Technical options in surgery for artery-involving pancreatic cancer: Invasion depth matters

Abstract

Background

The artery involvement explains the majority of primary unresectability of non-metastatic pancreatic cancer patients and both arterial resection and artery-sparing dissection techniques are utilized in curative-intent pancreatectomies for artery-involving pancreatic cancer (ai-PC) patients.

Methods

This narrative review summarized the history of resectability evaluation for ai-PC and attempted to interpret its current pitfalls that led to the divergence of resectability prediction and surgical exploration, with a focus on the rationale and the surgical outcomes of the sub-adventitial divestment technique.

Results

The circumferential involvement of artery by tumor currently defined the resectability of ai-PC but insufficient to preclude laparotomy with curative intent. The reasons behind could be: 1. The radiographic involvement of tumor to arterial circumference was not necessarily resulted in histopathological artery wall invasion; 2. the developed surgical techniques facilitated radical resection, better perioperative safety as well as oncological benefit. The feasibility of periadventitial dissection, sub-adventitial divestment and other artery-sparing techniques for ai-PC depended on the tumor invasion depth to the artery, i.e., whether the external elastic lamina (EEL) was invaded demonstrating a hallmark plane for sub-adventitial dissections. These techniques were reported to be complicated with preferable surgical outcomes comparing to arterial resection combined pancreatectomies, while the arterial resection combined pancreatectomies were considered performed in patients with more advanced disease.

Conclusions

Adequate preoperative imaging modalities with which to evaluate the tumor invasion depth to the artery are to be developed. Survival benefits after these techniques remain to be proven, with more and higher-level clinical evidence needed.

Key message

The current resectability evaluation criteria, which were based on radiographic circumferential involvement of the artery by tumor, was insufficient to preclude curative-intent pancreatectomies for artery-involving pancreatic cancer patients. With oncological benefit to be further proven, periarterial dissection and arterial resection have different but overlapping indications, and predicting the tumor invasion depth in major arteries was critical for surgical planning.

Introduction

In recent years, pancreatic cancer has been the most lethal malignancy in human beings, with a 5-year overall survival of around 10 % in the US, making it the third-leading cause of cancer-related death in the US as well as the seventh-leading cause in China [1,2]. Radical surgery has long been the only chance of curing pancreatic cancer, and provides a 5-year survival rate of 30–50 % if combined with modern adjuvant chemotherapy [3,4]. However, almost half of pancreatic cancer patients are metastatic upon primary diagnosis, with an 11.1-month overall survival, even when receiving an aggressive regimen consisting of oxaliplatin, irinotecan, fluorouracil and leucovorin (FOLFIRINOX) [5]. Furthermore, the majority of non-metastatic patients are at a stage in which radical resection is impossible due to tumor involvement with the major visceral vessels. With surgical techniques and chemotherapy regimens continuously improving, however, such cases are now potential candidates for curative resection. In current neoadjuvant settings, surgical resection has improved overall survival, with the figure increasing from 12 months up to 27 months in comparison with patients whose lesions remain intact [6].

Arterial involvement with the tumor has been one of the most important factors hindering curative-intent resection in pancreatic cancer patients both strategically and technically [7,8]. Arterial infiltration alone and combined arterial–venous involvement have contributed to 49 % and 37 % of primary unresectability in cases with non-metastatic locally advanced pancreatic cancer, respectively [8]. The arteries most commonly involved in pancreatic cancer are the celiac artery (CA), common hepatic artery (CHA) and superior mesenteric artery (SMA). The involvement of these arteries has led to the divergence of resectability evaluation as well as surgical techniques with which to perform adequate resection [9,10]. The criteria for resectability evaluation for artery-involved pancreatic cancer (ai-PC) have long been radiography-based, and the range of perivascular circumference involvement of the lesion has defined the resectability [[10], [11], [12], [13]]. However, these radiographic findings have been proven to be insufficient when it comes to predicting histological tumor invasion of the artery wall or precluding radical pancreatectomies [14]. Thus, innovations in imaging techniques for better preoperative resectability prediction have been warranted [[15], [16], [17]]. On the other hand, the aforementioned state-of-the-art surgical techniques for pancreatic cancer have provided a greater chance of curative resection, with arterial resection or arterial sparing techniques being possible in both treatment-naive ai-PC cases and those who have undergone neoadjuvant therapies [[18], [19], [20], [21], [22]].

In this review, we summarize the history of resectability evaluation and the evolution of surgical options for ai-PC, with a focus on the rationale and the surgical outcomes of the sub-adventitial divestment technique.

History of resectability evaluation for ai-PC and its current pitfalls

The vascular involvement explained around 75 % of primary unresectability during surgical exploration of non-metastatic pancreatic patients [8]. To achieve residue-free resection, Fortner et al. performed multivascular resection in a 1970s preliminary pancreatectomy cohort [23]. However, the aggressive procedures were considered to be infeasible, given the 30-day mortality of 17 %, as well as the 1-year survival rate of only 40 % in this cohort, with resectability evaluation being rarely discussed at that time [23,24]. However, major progress in pancreatic surgery has been observed in recent decades, whereby improving perioperative and long-term prognoses for pancreatic cancer patients and leaving peripancreatic vascular involvement the major obstacle for radical pancreatectomies; thus, resectability evaluation regarding pancreatic cancer patients has become a much debated topic [8,25,26].

In 1981, Megibow et al. established computed tomography (CT)-based resectability criteria for ai-PC for the first time, as an increased diameter of the superior mesenteric artery (SMA) or the celiac axis (CA) was associated with unresectability of the tumor. It was proposed that the vessel dilatation secondary to the obscuring of the periarterial region was possibly due to the tumor spreading in the perivascular lymphatics [27]. Warshaw et al. proposed angiographic criteria in 1990 which stated that narrowing or occlusion of the SMA or CA upon assessment via preoperative visceral angiography indicated unresectable cancer, which has a positive predictive value of 95 % and a sensitivity of 66 % [28]. Moreover, Loyer et al. and Lu et al. reported the prototype of the present-day radiography-based ai-PC resectability criteria that define the borderline and unresectable stages by examining the circumferential contiguity of the tumor to the vessel [29,30]. Since then, most ai-PC resectability evaluation systems have been based on the circumference of arterial involvement, which is determined by imaging and utilizing 180° as the cut-off value; indeed, such systems have been referred to by various guidelines all over the world [[10], [11], [12], [13],31].

However, ever since these criteria were proposed, inconsistency has persisted between the outcome of laparotomies for pancreatic cancer patients and radiographic resectability evaluation. In the 1990s, clinical scenarios were reported in which lesions were radiographically resectable but failed to be resected upon surgical exploration. CT-based resectability prediction criteria then provided almost 100 % sensitivity, while positive predictive values ranged from 22 % to 56 %, and the overall accuracy from 70 % to 79 % [[32], [33], [34]]. It was supposed that because of the relatively small size of these arteries, subtle invasion of the peripancreatic major vessels was difficult to detect [30]. With the progress of surgical techniques, subsequent studies after 2000 reported that radiographic suggestion of vascular circumferential involvement did not necessarily preclude curative intent after a laparotomy [14], which could probably be interpreted via two rationales.

First, the circumferential abutment or encasement of major peripancreatic vessels was not equivalent to histological tumor invasion of the artery. In a cohort studied by Bachellier et al., only 15 % (4/26) of resected arterial segments were invaded by pancreatic cancer upon histological examination, compared to 71 % (29/41) of resected portal-superior mesenteric veins (P-SMV) [35]. Furthermore, Watanabe et al. identified arterial involvement in 55 % (11/20) of left pancreatectomy specimens for pancreatic cancer, of which 91 % (10/11) had only invaded the tunica adventitia, with only 9 % (1/11) being found to have achieved deeper invasion [36]. A recent study published by Boggi et al. revealed that around 80 % of resected SMAs or CHAs were not histologically invaded [37]. Secondly, the developed surgical techniques, i.e., both arterial resection and arterial sparing, facilitated pancreatectomies for cases with major vascular involvement and provided preferable survival benefits, combined with modern adjuvant or neoadjuvant treatment [[20], [21], [22],[38], [39], [40]]. Fong et al. reported in 2013 that 24/25 (96 %) cases with major vascular involvement, as detected via CT, curative-intent resection could be achieved without vascular resection, and 17 (68 %) were detected residual free on the vascular margin [14]. In a more selective cohort consisting of 73 borderline and locally advanced ai-PC cases, Del Chiaro et al. reported 34 patients received arterial resection combined pancreatectomy, and the 1-, 3- and 5-year survival rates were 64 %, 23 % and 23 %, respectively [41].

Surgical options in pancreatectomies for ai-PC patients

It is of great importance to apply safe and effective surgical techniques to achieve curative pancreatectomies, especially in the current era, wherein systemic chemotherapy with modern regimens is providing promising survival benefits for patients with pancreatic cancer, and only patients receiving R0 resection have the best prognosis [42,43]. Versteijne et al. reported the results of their PREOPANC trial in 2020; they enrolled 54 (45 %) borderline and 65 (55 %) resectable pancreatic cancer patients in the neoadjuvant arm. The resection rate was 61 % and the median overall survival (MOS) was 16 months in this neoadjuvant arm, while the MOS of the resected subgroup in the same arm was 35 months [42]. The interim analysis of the CONKO-007 trial reported by Fietkau et al. presented a 27-month MOS for patients receiving neoadjuvant chemoradiotherapy and R0 pancreatectomies, and that the MOS of patients who did not receive resection was 17 months [43]. However, the neoadjuvant therapy rarely guaranteed radiographic regression of circumferential contiguity or the improvement of surgical resectability [44,45]. Dudeja et al. presented a patient cohort of 16 cases of locally advanced pancreatic cancer (LAPC) receiving neoadjuvant chemoradiotherapy with cisplatin, alpha-interferon, and 5-FU, none of which were observed to have circumferential regression on the peripancreatic major vessels involved [44]. A more delineated study undertaken by Mayor et al. showed an 18 % (4/22) circumferential decrease of arterial involvement when pathological invasion was detected in ai-PC patients with >180° radiological arterial involvement after neoadjuvant treatment, in comparison to 48 % (23/48) radiographic regression when a milder invasion depth was detected; this suggests that the invasion depth possibly plays a more important role than the circumference when resectability is evaluated [45]. Furthermore, in our experience, tumor residue as well as the extended desmoplasia around the artery complicated surgical dissection after neoadjuvant therapy, especially with a course longer than 3 months [46]. Thus, even neoadjuvant treatment was proven to devitalize periarterial tumors, the resection rate after treatment remaining unsatisfactory (28 %) prior to an aggressive FOLFIRINOX regimen [47].

The most common and traditional technical option in radical resection for ai-PC has long been combined resection of the arteries involved. However, the safety of this procedure is still seen to be very concerning since being preliminarily performed by Fortner et al. [23] Even in high-volume institutions for pancreatic surgery, the perioperative major morbidity rate after this high-risk procedure was 25–59 %, and the postoperative mortality rate was 10–16 % [40,48,49]. Mihaljevic et al. recently reported the largest single institutional cohort of Whipple's procedure, which showed that combined arterial resection during a pancreaticoduodenectomy was complicated with significantly higher surgical morbidity (59 %, 23/39) and 90-day mortality (10 %, 4/39) in comparison to a standard pancreaticoduodenectomy (42 %, 1222/2931 and 2.9 %, 86/2031, respectively) [50]. Specifically, when combined with SMA resection and reconstruction, a pancreaticoduodenectomy (or total pancreatectomy) was complicated with 39–91 % morbidity, and the mortality rate was up to 20 % [51]. As for a distal pancreatectomy, a multi-institutional retrospective study reporting on left pancreatectomies with celiac axis resection (DP-CAR) was carried out by Klompmaker et al.; the study enrolled patients from 20 hospitals in 12 countries and demonstrated 9 % (18/191) 90-day mortality and a 27 % (51/191) Clavien–Dindo 3a to 4b complication rate [40].

It was considered rational to perform arterial resection combined pancreatectomy only in hands of surgeons with high caseload after the learning curve [18,40,52,53]. Finks et al. reported that higher hospital volume explained 67 % decline in post-operative 30-day mortality for oncological pancreatectomies based on an US national-wise data base, with upper quadrants of the pancreatectomy volume were ranged from 14 to 35 cases/year, from 1999 to 2008 [54]. While the widely accepted cut-off to define high-volume or high case load is 15–20 pancreatectomies per year, evidence was needed to clarify the pancreatomy caseload on the outcome of patients receiving arterial the resection combined pancreatectomy [10]. When performed in centers with a DP-CAR of more than one case per year, postoperative mortality after this procedure was 5.5 %, compared to 18 % in centers with lower caseloads [40]. Tee et al. reported a 90-day mortality decrease from 29 % of the first 24 cases to 9 % of the following 87 cases when arterial resection combined with a pancreatectomy was performed [48]. Loos demonstrated that 15 was the cut-off number of arterial resection procedures performed to make the mortality rate lower than 10 %, with an actual 13 % (25/195) mortality rate [53]. The major safety concern surrounding arterial resection was a clinically significant postoperative hemorrhage, which occurred in around 15 % of patients [48,53]. Ischemic complications constituted another immanent issue with respect to this technique; such complications developed from the ceasing of arterial continuity when resection was performed alone, or thrombosis when arterial reconstruction was carried out. It was reported that ischemic complications contributed to as much as 45 % (5/11) to 73 % (11/15) of postoperative deaths after arterial resection combined pancreatectomies [48,55].

More recently, arterial-sparing procedures involving specific techniques have been proposed to maintain arterial lumen integrity while fulfilling the curative intention of ai-PC patients, notably periarterial divestment, periadventitial dissection, sub-adventitial divestment, or level 3 SMA dissection [20,22,39,56]. Inoue et al. carried out level 3 and extended level 3 SMA dissection with a monopolar electric cautery or energy device to perform a lymphadenectomy or a combined neural clearance around SMA adventitia extended >180°, depending on whether the artery was involved for >180° [22,56]. Taking the tumor invasion depth into account, Habib et al. reported an SMA periadventitial dissection to strip neural and lymphatic tissue around the SMA adventitia; they proposed the “halo sign” and “string sign” to predict pathological tumor invasion of the arterial adventitia and the feasibility of periadventitial dissection [20]. Diener et al. described a periarterial divestment technique utilizing bipolar-fine forceps to vaporize and loosen the periarterial tissue, as well as scissors to remove the adventitia involved [39]. A more aggressive sub-adventitial divestment was reported to perform en bloc resection of a tumor-involved adventitia on the dissection plan of the arterial elastic external lamina (EEL), emphasizing blunt separation of the tumor-invaded adventitia on the EEL and the dissection of ultrasonic scalpels [21] (Fig. 1).

Fig. 1.

Pathological invasion depth indicated the peri-arterial dissections. Three representative cases were aligned in column to illustrate the tumor-artery interfaces as different tumor invasion depth to artery occurred, as well as their impact on the feasibility of peri-artery dissections. The ex-vivo sub-adventitial divestment (SDT) was performed on the three fresh surgical samples of splenic artery in radical left pancreatectomy for pancreatic cancer patients. When there was no pathological invasion to the artery, or the invasion was restrained in the arterial tunica adventitia, the SDT could be performed keeping the external elastic lamina (EEL) intact. Any peri-arterial dissection would be infeasible if the EEL was invaded.

All slides were stained with hematoxylin and eosin and scale bars were presented at the right-lower portion of each slide panel. Red triangle: splenic artery lumen; blue triangle: splenic vein lumen; black arrowed line: the EEL of splenic artery after SDT; red arrowed line: dissection margin of SDT on tumor's end.

These emerging artery-sparing procedures were histopathologically based not only on the fact that pathological invasion of the artery rarely occurred, but also on the strength of the holding structure of the arterial wall and whether it was left intact after the aggressive periarterial maneuver. The EEL, consisting of compact collagen and elastin, was a firm layer between the tunica adventitia and vascular smooth muscle, which was proven to remain intact after both blunt and sharp stripping [57]. Similar adventitial stripping and denervation of the internal carotid artery were performed to treat carotid sinus syndrome with proven long-term safety [58,59]. When this similar arterial-sparing dissection was adopted for pancreatic cancer surgery, the smooth and the tough EEL were found to be a “holy plane” which could separate the involved adventitia from the artery (Fig. 2), since the potential space between the firm desmoplastic tumor invading the adventitia and the resilient artery was technically feasible for blunt dissection [21,46,60]. A pathological grading system was also proposed to appraise the EEL as the reference for the tumor invasion depth and evaluate the feasibility of sub-adventitial divestment. When sub-adventitial divestment was feasible, 100 % (12/12) of tumor invasion of the artery was external to the EEL. In arteries which were failed to be divested, 46 % (6/13) were detected to have tumor invasion within a 1 mm range from the EEL, and 54 % (7/13) were demonstrated to have invasion of the EEL (Fig. 2) [61].

Fig. 2.

Diagram of the pathological grading system and sub-adventitial divestment feasibility. (A) Grade 0: no tumor invasion of the artery, and R0 resection was to expect; (B) Grade I: tumor invasion of the tunica adventitia, with a tumor free distance from the external elastic lamina ≥1 mm and R0 resection was to expect with sub-adventitial divestment; (C) Grade II: tumor invasion of the tunica adventitia <1 mm of the external elastic lamina and R1 resection was to expect with sub-adventitial divestment or arterial resection was needed; (D) Grade III: tumor invasion of the external elastic lamina, arterial resection was needed or case unresectable.

A consecutive retrospective ai-PC patient cohort receiving sub-adventitial divestment combined with a pancreatectomy was studied; 86 % (36/42) of the sample presented with tumor involvement with the SMA. Combination with this artery-sparing procedure took approximately 30 min longer in terms of surgical duration, with a comparable estimated blood loss, compared to standard resections. The overall postoperative morbidity rate (40 %, 29/72) was similar to that of the contemporary control cases receiving standard resection for the disease at an earlier stage (37 %, 88/235). The 90-day mortality rate applied to 8 % (5/60) of the cohort, optimal to the historical reports of SMA resection cohorts [21,47]. Loos et al. compared outcomes after arterial resection and those after periadventitial dissection among ai-PC patients, 190 of whom received periadventitial dissection, with 195 receiving arterial resection. Arterial resection combined with a pancreatectomy was performed for a more advanced disease, with significant increases in surgical blood loss and operation time (median, 1300 mL vs. 900 mL and 420 min vs. 366 min). The arterial resection arm demonstrated significantly higher in-hospital mortality (13 % (25/195)) in comparison to patients receiving periadventitial dissection (5 % (9/190)). The chance of a postoperative pancreatic fistula, hemorrhage or ischemia was also increased in the arterial resection arm [53]. A major medical concern after arterial divestment was refractory diarrhea following denervation of the GI tract, which would impede in-time adjuvant therapy and, thus, negatively impact on patients' survival. Inoue et al. reported that diarrhea requiring opium tincture admission occurred in 21 % (LV2, 16/75), 34 % (LV3, 21/117) and 76 % (E-LV3, 31/41) of patients who received no neural clearance, hemi-circumferential denervation, or lympho-neural resection >180° around the SMA, while the proportions for patients receiving adjuvant chemotherapy were similar (76 % (57/77), 81 % (95/115) and 88 % (36/41)) [22]. In terms of the aspect of survival benefit, the 5-year survival rate was 27 %, 22 % and 26 % for LV2, LV3 and E-LV3, respectively. It was also reported that even a more aggressive arterial dissection resulted in a more severe long-term diarrhea, and the local resection rate was significantly decreased in the diarrhea group [62]. Compared to those receiving arterial resection, the MOS of patients receiving periadventitial dissection seemed to be prolonged (21.5 vs. 17.7 months), although statistical significance was not reached. Given the fact that patients receiving arterial resection might be complicated with a more advanced disease, further evidence was needed in order to justify the indications and oncological efficacy of these two surgical techniques [53,60].

Future considerations regarding the sub-adventitial divestment technique and ai-PC treatment

Resectability of pancreatic cancer should be defined in a multifaceted way, both technically and biologically. Technical unresectability was due to surgical obstacles caused by a locally advanced growth pattern of the tumor, surgical facilities, and team support at reach, case volume of the surgeon in complex pancreatic surgery [[63], [64], [65]]. Meanwhile, biological unresectability was due to the poor biological behavior of the tumor, such as early tumor metastasis, which restrained the survival benefit after tumor resection, regardless of the aforementioned technical obstacles [66]. Precise and objective evaluation of resectability was of paramount importance to ai-PC; indeed, without such evaluation as the reference and guidance for treatment, the patients would either lose the chance to receive curative resection or suffer from unbeneficial or even harmful surgery.

For ai-PC patients with tumor invasion deeper than the EEL, it was not possible to perform surgery with peri- or sub-adventitial dissection in a safe and radical manner. It is still necessary to equip with adequate arterial resection and reconstruction techniques for pancreatic surgeons performing radical surgery for ai-PC patients. Although the artery-sparing technique was highlighted for its potential safety and the oncological benefit for the majority of ai-PC patients, whose tumor invasion was out of the arterial EEL, high-level evidence was needed in order to verify the surgical security and the long-term outcome of said technique. Periarterial dissection and arterial resection have different but overlapping indications and predicting the tumor invasion depth in major arteries was critical for surgical planning.

Habib et al. proposed radiographic characteristics in preoperative CT scans to predict adventitial invasion, although the effectiveness of this approach in large-volume cohorts warrants further validation [20]. Larena-Avellaneda et al. reported a preliminary study on evaluating arterial invasion depth via intraarterial ultrasound, which could delineate the tumor–artery interface in a higher-resolution fashion. However, additional complications are inevitable due to the interventional nature of the technique [15]. X-ray microtomography could depict the EEL directly, whereby holding promise in depicting the abnormal structure of the arterial wall in ai-PC [16]. Elastin-specific molecular magnetic resonance imaging (MRI) was considered to be effective in the assessment of aortic walls and, thus, was seen to be a potentially feasible option with which to evaluate peripancreatic arteries [17]. Currently, however, surgical exploration and periarterial dissection remain the only way in which to determine the resectability for ai-PC patients and the feasibility of arterial resection or sparing techniques [46,67,68]. Future radiographic innovations should delineate the tumor–artery interface and invasion depth, facilitating a more precise and more objective preoperative resectability evaluation as well as directed utilization of either the arterial dissection or resection technique in radical surgery for ai-PC patients.

Meanwhile, a thorough understanding of the underlying biological mechanism in arterial invasion by pancreatic cancer is also warranted. In addition to the depth, the arterial involvement length was also considered implying the resectability of ai-PC and resulted in a 64 % of diagnostic accuracy to predict pathological tumor invasion [45,69]. However, it was also calling for further evidence whether the length of involvement, as the circumference, was a phenomenon resulted from tumor pathological local progression that led to technical difficulty of radical resection. The involvement of major arteries could still be considered resulted from the locus of ai-PC orientation, which abutted closer to the CA and SMA et al. [68,70] Different from perineural invasion, aggressive lymph metastasis, poorer differentiation, and other biological risk factors [71,72], arterial involvement alone was probably a situation that increased the technical difficulty of radical resection [70]. Our ever-deepening understanding of tumor biology would be helpful in predicting cases with a high risk of early metastasis, even if the lesion were small or free of vascular involvement. Moreover, ai-PC patients with less metastatic potency, even with multiple or long-range vascular invasion, would benefit from aggressive surgery such as arterial resection or divestment [[73], [74], [75], [76], [77], [78]].

Conclusion

Both arterial divestment and resection are feasible in curative-intent surgery for ai-PC, and determining the depth of tumor invasion of the artery (beyond the EEL or not) provides the base of the rationale behind choosing between these two techniques. Adequate preoperative imaging modalities with which to delineate the tumor–artery interface are to be developed. Survival benefits after these techniques remain to be proven, with more and higher-level clinical evidence needed.

Ethical approval

The study was approved by the Ethics Committee of the First Affiliated Hospital of Nanjing Medical University (Nanjing, China, Reference Number 2022-SR-001).

Funding sources

This work was supported by the grant from the National Natural Science Foundation of China (81672449, 82173206), Construction Program of Jiangsu Provincial Clinical Research Center Support System (BL2014084), the Project of Invigorating Health Care through Science, Technology and Education, Jiangsu Provincial Medical Outstanding Talent (to YM, JCRCA2016009), and Innovation Capability Development Project of Jiangsu Province (No. BM2015004).

CRediT authorship contribution statement

Yi Miao made critical review of the manuscript and final approval of the submission; Baobao Cai screened the articles and wrote the manuscript; Zipeng Lu interpreted the data and review the manuscript.

Conflict of interest

The authors declare that no conflicts of interest exist.