risk of mortality. In terms of tumor grade, early-stage tumors are clearly linked to a reduced risk of mortality. Most notably, patients with metastases to other organs—whether the brain, lungs, or bones—face a substantially increased risk of death (Table S1).

Cox proportional hazards regression analysis of the relationship between treatment modality and OS

In the Cox proportional hazards regression analysis (Table 2), with OS as the outcome variable, we fully adjusted for factors including age, sex, race, year of diagnosis, primary site, histology, T stage, N stage, radiotherapy, bone metastases, brain metastases, lung metastases, sequence number, marital status, and tumor grade. The results demonstrated that patients who underwent surgery combined with chemotherapy had significantly better OS than did those who received chemotherapy alone (HR =0.4, 95% CI: 0.4–0.5; P<0.001) in the unmatched cohort.

Several other variables emerged as independent prognosors of survival. Age ≥65 years was associated with increased mortality risk (HR =1.3), whereas patients with tumors located in the pancreatic body (HR =0.5) or tail (HR =0.4) had a better prognosis than those with head tumors. Advanced T stage (T3–4 vs. T1–2: HR =1.2), the presence of bone metastases (HR =1.3), and lung metastases (HR =1.4) were associated with worse survival. Notably, poorly differentiated tumors (grades III–IV vs. I–II: HR =2.8) represented the strongest independent predictor of mortality.

Kaplan-Meier analysis

In the Kaplan-Meier survival analysis, patients who received combined surgical intervention and chemotherapy exhibited a significant extension in OS compared with those treated with chemotherapy alone (median OS: 18 vs. 6 months; P<0.001; Figure 2A). The median OS was 21 months (95% CI: 18–25) in patients aged under 65 years, 13 months (95% CI: 12–16) in those aged 65 years or older, and 13 months (95% CI: 8–18) and 8 months (95% CI: 3–13) in patients with bone, brain, or lung metastases, respectively. The group under 65 years exhibited the highest OS rate. Across all stratified subgroups, the combined surgery and chemotherapy group consistently presented superior OS curves than did the chemotherapy-only group.

PSM analysis

To address potential selection bias, we performed 1:4 PSM using 15 baseline covariates (Table 1). Among the 686 surgery patients, 645 (94.0%) were successfully matched with 2,580 nonsurgery patients. Before matching, substantial imbalances were observed in multiple variables, including T stage (SMD =0.475), N stage (SMD =0.588), histology (SMD =0.431), lung metastases (SMD =0.434), and particularly tumor grade (SMD =1.005).

After PSM, significant improvements in covariate balance were achieved for several variables. Age (SMD: 0.257→0.027), sex (SMD: 0.002→0.009), histology (SMD: 0.431→0.011), radiotherapy (SMD: 0.188→0.004), and marital status (SMD: 0.085→0.064) all achieved good balance (SMD <0.1). However, several key prognostic factors remained inadequately balanced, including T stage (SMD =0.435), N stage (SMD =0.488), primary site (SMD =0.253), lung metastases (SMD =0.172), sequence number (SMD =0.174), and tumor grade (SMD =0.123). Overall, only 8 of 15 covariates (53.3%) achieved SMD <0.1 after matching.

In the matched cohort (Tables 2,3), the surgery group had a significantly longer median OS than the nonsurgery group did (17.0 vs. 8.0 months, P<0.001; Figure 2A,2B). However, the survival difference was attenuated from 12.0 months in the unmatched cohort to 9.0 months after PSM, representing a 25.0% reduction. Cox regression analysis of the matched cohort yielded a HR of 0.46 (95% CI: 0.42–0.51, P<0.001) compared with 0.4 (95% CI: 0.4–0.5) before matching. The persistent survival advantage in the surgery group after PSM, despite adjustment for measured confounders, may reflect either (I) the true benefit of surgery in highly selected patients or (II) residual confounding from inadequately balanced covariates and unmeasured factors such as performance status, exact number and size of liver metastases, and response to initial chemotherapy.

Discussion

Impact of selection bias and PSM

In this retrospective analysis utilizing the SEER 2004–2021 database, we observed that patients who underwent surgical treatment demonstrated substantially longer OS than those receiving chemotherapy alone did (18 vs. 6 months, HR =0.4). However, the most critical finding of our study is the demonstration of profound selection bias. Only 2.8% (686/24,802) of patients underwent surgery, indicating highly selective criteria not captured in the SEER database. This extreme disparity strongly suggests that surgeons select patients with favorable characteristics—likely superior performance status, limited metastatic burden (oligometastatic disease with ≤3 lesions), favorable tumor biology, and good response to initial chemotherapy—all of which are unmeasured confounders that independently predict better survival.

To quantify the magnitude of selection bias, we performed PSM using 15 baseline covariates. The matching achieved a 94.0% success rate (645/686), creating a more comparable control group of 2,580 patients. Critically, the survival difference was attenuated by 25.0% after PSM (from 12 to 9 months), confirming the substantial impact of the measured confounders. However, even after PSM, only 53.3% of the covariates achieved good balance (SMD <0.1). Notably, key prognostic factors remained inadequately balanced: T stage (SMD =0.435), N stage (SMD =0.488), primary site (SMD =0.253), lung metastases (SMD =0.172), and tumor grade (SMD =0.123). These unbalanced variables are strong predictors of survival in patients with pancreatic cancer, indicating persistent residual confounding even after PSM.

More importantly, the SEER database lacks critical prognostic variables that likely drive surgical selection decisions: Eastern Cooperative Oncology Group (ECOG) performance status, exact number and anatomic distribution of liver metastases, size of metastatic lesions, response to neoadjuvant chemotherapy, carbohydrate antigen 19-9 (CA19-9) levels, and nutritional status. These unmeasured confounders, which cannot be adjusted for in any retrospective analysis, represent the fundamental limitation of observational studies in this setting. The observed survival difference, therefore, likely predominantly reflects which patients were selected for surgery rather than the causal effect of surgery itself. This represents classic confounding by indication, where treatment assignment is determined by prognosis, rendering the surgery and chemotherapy-only groups inherently incomparable despite statistical adjustment.

Interpretation in the context of existing literature

Our findings must be interpreted cautiously in light of these methodological limitations. Current guidelines do not recommend surgical intervention for patients with PCLM (13,20). Considerable controversy remains regarding whether these patients should undergo surgical intervention. Some scholars, such as Shrikhande et al. and Ouyang et al., contend that surgery can significantly increase OS in these patients (21,22). Notably, tumors originating in the pancreatic tail (median OS: 16.8 months) have a median survival time closely aligned with that observed in patients within the SEER database (23). However, some scholars, particularly in earlier studies, argued that surgery involving the resection of both the primary tumor and metastatic lesions does not yield better outcomes for patients. This approach yields outcomes indistinguishable from those achieved with chemotherapy alone (24,25). A primary limitation of these studies is the insufficient sample size and lack of rigorous patient evaluation and selection, resulting in significant selection bias. Larger, more comprehensive patient cohorts should be selected to further validate the effectiveness of surgical treatment. In addressing this contentious issue, the vast majority of clinicians now opt for a more conservative approach, favoring chemotherapy alone to prolong patient OS. This propensity is attributable primarily to the exceptionally poor OS and high recurrence rates associated with pancreatic cancer, coupled with the formidable risks posed by surgical intervention, including postoperative complications (6,26,27). Poor patient prognosis and severe declines in quality of life, coupled with the lack of rigorous screening to assess surgical feasibility, have further contributed to this approach. In the current era of immunotherapy, advancements in the treatment of pancreatic cancer have been relatively slow, primarily due to its classification as a “cold tumor” with significant infiltration by immunosuppressive cells. This low immunogenicity and noninflammatory phenotype of PDAC contribute to the limited therapeutic options available for pancreatic cancer. Professor Yang reported that simultaneous resection of the primary tumor and oligometastatic lesions in PCLM patients can provide clinical benefits, particularly for those with primary tumors located in the pancreatic body and tail (23).

With the growth of the economy, particularly the digital economy, recent advancements in surgical instruments, refined surgical techniques, precise localization, improved preoperative diagnostics, and enhanced postoperative recovery conditions, surgical risks have diminished, leading to greater acceptance of surgical interventions among patients. A consensus on this issue has not yet been reached. However, with an increasing number of researchers adopting a favorable perspective, we undertook further analysis using the SEER database. Therefore, we utilized the SEER database, which includes data on treatment, metastatic sites, and metastatic pathways, to conduct univariate, multivariate, and stratified analyses to address these limitations. In our study, a multivariate risk-adjusted analysis based on surgical intervention demonstrated that surgical treatment significantly improved OS compared with nonsurgical approaches. This finding strongly substantiates the substantial efficacy of surgery, suggesting that, relative to the side effects of chemotherapy, surgical treatment may result in less suffering for patients. Owing to the advantages of enhanced therapeutic efficacy and reduced side effects, surgical intervention undoubtedly emerges as a promising treatment avenue for further exploration. However, our dataset does not specify the surgical modalities undertaken by patients, leaving it unclear whether the treatments adhered to guidelines recommending combined resection of both primary tumors and metastatic lesions for oligometastatic cases. Some patients may have undergone palliative surgeries, radiofrequency ablation, or cryoablation, all of which could impact outcomes. Whether these methods yield better or worse outcomes than open surgery remains undetermined, and the SEER database lacks such data, precluding the inclusion of these variables in our analysis. Nonetheless, this limitation does not necessarily undermine the robustness of our conclusions.

For patients with PCLM being considered for surgical treatment, rigorous selection criteria are essential for those with resectable primary lesions who are surgically fit. Specifically, candidates should have no metastases outside the liver and no more than three hepatic lesions. Ideally, metastatic sites would be confined to a single liver lobe with smaller, well-defined lesions, as these characteristics suggest lower tumor aggressiveness, potentially more localized spread, and a greater likelihood of achieving complete resection in a single operation (28). However, the size of metastatic lesions is not an absolute determinant of surgical eligibility. On the basis of extensive data from the SEER database, we propose that, in highly selected patients, combined resection of primary and metastatic lesions is likely to confer substantial benefit, potentially significantly extending OS. Regardless of the surgical approach, chemotherapy should be administered (6). For patients who are initially inoperable, this remains the most effective treatment option. In cases where chemotherapy yields favorable outcomes, the subsequent application of surgical intervention may significantly increase the likelihood of achieving superior therapeutic results (28). We also compared the postoperative outcomes of other stage IV cancers, such as those reported by Sara Benitez Majano in her study on stage IV colorectal cancer patients in Norway. Her findings indicate a remarkable 3-year survival rate of 33.0% (31.0–35.1%), with even the lowest survival rate observed in Sweden, reaching 26.7% (24.7–28.8%) (29). This is notably superior to the prognosis for pancreatic cancer, which is likely influenced by factors such as the extent of local invasion, the relationship with surrounding tissues, the presence of more prominent symptoms, and the possibility of earlier detection. Furthermore, Benitez Majano’s study underscores the significant impact of age, revealing that the proportion of patients aged 75 years or older undergoing resection is consistently lower than that of younger patients (29). This disparity increases with advancing age, a phenomenon that resonates with our own findings. It is largely attributable to patient preferences, individual physical conditions, and the capacity to tolerate subsequent treatments. In our study, we observed that the anatomical location of pancreatic tumors does not significantly affect postoperative prognosis; the most substantial factor influencing outcomes was tumor grade. Notably, when the grade was low, the prognosis resembled that of nonsurgical patients, suggesting that such findings could broaden our surgical indications. Conversely, for patients with higher-grade tumors, a more cautious approach is warranted, and careful preoperative evaluation is crucial in optimizing survival outcomes. Our discussion led us to propose several potential explanations for the OS benefit of surgery for stage IV pancreatic cancer patients over chemotherapy alone: (I) reduction in tumor burden, surgical resection of both primary and metastatic lesions can significantly alleviate the tumor burden while also mitigating symptoms resulting from local tissue invasion. This leads to notable improvements in both the physical and psychological well-being of patients. (II) Enhanced chemotherapy efficacy: postoperative resection may increase the effectiveness of chemotherapy. Pancreatic cancer is often referred to as a “cold tumor”, and resection may, to some extent, facilitate the release of immune factors that can induce effects akin to those observed with cryotherapy. Additionally, the reduction in tumor load allows for a more concentrated effect of chemotherapeutic agents on micrometastatic sites. (III) Heterogeneity in tumor biology: the biological variability of pancreatic cancer is considerable, with different patients displaying varying levels of sensitivity to chemotherapy. However, in cases where surgery is performed without postoperative complications, it can provide symptomatic relief, offering a greater benefit than chemotherapy alone.

Limitations and implications

This study has several critical limitations that must be acknowledged. First and foremost, the retrospective observational design precludes causal inference. Despite PSM, we cannot exclude substantial residual confounding from unmeasured variables. The SEER database lacks essential information, including (I) performance status (ECOG/Karnofsky scores), which is perhaps the strongest predictor of surgical candidacy and survival; (II) precise characteristics of liver metastases (number, size, bilobar vs. unilobar distribution); (III) specific chemotherapy regimens and response status; (IV) surgical details (R0/R1/R2 resection status, extent of hepatectomy, perioperative complications); (V) timing of treatments (neoadjuvant vs. adjuvant chemotherapy, interval between chemotherapy and surgery); and (VI) key laboratory values (CA19-9, albumin, bilirubin). These unmeasured factors collectively represent overwhelming confounders that cannot be addressed through any statistical adjustment of retrospective data.

Second, the 25% attenuation of survival benefit after PSM, combined with failure to achieve balance in 47% of matched covariates, strongly suggests that much of the observed survival difference reflects patient selection. The inadequately balanced variables—T stage, N stage, primary tumor site, lung metastases, and tumor grade—are precisely those that determine tumor biology and prognosis. Patients with lower T/N stages, better tumor differentiation, and the absence of extrahepatic metastases inherently have better prognoses regardless of treatment. Our inability to balance these factors despite PSM indicates that the surgery group fundamentally comprises patients with more favorable disease biology.

Third, the heterogeneity in surgical procedures constitutes a major limitation. The SEER database does not distinguish between curative-intent resections (combined resection of the primary tumor and liver metastases), palliative surgeries (tumor debulking, biliary bypass), ablative procedures (radiofrequency ablation, microwave ablation), or combinations thereof. Some patients may have undergone limited hepatic metastasectomy for oligometastatic disease per the NCCN recommendations, whereas others may have had only palliative interventions. This procedural heterogeneity introduces substantial noise and potentially dilutes any true treatment effect.

Fourth, and most fundamentally, prospective randomized controlled trials are ethically and practically unfeasible for this purpose. Randomizing patients with oligometastatic PCLM to surgery versus chemotherapy alone would be considered unethical by many clinicians, given the emerging evidence of potential benefit. Conversely, the poor overall prognosis of PCLM (median OS <1 year with chemotherapy alone) creates recruitment challenges. This impasse leaves observational studies as the only available evidence source, with their inherent limitations.

Clinical implications and future directions

Despite these profound limitations, our findings provide insights for clinical decision-making and research directions. In current clinical practice, surgery for PCLM should remain highly selective and restricted to patients meeting stringent criteria: (I) excellent performance status (ECOG 0–1); (II) oligometastatic disease with ≤3 hepatic lesions, ideally confined to one lobe; (III) technically resectable primary tumor; (IV) absence of extrahepatic metastases; (V) favorable response or stable disease on neoadjuvant chemotherapy; and (VI) adequate hepatic reserve for resection. Even among this highly selected group, the decision should involve multidisciplinary tumor board discussion and thorough informed consent regarding limited evidence and substantial recurrence risk.

Conclusions

In this large retrospective cohort study of 24,802 patients with pancreatic cancer and liver metastasis, we observed that highly selected patients (2.8%) who underwent surgery combined with chemotherapy had longer median OS than did those who underwent chemotherapy alone (18 vs. 6 months before matching; 17 vs. 8 months after PSM). However, PSM revealed a 25% attenuation of survival benefit and failure to achieve balance in key prognostic covariates, indicating that the observed survival difference likely reflects patient selection rather than a causal treatment effect.

The SEER database lacks critical variables, including performance status, precise metastatic burden, treatment sequence, and response to chemotherapy, all of which are likely to determine surgical candidacy and independently affect survival. Therefore, while surgery may benefit exceptionally well-selected patients with oligometastatic disease, the observed survival advantage in this retrospective analysis predominantly reflects the identification of patients with favorable tumor biology and clinical characteristics rather than the efficacy of surgical intervention per se.

In clinical practice, surgical resection for PCLM remains a highly selective option and is restricted to patients with excellent performance status, limited hepatic oligometastases (≤3 lesions), technically resectable primary tumors, no extrahepatic disease, and a favorable response to neoadjuvant chemotherapy. Even among this select group, patients should be counseled that evidence supporting surgery is limited to retrospective analyses with profound selection bias and that prospective data are lacking. Prospective multicenter registries with comprehensive data collection are urgently needed to better define the role of surgery and identify optimal candidates for combined modality treatment in this challenging clinical scenario.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://gs.amegroups.com/article/view/10.21037/gs-2025-269/rc

Peer Review File: Available at https://gs.amegroups.com/article/view/10.21037/gs-2025-269/prf

Funding: This study was supported by Sichuan Science and Technology Program (Nos. 2023NSFSC0726 and 2024NSFSC0740).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://gs.amegroups.com/article/view/10.21037/gs-2025-269/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Siegel RL, Giaquinto AN, Jemal A. Cancer statistics, 2024. CA Cancer J Clin 2024;74:12-49. [Crossref] [PubMed]
2. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
3. Sung H, Ferlay J, Siegel RL, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 2021;71:209-49. [Crossref] [PubMed]
4. Vincent A, Herman J, Schulick R, et al. Pancreatic cancer. Lancet 2011;378:607-20. [Crossref] [PubMed]
5. Hackert T, Niesen W, Hinz U, et al. Radical surgery of oligometastatic pancreatic cancer. Eur J Surg Oncol 2017;43:358-63. [Crossref] [PubMed]
6. Tempero MA, Malafa MP, Al-Hawary M, et al. Pancreatic Adenocarcinoma, Version 2.2021, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw 2021;19:439-57. [Crossref] [PubMed]
7. Al Efishat M, Wolfgang CL, Weiss MJ. Stage III pancreatic cancer and the role of irreversible electroporation. BMJ 2015;350:h521. [Crossref] [PubMed]
8. Werner J, Combs SE, Springfeld C, et al. Advanced-stage pancreatic cancer: therapy options. Nat Rev Clin Oncol 2013;10:323-33. [Crossref] [PubMed]
9. Neoptolemos JP, Kleeff J, Michl P, et al. Therapeutic developments in pancreatic cancer: current and future perspectives. Nat Rev Gastroenterol Hepatol 2018;15:333-48. [Crossref] [PubMed]
10. Del Chiaro M, Sugawara T, Karam SD, et al. Advances in the management of pancreatic cancer. BMJ 2023;383:e073995. [Crossref] [PubMed]
11. Park W, Chawla A, O'Reilly EM. Pancreatic Cancer: A Review. JAMA 2021;326:851-62. [Crossref] [PubMed]
12. Casolino R, Braconi C, Malleo G, et al. Reshaping preoperative treatment of pancreatic cancer in the era of precision medicine. Ann Oncol 2021;32:183-96. [Crossref] [PubMed]
13. Sohal DPS, Kennedy EB, Khorana A, et al. Metastatic Pancreatic Cancer: ASCO Clinical Practice Guideline Update. J Clin Oncol 2018;36:2545-56. [Crossref] [PubMed]
14. Fu N, Jiang Y, Weng Y, et al. Worth it or not? Primary tumor resection for stage IV pancreatic cancer patients: A SEER-based analysis of 15,836 cases. Cancer Med 2021;10:5948-63. [Crossref] [PubMed]
15. Wray CJ, Ahmad SA, Matthews JB, et al. Surgery for pancreatic cancer: recent controversies and current practice. Gastroenterology 2005;128:1626-41. [Crossref] [PubMed]
16. Birnbaum DJ, Turrini O, Vigano L, et al. Surgical management of advanced pancreatic neuroendocrine tumors: short-term and long-term results from an international multi-institutional study. Ann Surg Oncol 2015;22:1000-7. [Crossref] [PubMed]
17. Jin K, Xu J, Chen J, et al. Surgical management for non-functional pancreatic neuroendocrine neoplasms with synchronous liver metastasis: A consensus from the Chinese Study Group for Neuroendocrine Tumors (CSNET). Int J Oncol 2016;49:1991-2000. [Crossref] [PubMed]
18. Sargent M, Boeck S, Heinemann V, et al. Surgical treatment concepts for patients with pancreatic cancer in Germany--results from a national survey conducted among members of the "Chirurgische Arbeitsgemeinschaft Onkologie" (CAO) and the "Arbeitsgemeinschaft Internistische Onkologie" (AIO) of the Germany Cancer Society (DKG). Langenbecks Arch Surg 2011;396:223-9. [Crossref] [PubMed]
19. Strobel O, Neoptolemos J, Jäger D, et al. Optimizing the outcomes of pancreatic cancer surgery. Nat Rev Clin Oncol 2019;16:11-26. [Crossref] [PubMed]
20. Tempero MA. NCCN Guidelines Updates: Pancreatic Cancer. J Natl Compr Canc Netw 2019;17:603-5. [Crossref] [PubMed]
21. Shrikhande SV, Kleeff J, Reiser C, et al. Pancreatic resection for M1 pancreatic ductal adenocarcinoma. Ann Surg Oncol 2007;14:118-27. [Crossref] [PubMed]
22. Ouyang HH, Pan ZY, Ma WD, et al. Multidisciplinary treatment and survival analysis for 497 cases of pancreatic cancer with liver metastases. Zhonghua Yi Xue Za Zhi 2016;96:425-30. [Crossref] [PubMed]
23. Yang J, Zhang J, Lui W, et al. Patients with hepatic oligometastatic pancreatic body/tail ductal adenocarcinoma may benefit from synchronous resection. HPB (Oxford) 2020;22:91-101. [Crossref] [PubMed]
24. Dünschede F, Will L, von Langsdorf C, et al. Treatment of metachronous and simultaneous liver metastases of pancreatic cancer. Eur Surg Res 2010;44:209-13. [Crossref] [PubMed]
25. Takada T, Yasuda H, Amano H, et al. Simultaneous hepatic resection with pancreato-duodenectomy for metastatic pancreatic head carcinoma: does it improve survival? Hepatogastroenterology 1997;44:567-73.
26. Rahib L, Wehner MR, Matrisian LM, et al. Estimated Projection of US Cancer Incidence and Death to 2040. JAMA Netw Open 2021;4:e214708. [Crossref] [PubMed]
27. Rahib L, Smith BD, Aizenberg R, et al. Projecting cancer incidence and deaths to 2030: the unexpected burden of thyroid, liver, and pancreas cancers in the United States. Cancer Res 2014;74:2913-21. [Crossref] [PubMed]
28. Liu Z, Gou A, Wu X. Liver metastasis of pancreatic cancer: the new choice at the crossroads. Hepatobiliary Surg Nutr 2023;12:88-91. [Crossref] [PubMed]
29. Benitez Majano S, Di Girolamo C, Rachet B, et al. Surgical treatment and survival from colorectal cancer in Denmark, England, Norway, and Sweden: a population-based study. Lancet Oncol 2019;20:74-87. [Crossref] [PubMed]