Pancreaticoduodenectomy with right hemicolectomy for advanced malignancy: a single UK hepatopancreaticobiliary centre experience

Abstract Aim Locally advanced intestinal neoplasms including colon cancer may require radical en bloc pancreaticoduodenectomy and right hemicolectomy (PD‐RC) to achieve curative, margin‐negative resection, but the safety and benefit of this uncommon procedure has not been established. The Association of Coloproctology of Great Britain and Ireland IMPACT initiative has also highlighted a lack of awareness about current services available within the UK for patients with advanced colorectal cancer and concerns about low‐volume centres managing complex cases. Thus, we aimed to review the feasibility, safety and long‐term outcomes of this procedure at a single high‐volume hepatopancreaticobiliary surgery unit in the UK. Method A retrospective cohort study was performed using a database of all consecutive patients with intestinal cancer who had been referred to our regional advanced multidisciplinary team and undergone PD‐RC in a 7‐year period (2013–2020). Clinico‐pathological and outcome data were reviewed. Results Ten patients (mean age 54 ± 13, 8/10 men) were identified. Final histology revealed the primary tumour sites were colon (n = 7) and duodenum (n = 3). R0 resection was achieved in all cases. The major complication rate (Clavien–Dindo ≥ 3) was 10% (1/10) with no deaths within 90 days of surgery. The Kaplan–Meier estimated 5‐year overall survival was 83.3% (95% CI 58.3%–100%). Univariate survival analysis identified perineural invasion and extra‐colonic origin as predictors of poor survival (log‐rank P < 0.05). Conclusion En bloc PD‐RC for locally advanced intestinal cancer can be performed safely with a high proportion of margin‐negative resections and resultant long‐term survival in carefully selected patients.


BACKG ROU N D
Locally advanced intestinal neoplasms presenting with coloduodenal fistulas may require radical en bloc pancreaticoduodenectomy (PD) and right hemicolectomy (RC) to achieve complete (R0) resection. Malignant colo-duodenal fistulas may be secondary to either advanced colon cancer or, more rarely, primary duodenal cancer [1,2]. Both PD and RC in isolation, however, remain major surgical procedures with considerable morbidity and mortality. The complication rate of PD can exceed 40% even in experienced hands, with postoperative pancreatic fistula (POPF) being a major determinant of short-term outcomes [3]. Although RC may have lower rates of complications, the overall mortality is still estimated at 2%-3%, with colonic anastomotic leak as the key determinant of short-term outcomes [4,5].
Centralization of high-risk operations to high-volume centres has been associated with improved clinical outcomes [6]. For PD for cancer, centralization has been associated with both shortand long-term improvements in the quality of cancer surgery and perioperative care. This includes an increase in margin-negative (R0) resections, higher nodal clearance, fewer overall complications and improved management of post-pancreatectomy complications [3,7,8]. These improvements, in turn, have been associated with a reduction in postoperative mortality to <5% and improved overall survival [7,9]. The addition of RC to PD increases the complexity of the procedure and adds another anastomosis which, in theory, may increase the morbidity and mortality. This was suggested by a previous review of multivisceral resections with PD which showed a substantial increase in mortality (3-fold) and morbidity compared to a standard PD [10]. With increasing surgical and institutional expertise in complex pancreatic resections at high-volume units, however, the outcomes for multivisceral cancer resections in carefully selected patients may be equivalent to single-organ resections.
Combined PD-RC is an uncommon operation and a recent systematic review of the literature has reported that overall outcomes following PD with colonic resections are poor, with a morbidity of 12%-65% and surgery-related mortality of 10% [11]. Primary tumour histology was a critical determinant of oncological outcomes in this review, as patients with well-differentiated colonic adenocarcinomas without regional lymph node metastases had the best overall survival. In addition, previous reports have shown that curative multivisceral resection for colorectal cancer (CRC) is similar to standard resection [12]. The majority of cohorts included in this review, however, were from low/medium-volume centres and no studies have reported on the feasibility, safety and outcomes of elective and emergency PD-RC in a UK setting.
The Association of Coloproctology of Great Britain and Ireland has recently highlighted the priorities for advanced CRC patients after engaging with patient groups and clinical stakeholders through the Improving the Management of Patients with Advanced Colorectal Tumours (IMPACT) initiative [13]. This was in response to the considerable variation in the surgical management of these patients and lack of evidence from high-quality clinical trials. A major concern was the lack of clarity and awareness about current services available within the UK for patients with advanced CRC and concerns about low-volume centres managing complex cases. Our unit is a tertiary referral centre for hepatopancreaticobiliary (HPB) cancer surgery in England with an annual caseload of >50 PD, which can therefore be defined as a 'high-volume' institution [4,12]. Our primary objective was to review the management and clinical outcomes of all patients undergoing PD-RC in our region in the UK over a 7-year period in line with the themes of the IMPACT initiative.

ME THODS
A retrospective cohort study was performed using a prospectively maintained database of all consecutive adult patients (age >18 years) who had undergone PD-RC in a 7-year period (January 2013 to April 2020) at Hammersmith Hospital HPB Surgery Unit (London, UK) for intestinal cancer. The study has been reported in accordance with the STROBE guidelines (Table S1) [14]. All cases had been referred Endoscopic ultrasound was used selectively to assess the degree of organ invasion to aid operative planning when imaging was not conclusive.
Patients were selected for surgery using the general criteria • Good oncological outcomes can be achieved after PD-RC, particularly for primary colon cancers.
was performed for all patients by a consultant anaesthetist to confirm fitness for major surgery.
Neoadjuvant chemotherapy (NAC) was offered to patients on a selective basis when the resection margin was deemed to be at risk, based on an MDT assessment of preoperative CT imaging.
This was based on data from the FOxTROT trial which showed significantly lower margin involvement after NAC for locally advanced CRC [15].
The preliminary decision to perform a PD-RC was made after MDT review of the histology, CT imaging and upper GI endoscopy ± endoscopic ultrasound findings. Radical en bloc resection (PD-RC) was recommended for locally advanced tumours which were invading the second part of the duodenum (and/or the pancreatic head directly) adjacent to the ampulla of Vater and distal common bile duct. This would ensure wide margins and the greatest chance of a curative R0 resection. Pancreas-preserving duodenal resections would only be recommended in cases where the duodenal involvement was limited and distant from the ampulla, and there was no direct involvement of the pancreas. The decision to perform a PD-RC was then finalized after intra-operative assessment of the degree of organ invasion by the operating surgeons.
All operations were performed using an open approach by experienced consultant HPB and colorectal surgeons. The procedure of choice was en bloc pylorus-preserving PD with standard lymphadenectomy [16] and conventional lateral-to-medial RC. Reconstruction consisted of an end-to-side pancreatico-jejunostomy, end-to-side hepaticojejunostomy, end-to-side duodenojejunostomy and a sideto-side ileo-colic anastomosis. Postoperative complication severity was graded using the Clavien-Dindo classification [17] and pancreatectomy complications were recorded using the International Study Group of Pancreatic Surgery classifications [18][19][20]. Histology was reviewed by a consultant histopathologist with specialist experience in GI malignancy. In the case of any disagreement, the histology was reviewed by a second consultant histopathologist. After discharge, patients were followed up at 6-8 weeks after surgery and then at 3-6-month intervals. Adjuvant chemotherapy or targeted therapy was offered to patients on a selective basis after histopathology review and MDT discussion.
Data on patient demographics, staging, treatment, clinical outcomes and overall survival were retrieved from the database. Categorical clinico-pathological data were compared using a two-tailed Fisher's exact test. Overall survival was calculated using the Kaplan-Meier method and survival outcomes were compared using the log-rank test.
All statistical analyses were conducted using SPSS® version 27 (IBM).

RE SULTS
Over a 7-year period, 10 patients (mean age 54 ± 13, 8/10 men) underwent en bloc PD-RC for cancer at our unit ( Table 1) Figure S1). Perineural invasion was also associated with significantly worse overall survival. There were no significant associations  Table 3 shows associations between tumour site and histological markers of lymphatic, vascular and perineural invasion. Duodenal tumours tended to have a higher rate of perineural invasion (P = 0.03).
Pancreaticoduodenectomy with right hemicolectomy was per- Resection margin quality is a major indicator of surgical quality for pancreatic surgery units [21] and it is significant that all cases had a negative margin despite requiring en bloc multivisceral resection of large, locally advanced tumours. This is also clinically significant, since achieving margin-negative resection in locally advanced tumours in multivisceral resection has been shown to produce similar survival to cases where there has been no adjacent organ involvement in stage-matched patients [22].
The postoperative morbidity was also low as the rate of major complications (Clavien-Dindo ≥ 3) was only 10% (1/10). The rate of Grade B POPF was 20% (2/10) which is similar to reported POPF rates in the literature after PD and PD-RC [11,23]. There were no colonic anastomotic leaks in this series whereas previously reported rates after PD-RC were as high as 33% [11]. No patients in this series required reoperation or died within 90 days of surgery which confirms that this procedure can be performed safely in experienced hands. Birkmeyer and others have previously noted that there is a strong inverse relationship between adverse clinical outcomes, specifically postoperative mortality, and hospital volume in relation to complex cancer surgery [24,25]. This evidence has paved the way for centralization of complex operations such as PD to regional centres of excellence. Current evidence suggests a large part of the reduction in mortality at high-volume hospitals can be attributed to a reduction in 'failure to rescue' scenarios [8,26]. This requires robust institutional frameworks to recognize and react to complications after specialist surgery. All PD-RC patients at our unit are therefore admitted to a specialist surgical intensive care unit after This ensures that surgery takes place at a high-volume centre after coordinated discussion between specialists.
Factors identified in this series associated with worse prognosis on univariate survival analysis were perineural invasion and extra-colonic origin of the primary tumour on final histology. Pancreatico-duodenal tumours are known to be biologically more aggressive tumours than colonic tumours, with a greater rate of metastasis and therapy resistance [27]. Nodal status was not an independent factor affecting prognosis as in other series [11], but this is probably due to the heterogeneous case mix (e.g., cancer type and use of systemic therapy).
The highlighted case (patient 10) also demonstrates that radical single-stage multivisceral resection with negative margins can be performed safely on an urgent basis after referral to a specialist unit.
Previous reports of emergency PD-RC have shown that early and long-term outcomes can be similar to those of non-emergency PD-RC when performed by experienced surgeons [28].
The role of neoadjuvant therapy versus upfront PD-RC surgery has not been thoroughly evaluated in this setting due to the rarity of this presentation, but it would seem reasonable to consider downsizing tumours where the resection margin may be at risk. Although currently not recommended in international guidelines, previous clinical trials such as FOxTROT have shown potential benefit from NAC in increasing the likelihood of margin-negative resection and overall survival, compared to a direct-to-surgery approach, in locally advanced CRC [15]. It is unknown, however, whether NAC could downstage tumours sufficiently to avoid the potential morbidity of PD-RC altogether. Furthermore, a recent meta-analysis found previous trials of NAC were limited by a lack of randomization and heterogeneity in NAC regimens [29]. Therefore, patient selection and choice of NAC for locally advanced CRC prior to PD-RC is an important area of future research.
It is interesting to note that at least half of the tumours in this series had defective DNA mismatch repair (MMR), compared to an overall frequency of 15% among sporadic colonic tumours [30]. It is known that most sporadic MMR tumours occur in the right side of the colon which present later than left-sided tumours [31], and hence colonic adenocarcinoma requiring PD-RC may be more likely to harbour defective DNA MMR. While MMR-deficient CRC may have a more favourable stage-matched prognosis compared to MMR-proficient tumours [32], they are also relatively TA B L E 2 Univariate analysis of survival outcomes Although all PD-RC procedures in this case series were per- formed using an open approach, there has been a growing trend to perform PD using a minimally invasive approach (laparoscopic or robot-assisted) which may further reduce morbidity [34]. A totally laparoscopic technique to perform PD-RC has been described recently [35]. However, there is currently a lack of high-quality data confirming the safety or oncological superiority of a minimally invasive approach over an open approach for PD [36,37]; therefore this procedure remains under investigation at pancreatic surgery units with extensive experience in minimally invasive surgery. Future studies may also reveal the utility of extended colonic resection (complete mesocolic excision and D3 lymphadenectomy) combined with PD in improving clinical outcomes for locally advanced CRC. The oncological superiority and safety of this approach for CRC is still under investigation but low-quality evidence from a recent systematic review appears to suggest that better overall and disease-free survival can be achieved with a more radical approach, particularly for more advanced (stage 2 and 3) CRC [38].
The main limitations of this work are the small sample size and case mix heterogeneity from a single centre which may limit the generalizability of the results. The good postoperative outcomes may, in part, be attributed to the fact that the cohort was carefully selected and relatively young without major cardiovascular comorbidities. A future multicentre national audit of PD-RC practices and outcomes could reveal additional information on volume-outcome relationships and regional differences in practice (e.g. referral criteria, use of NAC, patient selection criteria) that could help inform and standardize practice nationally.
In conclusion, in a UK setting, en bloc PD-RC for locally advanced intestinal cancer can be performed safely with a high proportion of margin-negative resections and good long-term outcomes. Due to the rarity and complexity of this procedure, regional referral networks with standardized referral criteria to advanced MDTs at specialist units may be required for PD-RC and other types of multivisceral resection. This is based on the established relationship between volume and outcome in other complex surgical procedures. Pooling patients at high-volume centres could also aid recruitment to clinical trials and improve access to experimental therapies. Further research is needed to determine optimum patient and tumour characteristics for this procedure and better define the role of neoadjuvant therapy.

AUTH O R CO NTR I B UTI O N S
B.D., M.F. and S.H.Y. contributed to data collection, data analysis and drafting the paper. M.P., L.R.J., J.T.J. and D.R.C.S. contributed to conception and design, critical revision of the paper, and approval of the paper.

FU N D I N G I N FO R M ATI O N
None.

CO N FLI C T O F I NTE R E S T
None declared.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data generated or analysed during this study are included in this published article (and its supplementary information files).

E TH I C A L S TATEM ENT
As a retrospective service evaluation, NHS REC review was not required.