Is there a chance for curative treatment for metastatic pancreatic adenocarcinoma? A systematic review with meta-analysis

Abstract

Introduction

Pancreatic adenocarcinomas (PDAC) have a poor prognosis, with a 5-year relative Survival rate of 11.5%. Only 20% of patients are initially eligible for resection, and 50% of patients presented with metastatic disease, currently only candidates’ palliative treatment. This meta-analysis compares the outcomes of surgical treatment versus chemotherapy or palliative care for M-PDAC.

Methods

A literature search was performed on May, 9th of 2024. A meta-analysis was then conducted. The primary outcome was Overall Survival and the secondary outcomes were 1-year Survival and 3-year survival. Subgroup analysis was also performed, based on metastatic sites.

Results

The analyses showed a significant benefit of surgical treatment in terms of overall survival (HR: 0.42, 95% CI [0.33– 0.53]), 1-year survival (OR: 0.37, CI 95% [0.26, 0.52]), and 3-year survival (OR: 0.16, CI 95% [0.07, 0.383]). In the subgroup analysis based on metastatic site, the benefit of surgical treatment persisted for liver-only metastases (HR : 0.40, CI 95% [0.29, 0.53]), but not for lung-only metastases.

Conclusion

This meta-analysis showed a survival advantage for surgical treatment in patients with M-PDAC and especially in liver-only metastases. Prospective trials are needed to confirm these findings and refine patient selection criteria for surgical treatment.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00423-025-03778-6.

Introduction

Pancreatic ductal adenocarcinoma (PDAC) has a poor prognosis because of its aggressive biology and late discovery. Despite being the 12th most common cancer worldwide, it ranks as the 7th in cancer- related mortality, with a 5-year relative Survival rate as low as 11.5% [1, 2]. Surgery remains the cornerstone of curative treatment enhancing Survival rates. However, only 20% of patients are initially qualified for resection, with another 30% possibly becoming eligible after neoadjuvant therapy [3]. This leaves out 50% of patients with metastatic disease, currently only candidates’ palliative treatment. The American Cancer Society has reported a 5-year Survival rate of 2.9%, for stage IV pancreatic cancer [4]. Despite the low survival rate, it remains significant, likely due to this group’s heterogeneity. We can assume that some stage IV patients may have characteristics that contribute to better survival, and performing a more thorough selection might even lead to the proposal of curative treatment. Several small series and case reports described prolonged survival after ablative treatments for metastatic pancreatic adenocarcinoma [5–7]. These reports suggested that stratification of the metastatic site could be an important parameter for outcome prediction following ablative treatment. To our knowledge, this is the first systematic review with meta-analysis aimed at investigating the outcomes of ablative treatments versus chemotherapy and/or palliative treatment for metastatic PDAC stratified by the metastatic site.

Methods

Inclusion and exclusion criteria

Types of studies

Due to the lack of randomized trials, we considered observational studies that compared surgical resection or other ablative treatments (radiofrequency/microwave) with chemotherapy or palliative treatment for the metastasis of pancreatic ductal adenocarcinoma. We excluded studies published as short abstracts that lacked sufficient data for the method assessment.

Types of participants

Adults (≥ 18 years old) with metastatic PDAC were included in the study. The metastases were either synchronous or metachronous. We excluded studies that included patients with metastases other than liver, lung, peritoneum or lymph node (i.e. : bone, brain, etc.), studies that included other histologic types of cancer than ductal adenocarcinoma (i.e. : neuroendocrine, acinar, etc.) and studies that included cancer locations other than the pancreas (i.e. : ampulla, bile duct).

Types of interventions

The included studies had at least one treatment arm using ablative treatment, including surgical resection and thermal ablation with either radiofrequency ablation (RFA) or microwave ablation (MWA).

Types of outcome measures

We sought information on the study’ characteristics, including the first author’s name, country, year of publication, publishing journal, study design, sample size, and recruitment period.

We also gathered participant characteristics, including the timing of liver or lung metastases (synchronous vs. metachronous), the number and dimensions of metastases (oligo- vs. polymetastases), the number of patients in each group, the types of ablative treatment (resection/RFA/MWA), types of surgical resection of metastases (wedge resection/segmentectomy/bisegmentectomy/lobectomy/hemihepatectomy/trisegmentectomy), the surgical approach for the primary lesion (pancreaticoduodenectomy (PD)/pylorus preserving pancreaticoduodenectomy(PPPD)/distal pancreatectomy(DP)/total pancreatectomy(TP)), the type and duration of neoadjuvant and adjuvant treatments, palliative chemotherapy details, and follow- up.

Primary outcome

The primary outcome was Overall Survival expressed as Hazard Ratio of death.

Secondary outcomes

The secondary outcomes were 1 Year-Survival and 3 Year-Survival.

Search strategy

Two authors (TK and MAT) performed bibliographic research on May 9, 2024, through MEDLINE (via PubMed), EMBASE (via Ovid), and CENTRAL (the Cochrane library). We restricted our search to articles published in the English language. References from selected papers were used to search for additional articles. Additionally, the two authors searched (clinicaltrials.gov/) and the International Clinical Trials Registry Platform (apps.who.int/trial search/) to scan ongoing or completed clinical trials.

See supplementary 1 for the MEDLINE and EMBASE search strategies.

Data collection and analysis

Selection of studies

Two authors (TK and MAT) used independently used the COVIDENCE Software (https://app.covidence.org) to screen the articles identified in the bibliographic search. Disagreements were resolved through discussion or the intervention of a third author (TP) when doubt persisted. The titles and abstracts were first screened to exclude articles that did not meet the inclusion criteria. The full texts of the remaining papers were then analyzed to determine their inclusion in the review.

Data extraction and entry

Two review authors (TK and MAT) independently extracted the data from each retained study using the COVIDENCE Software (https://app.covidence.org). Disagreements were resolved by discussion. Data suitable for the meta-analysis were entered into the Review Manager (RevMan), and a third author (TP) cross-checked the entries.

Assessment of risk of bias in the included studies

The Risk Of Bias in Non-Randomized Studies of Interventions (ROBINS-I) tool was applied to assess the risk of bias in observational controlled trials [8]. Discrepancies were resolved through consultation with a third reviewer (TP). The following domain were assessed for each study:

• - Bias due to confounding: We evaluated whether the observed effect might differ from the true effect due to uncontrolled confounding factors.

• - Bias in patient selection: We examined whether participants were included in the study based on characteristics that emerged after the intervention began, potentially influencing both the intervention and outcome measures. We considered at risk of selection bias if selection into the study was related to both the intervention and outcome.

• - Bias in classification of interventions.: We assessed whether the intervention groups were clearly defined and distinct without ambiguity.

• - Bias due to deviations from intended interventions: We evaluated whether the analysis was made in the intention to treat.

• - Bias due to missing data: We evaluated whether the available data were sufficient to ensure robustness in the findings, and if the proportion of missing data and reasons for missing data were comparable across intervention groups.

• - Bias in measurement of outcomes: All included studies were categorized as having low risk due to the nature of our primary and secondary outcomes (survival).

• - Bias in the selection of the reported result: We evaluated whether the reported effect estimate was chosen based on the results. We assumed a low risk of varying outcome measurements for our review, but checked if multiple subgroup estimates were selectively reported.

Measures of treatment effect

We used Review Manager (RevMan), with a random-effects model to generalize our results to the universe of possible studies.

To evaluate overall survival, we calculated the hazard ratio (HR) with 95% confidence intervals (95% CI) and to investigate 1-year and 3-year survival, we calculated the Odds Ratio (OR) with 95% confidence intervals (95% CI).

Finally, we conducted a subgroup analysis based on metastatic site to investigate the role of surgery for lesions confined to the liver or lungs.

Subgroups analyses based on the size, the number of metastases and the type and duration of chemotherapy were not possible due to the lack of data to analyse.

Assessment of heterogeneity

We visually examined heterogeneity by studying Forrest plots. Heterogeneity was suspected if the confidence intervals overlapped slightly or not at all. To quantify heterogeneity, I square statistic (I2) was used. Heterogeneity was considered low when I2 was between 0 and 40%, intermediate when it was between 40% and 60%, and high when it was between 60% and 100%. If high or intermediate heterogeneity was found, we performed sensitivity analysis and excluded studies with a serious or critical risk of bias.

Results

Study selection (Fig. 1 Flow diagram)

Fig. 1.

PRISMA flow-chart of the bibliographic research

One thousand four hundred and sixty-six references were imported from the databases and registers for screening. Additionally, 23 articles were found by searching the references of the selected articles and by searching the grey literature. After removing duplicates, 1385 remaining studies were screened by title and abstract. One thousand one hundred and ninety-one unrelated papers, 89 case reports, 48 reviews/meta-analyses and letters, 20 abstracts, and 2 research protocols were removed. Thirty-five full-text studies were assessed for their eligibility. Nineteen studies [6, 9–26] were excluded for reasons stated in supplementary 2 (characteristics of studies excluded and reasons for exclusion), leaving 16 studies for quantitative analysis.

Studies characteristics and participants characteristics

Sixteen retrospective studies were retained for the quantitative analysis [5, 27–29, 29–40]. These articles were published between 2016 and 2023 and compared 715 patients in the resection group with 1211 patients in the non-resected group. (Table 1: Main characteristics of the articles retained for meta-analysis + Risk of Bias + Table 2: Risk of Bias assessement) and (Table 3: characteristics of resected patients).

Table 1.

Characteristics of included studies + Risk of bias

Study Country of conduct	Study type	Inclusion period	Population	Key detail of intervention group/n of patients	Key detail of control group/n of patients	Outcomes measure	Risk of Bias (ROBINS-I)
Liver only metastases
CRIPPA 2016 Italy	retro	Jan 2003- Dec 2013	Synchronous liver only metastasis : 127	surgery of the metastasis and the primary : 11	Chemotherapy : 116	overall survival (OS) 1 year survival	critical
TACHEZY 2016 European multicentric	case control	Jun 1994- Jul 2014	Synchronous liver only metastasis : 138	surgery of the metastasis and the primary : 69	no resection : 69	overall survival (OS)	serious
KIM 2019 South Korea	retro	Jan 2000- Dec 2014	Metachronous liver only metastasis : 132	Resection : 13	no resection : 119	overall survival (OS) (from pancreatic resection to death)	serious
YANG 2019 China	case Control	Jan 2012- Oct 2017	synchronous pancreatic liver only metastasis : 89	surgery of the metastasis (resection or RFA) and the primary :48	No resection : 41	1 year survival 3 year survival	moderate
SHAO 2020 China	case control	Jun 2009- Nov 2018	Synchronous liver only metastasis : 100	surgery of the metastasis and the primary : 50	Chemotherapy : 50	overall survival (OS) 1 year survival 3 year survival	moderate

Table 2.

Characteristics of resected patients

Authors	Neoadjuvant treatment/(number of patients treated (%)	Type of pancreatectomy/number of patients (%)	Number and size of Liver metastasis	Type of liver resection/number of patients	Other metastatic sites resection/number of patients	Conditions of resection
Liver only metastasis
CRIPPA 2016 11 patients	Treated :100% - FOLFIRINOX : 3 (8%) - PEXG : 4 - GEMOX : 2 - PEFG : 1 - PDXG : 1 No treatment : 0	-PD : 6 (55%) - PPPD :0 -LP :5 (44%) - TP :0	Number: - No limit of number at diagnosis. - Single at restaging Size : - NI	- Atypical : 2 - Segmentectomy : 2 - No resection : 7 (complete response)	-NO	After neo-adjuvant treatment : - Primary resectable or borderline - Complete or major liver metastases response (RECIST) - CA 19/9 decrease > = 90%
TACHEZY 2016 69patients	Treated : 9 (%) - FOLFIRINOX : 4 (6%) - gemcitabine : 3 (4%) - unknown CTX : 2 (3%) No treatment : 60 (%)	-PD : 30 - PPPD : 12 -LP : 25 - TP : 2	Number: - 2–11 Size: - NI	-Atypical : 69	-NO	- Patient decision++ - Surgeon decision if discovered peroperatively
SHAO 2020 50 patients	Treated : 41 (%) - FOLFIRINOX - Gemcitabine based No treatment : 9 (%)	-PD : 50 (100%) - PPPD : 0 - LP : 0 - TP : 0	- NI	- Hepatic resection : 45 - RFA + hepatic resection : 5	-NO	- If preoperative diagnosis : when neaoadjuvant treatment is effective - if discovered peroperatively : oligo-metastase (< 4)
PAUSCH 2021 259	Treated : 174 (67.2%) -Not specified No treatment : 85 (32.8%)	-NI	Number: - < 5 :259 Size: - NI	- Minor resection or ablation	- NO	- < 5 metastases - Resectable lesions with minor resection or ablation
KIM 2019 13 patients	- NI	- NI	- NI	- NI	- NO	- Patient decision - surgeon decision
YANG 2019 48 patients	Treated : 13 (%) -Gem + nab- paclitaxel : 13 No treatment : 35	-PD : 20 - PPPD : 0 -LP : 28 - TP : 0	Number: - Oligometastatic (1–3): 23 - Non oligometastatic(> 3): 25 Size: - NI	- Atypical : 43 - Segmentectomy : 4 - Hemihepatectomy : 1	- NO	- No locally advanced primary - R0 possible for primary - Metastases can be completely extirpable
SCHWAR Z 2020 25 patients	Treated : 14 (%) -not specified No treatment : 11 (%)	-PD : 20 - PPPD : 0 - LP : 5 - TP : 0	Number: - 1–3 metastases: 25 Size: - NI	- Atypical : 8 - Anatomical minor : 15 - Anatomical major : 2	- NO	- Good response to chemotherapy+++
LEE 2020 60 patients	Treated : 4 (%) -Not specified No treatment : 56 (%)	-PD : 23 - PPPD : 15 -LP : 20 - TP : 2	Number: - Single:39 - Multiple: 22 Size: - Max < 1.5 cm: 25 - Max > 1.5 cm: 35	- RFA : 60	- NO	- < = 5 lesions - < = 5 cm in max diameter - No vascular invasion - Stable extra-hepatic disease by chemo
Lung only metastasis
ARNAOU TAKIS 2011 31 patients	- NI	- PD : 31	- NI	- NO	-Pulmonary resection through standard thoracotomy : 31	- Surgeon decision
KIM 2019 15 patients	- NI	- NI	- NI	- NI	- NO	- Patient decision - surgeon decision
YUN 2022 15 patients	Treated : not informed - FOLFIRINOX - Gemcitabine + Abraxa ne - Gemcitabine + other combination - Gemcitabine monotherapy No treatment : not informed	- NI	- NI	- NO	-Pulmonary resection :15	- NI
Multi-metastatic sites
KANDEL 2018 6 patients	Treated : 100% - FOLFIRINOX - Gemcitabine + Nab -Paclitaxel No treatment : 0	- PD : 3 - LP : 3	Number: - < 3 lesions:6 - < 4 cm:6 Size: - NI	- Hepatic resection only : 1 - Hepatic resection + RFA : 2 - Radio-embolization : 1	-RFA only : 2	- Matastases limited to the liver and lung - < 3 lesions - < 4 cm - After neo-adjuvant treatment
TANAKA 2019 43 patients	Treated : 100% -FOLFIRINOX No treatment : 0	-PD : 16 -LP : 19 - TP : 8	Number: - None: 5 - 1: 19 - 2: 11 - 3 : 4 - 4 : 0 - 5 : 2 - 6 : 2 Size NI	- Metastasectomy : 30	-Metastasectomy pertoneum : 7 - Metastasectomy other sites : 6	- No progressive tumor after chemotherapy - < 7 lesions - Technically resectable disease
BYUN 2019	Treated : 100% -FOLFIRINOX No treatment : 0	- NI	- Number and size NI	- NI	- NI	- Response or stable disease after chemotherapy (RECIST- CA19-9 OR metabolic uptake) - Patient willingness
GU 2020 36 patients	Treated : 0 No treatment : 100%	- NI	- Number and size NI	- NI	- NI	- Surgeon decision - Patient decision
HANK 2023 48 patients	Treated : 100% -Not informed No treatment : 0	-PD : 15 -LP : 24 - TP : 9	- Number and size NI	- NI	- NI	- For primary : stable, partial or complete response after chemotherapy - For metastases : partial or complete response with CA19- 9 and ACE decrease

Abbreviations : NI not informed, PD pancreaticoduodenectomy, PPPD Pylorus preserving pancreaticoduodenectomy, LP left pancreatectomy, TP Total pancreatectomy, RFA radiofrequency ablation

Table 3.

Risk of bias assessement

Risk of bias judgement
Study1	Counfounding	Selection of participants	Classification of interventions	Deviations from intended interventions	Missing data	Measurement of outcomes	Selection of the reported result	Risk of bias
ARNAOUTAKIS 2011	moderate	low	low	low	low	low	low	moderate
CRIPPA 2016	critical	low	Low	low	low	low	low	critical
TACHEZY 2016	moderate	low	serious	low	low	low	low	serious
KANDEL 2018	moderate	Low	Low	Low	low	Low	low	moderate
BYUN 2019	serious	Low	Low	Low	Low	Low	Low	Serious
KIM 2019	serious	Low	Low	Low	Low	Low	Low	Serious
TANAKA 2019	serious	Low	Low	Low	Low	Low	Low	Serious
YANG 2019	moderate	Low	Low	Low	Low	Low	Low	moderate
GU 2020	serious	Low	Low	Low	Low	low	low	Serious
SCHWARTZ 2020	serious	low	Low	low	low	low	Low	Serious
SHAO 2020	moderate	low	Low	low	low	low	low	moderate
LEE 2020	low	low	Low	low	low	low	low	moderate
PAUSCH 2021	low	low	serious	low	low	low	low	Serious
YUN 2022	serious	Low	Low	Low	low	Low	low	Serious
HANK 2023	moderate	low	moderate	low	Low	low	Low	moderate

Main outcome: overall survival

For Overall Survival, seven studies [27, 28, 31, 33, 35, 36, 40] involving 1106 participants were included in the analysis. A total of 458 M-PDAC patients receiving Surgical treatment were compared to 648 patients receiving palliation. The meta-analysis indicated a statistically significant benefit of surgical treatment (HR:0.42, 95% CI [0.33–0.53]). Between-study heterogeneity was low (I² = 31%) (Fig. 2).

Fig. 2.

Forest plot overall survival

Subgroup analysis was performed based on metastatic sites (Fig. 3). Four studies [27, 28, 31, 33] examined the role of surgical resection in liver only metastases. The analysis showed that surgical resection provided a survival benefit compared to palliative care (HR : 0.40, CI 95% [0.29, 0.53]).

Fig. 3.

Overall survival: subgroup analysis based on metastatic site

Between-study heterogeneity was moderate (I2 = 44%). For lung-only metastasis, only one study [35] has compared the two treatment modalities. Therefore, it was not possible to perform a meta- analysis.

One-year survival

Five hundred ninety-eight resected patients were compared with 1056 palliative patients in terms of 1-year survival [5, 27, 29–39]. The analysis revealed a statistically significant advantage of surgical treatment (OR: 0.31, CI 95% [0.21, 0.46]). The heterogeneity of the included studies was intermediate (I2 = 39%) (Fig. 4). A sensitivity analysis was conducted by excluding papers with a serious or critical risk of bias. This did not alter the results for 1-year survival (OR: 0.37, CI 95% [0.26, 0.52]) and heterogeneity was reduced (I2 = 11%). Subgroup analysis was performed based on metastatic sites (Fig. 5). Seven studies [5, 27, 29, 31–34] examined the role of Surgical resection in liver only metastases. Four hundred and sixty-six patients were compared with 653 unresected patients. The 1 year survival was better in the resected group (OR : 0.28, CI 95% [0.17, 0.45]). The heterogeneity between studies was moderate (I2 = 40%). For lung-only metastasis [29, 32, 34], no difference was found between the two groups (OR : 0.55, CI 95% [0.10, 3.02]).

Fig. 4.

Forest Plot of 1-year survival

Fig. 5.

1-year survival: subgroup analysis based on metastatic site

Three-year survival

Six hundred and twenty resected M-PDAC patients were compared with 958 patients receiving palliative treatment in terms of 3-year survival (5,29–33,35,37–40). The analysis showed a statistically significant superiority of surgical treatment (OR: 0.16, CI 95% [0.07, 0.383]). the heterogeneity among the included studies was low (I2 = 28%). (Fig. 6)

Fig. 6.

Forest plot of 3-year survival

In the Subgroup analysis according to the metastatic site, 6 studies [5, 29–33] investigated 992 patients and compared resection to no resection for liver-only metastases. The analysis found a statistically significant benefit of surgical treatment (OR : 0.22, CI 95% [0.06, 0.83]). Heterogeneity was intermediate between the studies. Regarding lung-only metastasis [29, 35], no significant difference was found between the two groups (OR : 0.25, CI 95% [0.03, 2.08]). (Fig. 7)

Fig. 7.

3-year survival: subgroup analysis based on metastatic site

Discussion

PDAC is one of the deadiest cancers, with a 5-year relative Survival of 11.5% [1, 2]. No more than 20% of patients are eligible for surgery at presentation, mainly due to vascular involvement or metastatic spread [41]. Currently, guidelines recommend FOLFIRINOX or Gemcitabine + Nab-paclitaxel as first-.

line treatments for metastatic disease; however surgical resection remains nonconsensual due to insufficient evidence.

To our knowledge, there are only two ongoing phase III trials investigating the role of surgical resection in this group of patients, and the results have not yet been published. The CSPAC-1 trial [42] evaluates Surgical resection after chemotherapy in patients with synchronous oligo-metastases limited to the liver. The authors plan to enroll 300 participants, and the study completion is estimated to be June 2025. The second trial is the METAPANC-trial [43], which shares the same protocol as the CSPAC-1 trial, including both synchronous and metachronous metastases.

Our systematic review and meta-analysis showed a statistically significant survival benefit of surgical treatment in patients with metastatic PDAC, mainly for lesions limited to the liver, whether synchronous or metachronous. Our findings suggest that certain subgroups of patients with stage IV PDAC may have a chance of prolonged survival. Identifying metastatic PDAC such patients with favorable biological profiles could potentially grant them consideration for curative treatment.

The ABCD metastasectomy criteria, introduced by Omiya et al. [44], are designed to identify patient populations suitable for surgery in the context of metastatic disease. A (Anatomy) refers to imaging evidence of tumor shrinkage; B (Biology) refers to CA19-9 levels being normal or normalizing after chemotherapy; C (Condition) refers to modified Glasgow prognostic score; and D (Duration) refers to a sufficient treatment period ≥ 6–8 months to assess response and stabilize disease before considering surgery. According to Omiya et al., applying these criteria resulted in promising outcomes for patients undergoing metastasectomy, with a median overall survival (OS) of 52.6 months and Survival rates at 1 year (98.2%), 3 years (78.4%), and 5 years (40.0%). These criteria align with the approach of prioritizing the response to chemotherapy as a key factor in selecting patients for surgery, aiming to maximize the likelihood of long-term survival and potential curative treatment.

Advancements in chemotherapy regimens have led to significant improvements in the Survival of patients with metastatic PDAC. FOLFIRINOX and Gemcitabine plus Nab-paclitaxel are now considered the standard of care, achieving response rates of 20 to 30%. Frigerio et al. [45] reported a 56 months overall survival after conversion surgery. Recently, the NAPOLI-3 phase III trial [46] yielded positive results for NALIRIFOX, leading to FDA approval of this regimen as a first-line treatment option for unselected stage IV PDAC. However, the response rate to chemotherapy can vary significantly among patients. Ten to twelve% of individuals exhibit germline alterations, that influence their response to treatment and overall disease behavior. Recognizing these mutations is crucial for determining prognosis and exploring targeted therapy options. In a phase II trial, O’Reilly et al. [47] observed a response rate twice as favorable in PDAC patients with BRCA1/BRCA2 and PALB2 mutations (present in 7% of PDAC cases) treated with platinum-based therapies. KRAS mutations, detected in 90 to 92% of PDAC cases, are predominantly found in G12, less frequently in G13, and rarely in Q61. Abdelrahmen et al. [48] highlighted that specific alleles of KRAS may have prognostic significance: KRAS wild-type (wt) alleles typically correlate with the most favorable outcomes, whereas KRAS G12R mutations may confer a better prognosis compared to KRAS G12D, possibly indicating a Subgroup with heightened immunogenicity. However, clinical detection of these abnormalities is not yet routine, and their application in patient selection remains confined to research settings. Response to neoadjuvant chemotherapy or chemoradiation remains the primary means to distinguish patients with a favorable prognosis Suitable for curative treatment from those with a poorer prognosis for whom surgery is unlikely to be successful. The radiological response of the primary pancreatic lesion is often underestimated on CT-scan images due to dense stroma and treatment-related fibrosis, which can lead to underestimation of tumor shrinkage. Retrospective studies of Bordeline PDAC and Locally advanced PDAC patients confirmed that radiological restaging, after neoadjuvant therapy, according to the RECIST or NCCN criteria cannot accurately predict the resection rate. In this context, evaluating the radiological response of metastases is more reliable than assessing the primary tumor. In terms of biological response, several studies have identified a decrease in CA 19 − 9 levels after neoadjuvant treatment as a strong predictor of Survival. However, the threshold values for the reduction in CA 19 − 9 levels or the post-chemotherapy CA 19 − 9 level that would motivate metastasectomy remain controversial. Hank et al. [40], reported no survival difference between resected patients with pre-operative CA19-9 levels < 37 U/ml or 37–400 U/ml.

However, patients with CA19-9 levels > 400 U/ml had worse Survival. Additionally, in multivariate analysis, a pre-operative CA 19 − 9 level > 400 U/ml was found to be a predictive factor for mortality. Similar findings were reported by Tanaka et al. [37] with a cut-off value of 150 U/ml. A “major biochemical response,” defined as a decrease of at least 90%, was considered conclusive and decisive by Crippa et al. [27] for resecting patients with metastatic PDAC. Resected patients demonstrated significantly higher survival than non-resected patients (39 months vs. 11 months).

In addition to evaluating tumor response to chemotherapy, the authors have highlighted considerations such as metastatic site, number of metastases, and synchronous or metachronous nature as crucial factors in determining the suitability for surgery. The liver is the most common site of metastasis in PDAC (90%), followed by the lymph nodes (25%), lungs (25%), and peritoneum (20%) [15]. Consequently, more studies have focused on surgery for hepatic metastases than for other locations. Our study showed a statistically significant survival benefit of surgery for liver-only metastases. However, outcomes vary widely between studies, depending on whether the metastases are metachronous or synchronous.

Metachronous liver-only metastases have Survival rates ranging from 15 to 37 months(5– [7, 32, 49]). In contrast, synchronous lesions exhibit Survival rates ranging from 5.9 to 15.6 months with upfront surgery [16, 24, 28, 30, 49–52], and 19 to 56 months after conversion surgery(27,40,45,53). Regarding lung-only metastases, the data we found were insufficient to draw definitive conclusions. Our analysis included three comparative studies that focused solely on the outcomes of lung metastases. Arnaoutakis et al. [34] reported in 2011, in a case–control study, a cumulative Survival of 51 months in the resected group versus 23 in the non-resected group. Kim et al. [29] reported a better outcome of Surgical resection when metastases were limited to the Lung compared to other metastatic sites. Survival after recurrence was 36⋅5 months in the resected group versus 9⋅5months in the non-resected group. Finally, Yun et al. [35] drew similar conclusions, reporting better outcomes for the Subgroup of patients with single metachronous Lung metastases. The reported 5-year Survival rate after resection was 60.6% vs. 6.2% in the chemotherapy group.

Our study had several limitations. First, the retrospective nature of the included studies reduced the strength of our results. Non-random bias may influence the comparison between resected and non- resected patients. Currently, there are only two ongoing Phase III trials, and the preliminary results are not yet available. To mitigate this bias, we conducted a sensitivity analysis that included only case-control studies, which did not alter the results, thereby, confirming the robustness of our findings.

Additionally, the studies included patients with metastases at multiple sites, which potentially introduces additional complexities into the analysis. Subgroup analysis by metastatic site was conducted, but there are insufficient studies specifically addressing resection outcomes for lung or multi-site metastases. Moreover, the selection criteria for resection varied among the studies included in our analysis, contributing to clinical heterogeneity, even though it did not result in statistical heterogeneity. As a result, our analysis did not establish evidence-based selection criteria for identifying the most suitable candidates for resection surgery.

Conclusion

To conclude, this systematic review with meta-analysis represents the first comprehensive effort to evaluate the efficacy of ablative treatments versus chemotherapy and palliative care for metastatic PDAC stratified by the site of metastasis. Our analysis demonstrated a significant survival advantage with surgical interventions compared with non-surgical treatments, particularly for lesions confined to the liver, both synchronous and metachronous. These findings underscore the potential benefits of aggressive treatment approaches in selected patients with metastatic pancreatic cancer. However, owing to the limitations of retrospective studies and variability in treatment criteria across the included studies, further prospective trials are crucial to validate and generalize these results across broader patient populations.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

T.K and M.A.T did the conception and design of the work, the data collection, and the data analysis and interpretation. A.A prepared figures. R.R did the critical revision of the article. F.G and T.P: did the final approval of the version to be published. All authors reviewed and approved the final manuscript.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.