Prospective minimally invasive pancreatic resections from the IGOMIPS registry: a snapshot of daily practice in Italy on 1191 between 2019 and 2022

This retrospective analysis of the prospective IGOMIPS registry reports on 1191 minimally invasive pancreatic resections (MIPR) performed in Italy between 2019 and 2022, including 668 distal pancreatectomies (DP) (55.7%), 435 pancreatoduodenectomies (PD) (36.3%), 44 total pancreatectomies (3.7%), 36 tumor enucleations (3.0%), and 8 central pancreatectomies (0.7%). Spleen-preserving DP was performed in 109 patients (16.3%). Overall incidence of severe complications (Clavien–Dindo ≥ 3) was 17.6% with a 90-day mortality of 1.9%. This registry analysis provided some important information. First, robotic assistance was preferred for all MIPR but DP with splenectomy. Second, robotic assistance reduced conversion to open surgery and blood loss in comparison to laparoscopy. Robotic PD was also associated with lower incidence of severe postoperative complications and a trend toward lower mortality. Fourth, the annual cut-off of ≥ 20 MIPR and ≥ 20 MIPD improved selected outcome measures. Fifth, most MIPR were performed by a single surgeon. Sixth, only two-thirds of the centers performed spleen-preserving DP. Seventh, DP with splenectomy was associated with higher conversion rate when compared to spleen-preserving DP. Eighth, the use of pancreatojejunostomy was the prevalent reconstruction in PD. Ninth, final histology was similar for MIPR performed at high- and low-volume centers, but neoadjuvant chemotherapy was used more frequently at high-volume centers. Finally, this registry analysis raises important concerns about the reliability of R1 assessment underscoring the importance of standardized pathology of pancreatic specimens. In conclusion, MIPR can be safely implemented on a national scale. Further analyses are required to understand nuances of implementation of MIPR in Italy. Supplementary Information The online version contains supplementary material available at 10.1007/s13304-023-01592-7.


Introduction
Minimally invasive (MI) surgery was probably the greatest innovation in general surgery in the last century, and was immediately rewarding in operations requiring large incisions to perform low complexity surgeries.
Feasibility of MI pancreatic resections (MIPR) was shown over 25 years ago [1,2].However, MIPR had a slow implementation due to the intrinsic complexity of pancreatic The following surgeons contributed patients to the IGOMIPS registry, but do not qualify for full authorships.They should be listed as Collaborators are listed in the Supplementary Appendix Section.
Extended author information available on the last page of the article surgery, often including multiple digestive reconstructions, and the lack of an obvious advantage in terms of improved outcomes [3].Oncologic adequacy was an additional concern, especially for pancreatic cancer [4].Thanks also to the advent of robotic technology, recent evidence shows that in selected patients, MIPR can offer clear advantages over the conventional open surgery [5][6][7].These newer data come from centers of excellence that have surpassed a quite steep learning curve.In MI pancreatoduodenectomy (PD; MIPD), true proficiency is acquired after 250 procedures [8], making generalizability of results achieved by champion surgeons questionable.
The most effective tool to depict daily practice of MIPR is probably a prospective registry.Participation in national and international registries was also recommended by the International Evidence-Based Guidelines on Minimally Invasive Pancreatic Resections with the purpose of ensuring safe and wide expansion of MIPR [9].At the moment, there are some generalist registries for pancreatic surgery [10], but only two registries for MI pancreatic surgery: the European Registry (http:// www.e-mips.com/ regis try) and the Italian Registry (IGOMPIS: https:// www.yours uite.it/ IGOMI PS/).
IGOMIPS is a prospective registry, established in 2019, that includes the majority of MIPR performed in Italy [11].We herein report the first comprehensive analysis of the IGOMIPS registry with the aim of providing a snapshot of daily practice of MIPR in Italy.Detailed data are provided for distal pancreatectomy (DP) and PD, as these procedures account for over 90% of all MIPR.

IGOMIPS
The Independent Ethics Committee of the Humanitas Institute (authorization number 2167) established the IGOMIPS registry in 2019 following approval.IGOMIPS was recorded in the Registry of Patient Registries (RoPR) of the Agency for Healthcare and Research and Quality, US Department of Health (Registry of Patient Registries.Content last reviewed April 2019 https:// www.ahrq.gov/ ropr/ index.html).
A detailed description of IGOMIPS was previously reported [11].Briefly, IGOMIPS is a prospective registry for MI pancreatic surgery capturing operative and outcome data up to 90-days after surgery.All Italian centers performing MI pancreatic surgery can apply to participate to IGOMIPS, following protocol approval by the local Ethical Committee.IGOMIPS has some unique features permitting several analyses that are not feasible in other registries.First, every procedure must be declared the day before surgery, so that concordance between planned and performed procedures can be defined.Second, IGOMIPS includes progressive case numbers for centers and individual surgeons, thus permitting to analyze performance based on experience.Third, participating centers are periodically audited to verify the quality of data and adherence to registry protocol.

Study design
This study provides a retrospective review of MIPR performed at 34 IGOMPIS centers between September 2019 and June 2022.Comparisons and statistical analysis aim to provide insights on current practice of MIPR on a national basis.
Statistical computations were performed using the software STATA 17.0 (StataCorp.2017.Stata Statistical Software: Release 15.College Station, TX: StataCorp LLC.).Descriptive and inferential statistics were carried out with the analytical models adequate for the type of variable studied (e.g., Mann-Whitney test, chi-square).Two-sided P values lower than 0.05 were considered statistically significant.All continuous variables were reported as the median and interquartile range (IQR).

Data analysis
A per protocol analysis of MIPR enrolled in IGOMIPS up to June 2022 was performed.Other MI pancreatic surgeries (e.g., diagnostic laparoscopy, and biliary or gastric by-pass) were not included in the analysis.
Operations converted to open surgery were still analyzed as MI procedures (intention-to-treat analysis).We also reported rates of conversion to open surgery and agreement between planned and performed procedures.

Center volume
The cut-off of ≥ 20 MIPR proposed by the International Evidence-Based Guidelines on Minimally Invasive Pancreatic Resections [9] was used to identify high-volume centers.Outcomes of centers above and below this cut-off were compared.

Postoperative complications
Pancreas-specific complications were defined and graded as proposed by the International Study Group of Pancreatic Surgery [12][13][14][15].Severity of all complications was assessed according to the Clavien and Dindo scale [16].Severe complications were those with a score ≥ 3.
Data on modality of pancreatic transection were available for 534 patients (80.0%).A stapler was used to divide and seal the pancreatic stump in most MIDP (n = 427, 80.0%).In the remaining patients, the pancreas was either divided by harmonic shears alone (n = 55; 10.3%) or by an energy device plus fish-mouth sutures (n = 52; 9.7%).
In patients with pancreatic cancer, despite similar T stage, nodal status, and number of examined lymph nodes, robotic DPWS was associated with lower rates of R1 resection (7.3% vs 28.0%; p = 0.002).

Pancreatoduodenectomy
Table 6 provides a summary of baseline characteristics, intraoperative data, and postoperative outcome for patients undergoing MIPD.
Most PD were performed at high-volume centers (n = 303; 69.3%) using robotic assistance (n = 242; 55.6%).Laparoscopy was used in 152 PD (34.9%).When the analysis was restricted to the 16 centers that have a robot, implementation of PD increased to 44.6% and the use of robotic assistance to 59.0%.Adoption of robotic assistance increased over time.It was 7.1% in 2019, 47.6% in 2020, 57.7% in 2021, and 71.1% in 2022 (p < 0.0001).A hybrid technique, mostly using a mini-incision for some open anastomosis, was used in 37 additional laparoscopic (8.5%) and 4 robotic PD (0.9%), respectively.

Laparoscopic versus robotic PD
As shown in Table 7, comparing laparoscopic PD and robotic PD showed that the latter was performed in younger patients (67 years versus 69 years; p = 0.017) and in persons with a In patients operated for a malignant tumor, with an identical prevalence of T3/4 tumors and a higher prevalence of node positive patients in robotic PD (68.7% versus 57.3%; p = 0.049), robotic PD showed a higher median number of examined lymph nodes (30 [20] versus 21.5 [14]; p = 0.0001) but a higher rate of R1 resection (23.7% versus 14.0%; p = 0.050).

Results of MIPR based on center volume
Results of MIPR based on center volume are presented in Table 8.Center volume had an effect on selected outcome measures.

Discussion
This report of 1,191 contemporary MIPR from the prospective IGOMIPS registry provides some important information.First, robotic assistance was preferred for all types of MIPR but DPWS.Second, robotic assistance was associated with reduced rates of conversion to open surgery and lower amount of estimated blood loss.Third, robotic PD had trend toward lower mortality when compared to laparoscopic PD.Fourth, despite the use of robotic assistance was  This registry analysis raises also important questions about the reliability of R1 assessment (i.e., the importance of standardized pathology of specimens) and the consequences of unplanned MIPR especially when this means an increase in technical complexity (e.g., from MIDP to MIPD).
Despite the lack of clear evidence of superiority of robotic assistance over the conventional laparoscopy in MIPR, the use of robotic assistance was prevalent for all types of MIPR but DPWS and increased over time.There has been a tenfold increase in robotic MIPR between 2019 and 2022, at hospitals where a robot is available.Recent evidence shows that robotic assistance outperforms laparoscopy in MIDP, for some outcome measures [17][18][19].Considering the high costs of robotic assistance and the need to select the procedures in which this new technology may be conveniently employed, it is not surprising that the use of conventional laparoscopy was prevalent for DPWS.For MIPD, advantages of robotic assistance are more evident [20][21][22] as shown also by the high implementation of robotic PD in the IGOMIPS registry.One of the most striking pieces of evidence favoring robotic PD is provided by the Dutch trial on laparoscopic versus open PD that was terminated due to excess mortality in the laparoscopic arm [23].Since then, Dutch surgeons have embraced robotic PD and have achieved excellent outcomes [24].
Little doubt exists that the use of robotic assistance reduces the rate of conversion to open surgery and the amount of blood loss [17][18][19][20][21][22].The IGOMIPS registry confirms these results in daily practice, showing that advantages of robotic assistance in MIPR are not reserved to the few centers that have pioneered robotic surgery.One striking result from this registry analysis is that robotic assistance, when compared to laparoscopy, reduced the incidence of severe postoperative complications, and could reduce postoperative mortality of MIPD.Despite higher prevalence of patients at increased operative risk (ASA score ≥ 3: 47.1% versus 36.8%;p = 0.045), incidence of severe postoperative complications was lower in robotic PD.Difference in mortality showed only a trend toward statistical significance.However, it may still be important to note that robotic PD was associated with a mortality rate of 3% at a national level.This mortality rate is equivalent to the value reported in the benchmark study for open PD when patients have an ASA score ≥ 3 [25].In robotic PD, approximately 50% of the patients had an ASA score ≥ 3.
Not surprisingly, MIPD performed at low-volume centers was associated with worse outcomes.Only 5 of 19 centers (26.3%) performing MIPD met the threshold of ≥ 20 procedures per year defined by the Miami guidelines [9].Considering that approximately 30-40% of all PD can be MIPD, meeting this cut-off means that at least 50 PD are performed annually.Although just few centers met this annual volume at a national level, the importance of annual volume for the outcome of PD is well established [26].A Dutch study showed that at least 40 PD per year are required to improve postoperative mortality [27].A more recent study from Norway showed that ≥ 40 PD may not be enough to reduce postoperative mortality [28], and a study Korea demonstrated that mortality improves if the annual volume of PD is ≥ 54 [29].Therefore, the annual volume ≥ 50 PD, permitting ≥ 20 MIPD, seems appropriate to offer good clinical outcomes.
This analysis also shows that most MIPR are performed by a single surgeon at most centers.This was especially true for MIPD.If, on the one hand, convincing evidence demonstrates that 250 robotic PD are required to achieve truly optimal outcomes [8], thus reinforcing the need for centralization, on the other hand, this high number of procedures raises the difficult question about how MIPD can safely diffuse on a large scale [30].This study raises also important questions on how to train the next generation of pancreatic surgeons, and how to retrain the current generation of pancreatic surgeons that is mostly composed by open surgeons.
SPDP was mostly performed at high-volume centers, further underscoring the importance of volume in MIPR.Preserving the spleen is believed to be important in the rare patients with a benign but symptomatic tumor or a premalignant pancreatic tumor that require an MIDP.Spleen preservation prevents overwhelming sepsis and thrombocytosis and preserves overall immune function [31][32][33].In addition, it could reduce blood loss and operative time, while limiting the rate of postoperative pancreatic fistula and delayed gastric emptying [31,[34][35][36].However, SPDP, especially when the splenic vessels are also spared (Kimura procedure), is technically demanding and requires greater technical skills when compared to DPWS.This is mostly why robotic assistance is believed to improve the ability to preserve the spleen during MIDP [17,18].The fact that SPDP was mostly performed at high-volume centers is a possible explanation for the lack of an increased spleen preservation rate in the robotic group in this registry analysis.
In the IGOMIPS registry, DPWS was associated with higher rates of conversion to open surgery when compared to SPDP.In general, SPDP is technically more complex than DPWS.However, SPDP can be converted to DPWS when spleen preservation is not feasible, while primary DPWS is more frequently associated with difficulty factors, such as malignant histology, tumor proximity to the superior mesenteric-portal vein, sinistral portal hypertension, and splenomegaly making conversion to open surgery more likely to occur in these patients [37].
This registry analysis also showed that in MIPD, the pancreatic anastomosis is nearly always an end-to-side pancreatojejunostomy, but the surgical technique varied considerably among centers.A Blumgart or a modified Blumgart pancreatico-jejunostomy was used in 168 patients (38.6%), being the technique used more frequently.Practice in MIPD is probably influenced by experience in open PD.However, the minimally invasive approach may put additional difficulties, sometimes forcing surgeons to oversimplify the technique.This is probably why some surgeons prefer single layer running pancreatojejunostomy [38], despite this technique was associated with increased rates of postoperative pancreatic fistula in a large multicenter study [39].The Blumgart technique is quite easy to perform during MIPD and combines the principle of duct-to-mucosa anastomosis to jejunal wrapping over the pancreatic stump.A recent study showed that modified Blumgart pancreatojejunostomy is associated with low rates of grade C postoperative pancreatic fistula in either open or robotic PD [40].A modified Blumgart anastomosis is included in the standardized training pathway developed by the Dutch Pancreatic Cancer Group [24].
One key information from this study is that implementation of MIPR was not associated with high rates of resection for benign tumors, such as serous cystadenoma.A recent study showed that approximately 2% of the patients undergoing surgery for an incidentally discovered pancreatic cystic lesion have a final histology of serous cystadenoma [41].Despite the different denominator in this study and in the IGOMIPS registry, the 3% rate of resection for serous cystadenoma reported herein is quite reassuring that availability of MI techniques does not result in unnecessary surgery [42].
This registry analysis showed conflicting data about R1 rates.DP at high-volume centers was associated with higher rates of R1 resection, which appears counterintuitive.Furthermore, robotic PD was associated with higher rates of R1 resection, while robotic DP showed lower R1 resection, versus laparoscopy.Margins status is an important quality metric in pancreatic surgery [43], but objective assessment relies on standardized and high quality of pathology.The higher number of examined lymph nodes suggests more accurate histology at high-volume centers, supporting the hypothesis of underestimation of R1 at low-volume centers.In addition, administration on neoadjuvant treatments decreases R1 rates [44], and a study from Esposito et al. showed that most resections for pancreatic cancer is R1 [45].It is therefore difficult to believe that at high-volume centers, where patients receive neoadjuvant chemotherapy more frequently, R1 rates are truly higher when compared to lowvolume centers.Clearly, quality of pathology makes most of the difference.Pathology of pancreatic specimens should become truly standardized to permit meaningful comparison on margin status.
This study has several limitations.First, despite prospective enrollment in the IGOMIPS registry, accuracy of information depends on individual centers.However, prospective data acquisition is the best possible method to ensure high quality of information.Second, some results may be influenced by local practice and/or quality of some hospital services (e.g., pathology).The large number of cases reported to the registry is expected to dilute the effect of these confounders.Third, relatively few centers provided most cases.Therefore, even in a national registry, quality of care mostly refers to specialized centers.
In conclusion, this registry analysis shows that MIPR can be safely implemented on a national scale.A few highvolume pancreatic centers perform most procedures, but results achieved at low-volume centers appears acceptable.Further analysis, on a larger sample, is required to understand nuances of implementation of MIPR in Italy.
reports the study flowchart.Overall 1191 MIPR were analyzed following exclusion of 102 procedures, because an MIPR was not performed (n = 60; 4.0%) or due to missing data (n = 42; 4.2%).Number of procedures per center ranged from 1 to 161 with a median of 12 [49].Most procedures (854/1191; 71.7%) were performed at nine centers performing ≥ 20 MIPR per year.Overall, the top five most active centers enrolled 618 MIPR (51.9%).DPWS was performed at all centers.PD was performed at 19 centers (55.9%) and SPDP at 21 centers (61.7%

Fig. 1 Fig. 2
Fig. 1 Number of cases reported (within columns) by trimester and number of active centers (above columns)

Fig. 3
Fig. 3 Proportion of DPWS and SPDP at high-and low-volume centers indicate statistical significance of p values DP distal pancreatectomy, DPWS distal pancreatectomy with splenectomy, SPDP spleen-preserving distal pancreatectomy, PD Pancreaticoduodenectomy, MIPD minimally invasive distal pancreatectomy, BMI body mass index, ASA American Society of Anesthesiologists Table 8 centers (23.5%) located in Central Italy reported 342 MIPR (28.7%), and 7 centers (20.6%) located in Southern Italy reported 106 MIPR (8.9%).Four of the five top recruiting centers were located in Northern Italy and reported a total of 461 MIPR.A single center, located in Central Italy, was the top recruiter with 155 MIPR (13.0%).

Table 3
Data available for 95 SPDP and 439 DPWS; °data available for 92 patients.§ Data available for 108 SPDP, and 556 DPWS.*Data calculated only for malignant pancreatic lesions BMI body mass index, ASA American Society of Anesthesiologists, RAMPS radical anterograde modular pancreatosplenectomy ç

Table 5
for DPWS and < 5% for SPDP at low-volume centers.The importance of these findings is emphasized by the fact that figures refer to contemporary daily practice of MIPR on a national basis.

Table 8
Baseline patient characteristics and outcomes of Minimally Invasive Pancreatic Resections (MIPR) based on the cut-off of ≥ 20 procedures per year