Para-aortic lymph node metastases in pancreatic cancer should not be considered a watershed for curative resection

No international consensus regarding the resection of the para-aortic lymph node (PALN) station Ln16b1 during pancreatoduodenectomy for pancreatic ductal adenocarcinoma (PDAC) has been reached. The present retrospectively investigated 264 patients with PDAC who underwent curative pancreatoduodenectomy or total pancreatectomy between 2005–2015. In 95 cases, the PALN were separately labelled and histopathologically analysed. Metastatic PALN (PALN+) were found in 14.7% (14/95). PALN+ stage was associated with increased regional lymph node metastasis. The median overall survival (OS) of patients with metastatic PALN and with non-metastatic PALN (PALN−) was 14.1 and 20.2 months, respectively. Five of the PALN+ patients (36%) survived >19 months. The OS of PALN+ and those staged pN1 PALN− was not significantly different (P = 0.743). Patients who underwent surgical exploration or palliative surgery (n = 194) had a lower median survival of 8.8 (95% confidence interval: 7.3–10.1) months. PALN status could not be reliably predicted by preoperative computed tomography. We concluded that the survival data of PALN+ cases is comparable with advanced pN+ stages; one-third of the patients may expect longer survival after radical resection. Therefore, routine refusal of curative resection in the case of PALN metastasis is not indicated.

. On univariate analysis, PALN status did not significantly correlate with the different overall survival data (P = 0.093). However, PALN status (P = 0.03) and failure to receive adjuvant treatment (P < 0.001) were the only significant factors associated with a shorter DFS on multivariate analysis. To compare para-aortic with regional lymph node metastases, we further stratified cases with non-metastatic PALN according to the pN-stage. The PALN− subgroup included 37 cases staged N0 and 44 cases with regional lymph node metastasis (N1). The median overall survival of the N0 subgroup was 23.9 (95% CI 18.8-48.0) months and significantly longer (P = 0.047), compared with N1 (median survival 19.3 months, 95% CI 10.9-22.9) or PALN+ cases (Fig. 1B). The overall survival of patients with PALN+ and those staged PALN− N1 was not significantly different (P = 0.743). Likewise, the PALN− N0 subgroup had a significant longer DFS (P = 0.013, Fig. 1C).
The new (8 th edition) classification of TNM staging of the Union for International Cancer Control (UICC) introduced a N2 stage for PDAC (≥4 regional lymph node metastases) because recent studies had underlined that the number of lymph node metastases significantly affected survival 6,7 . When we retrospectively applied the 8 th edition nodal staging system to the present cohort of patients, the median survival of the patients staged pN0, pN1, and pN2 was 23.9 months, 19.3 months, and 17.0 months, respectively. The pN2 subgroup (n = 24) included 8 patients with PALN metastases, and exclusion of these cases resulted in a median survival of 10.9 months (95% CI 9.4-28.0). In contrast, patients with 1-3 regional lymph node metastases (pN1 according to the 8 th edition of TNM staging) and no PALN metastases (n = 28) had a median survival of 19.5 months (95% CI 15.4-25.0). Thus, the survival prognosis of the analysed PALN+ patients was in-between the prognosis of pN1 and pN2 staged cases. To compare the survival probability of patients with metastatic PALN and palliatively treated patients, we further identified all patients who underwent surgical exploration or palliative bypass surgery for unresectable or metastatic PDAC within the study period. These 194 patients had a median survival of 8.8 (95% CI 7.3-10.1) months, which was not significantly shorter compared with the PALN+ subgroup (Fig. 1D). However, some patients with metastatic PALN demonstrated a long-term survival, not seen in the palliative setting: 5 patients (36%) with positive PALN survived more than 19 months (range 19.6-57.1), mirroring the median outcome of patients treated in curative intent.    (Table 4). Interestingly, the 2 cases had a high number of tumour-involved PALN (3 and 7), whereas the mean number of metastatic PALN was 1 in non-suspected (based on CT imaging) PALN (Table 4). 18 F-Fluorodeoxyglucose positron emission/computed tomography (FDG/PET-CT) scans were not performed routinely in the preoperative setting of patients with PDAC. A preoperative FDG/PET-CT was available for only 1 patient with PALN metastases, but the uptake values did not predict the PALN involvement in this case.

Discussion
The discussion of extended lymphadenectomy or PALN resection during PDAC resection started years ago, but is still an ongoing and controversial debate as demonstrated by the very recent literature. Consequently, no consensus has been reached on the management of PALN 3 . The present study and recent meta-analyses congruently demonstrate that approximately 14-18% of PDAC of the pancreatic head harbour PALN metastases at the time of resection 4,8 , and that this subgroup of patients has a relative poor prognosis with a median overall survival of 13-14 months. Recently, Hackert et al. published an overall median survival of 12.3 months after resection of synchronous oligometastatic PDAC (either liver or interaortocaval lymph node metastases) 9 . Taken together, these data further support that the Asian and European outcome data after PALN resection are comparable. These include the quality of PALN resection, indicated by the mean number of resected PALN: 6 (present study) compared with 4.3 10 and 5.0 11 in the Asian studies. Interestingly, our analysis showed that one-third of the patients with PALN metastases survived 19 months or longer after resection. This is in line with results from the largest multicentre study to date, which reported that 26 out of 102 patients with metastatic PALN survived for more than 2 years 10 . There are even reports of long-term survival after removal of PALN metastases 12 . When we compared the median survival duration of patients with palliative surgery (8.8 months) or PALN+ resection (14.1 months), patients in the latter subgroup demonstrated a longer survival of approximately 5 months, thus, supporting the resection of PALN (Ln16b1) at least in subgroups of patients. Based on the fact that PALN metastases are associated with early tumour recurrence, others have recommended that intraoperative sampling of PALN with frozen section analysis should be routinely performed, and consequently, that tumour resection in the case of such lymph node metastases should be avoided 5 . In contrast, we do not conclude that PALN metastases should be considered a general contraindication for curative resection or multimodal strategies, because of the lack of superior treatment alternatives. Moreover, the present data revealed that most frequently only one single lymph node of the Ln16b1 group was tumour infiltrated, which limits the prediction of intraoperative frozen section of single nodes. In future, the possibility of genetic subtype analysis may further define patients who would benefit from PALN resection 13 . The survival and DFS difference of patients with PALN metastases and regional pN1pM0 stage is marginal, and PALN+ cases are associated with a high burden of lymphovascular tumour spread. The survival prognosis is comparable to the new N2 stage of the UICC 8 th edition staging system, which should be evaluated in larger series. Therefore, PALN metastases may characterise a subgroup of PDAC with extensive lymph node metastases, which might benefit additionally from neoadjuvant treatment. The data further open the discussion whether PALN should be classified as regional lymph nodes for PDAC.
However, the preoperative prediction of PALN status is difficult, if not impossible 14 . Imai et al. could not detect PALN metastasis in any of their patients by preoperative imaging studies 15 . In the present study, only 2 cases with extensive PALN metastasis (3 and 7 Ln16b1 nodes infiltrated) were identified by the preoperative CT scans. Thus, in most cases the sensitivity of CT imaging is insufficient for prediction of PALN metastases. Other imaging modalities such as FDG/PET-CT or endoscopic ultrasound (EUS) have been investigated in staging of PDAC. Whereas FDG/PET-CT may be more sensitive in the diagnosis of primary PDAC and distant metastases compared with contrast-enhanced CT, its accuracy for detection of local nodal tumor metastases is poor and was reported between 37.5 and 42% 16,17 . The accuracy of EUS for staging of nodal involvement reached 65% or even higher in single studies, but a meta-analysis demonstrated its general limits in nodal staging based on the pooled accuracy parameters 18,19 . In summary, the accuracy of CT, FDG/PET-CT or EUS is insufficient to support selective treatment approaches, e.g. neoadjuvant therapy in case of FDG/PET-CT positive PALN.
The present study has some limitations that should be considered. Because PALN resection was routinely performed at our institution, there was no comparison of the morbidity of cases with and without PALN resection. One of the most relevant complications is the development of a POPF, which was diagnosed in 14.7% of all cases. We do not think that dissection of the PALN generally increases the risk for POPF because it does not affect the pancreatic neck dissection or the site of pancreatojejunostomy. A definite allocation of the resected lymph nodes to the Ln16b1 station (PALN) was only possible in 95 of the 264 cases, because PALN were not labelled routinely at the time of resection, which limited the available cases. Further, the low number of cases included for analysis of CT prediction of PALN status is insufficient to calculate valid test statistics, but was predetermined by the 10 available CT studies of patients with PALN metastases.
In conclusion, the present study confirms that patients with PALN metastases (pM1 LYM) have a relative poor prognosis, resulting in a median survival of approximately 14 months. However, one-third of these patients demonstrated a longer survival. Compared with other studies on PALN resection, the present study explicitly analysed the lymph node status and demonstrated that the survival of patients with positive PALN was comparable to patients with pN1pM0 disease, and even more to pN2pM0 patients according to the 8 th edition of TNM staging (UICC). Because preoperative imaging studies fail to reliably predict PALN metastases, PALN status prior to tumour resection can only be assessed by intraoperative frozen section analysis. However, abandoning tumour resection in the case of Ln16b1 metastases should not be routinely recommended; we would rather recommend a tailored approach taking into account the individual risk profile of the patient.

Methods
Study design and patients. Data were obtained from a prospective database and retrospectively analysed.
All patients who underwent partial pylorus-preserving PD (PPPD), classic PD (cPD), or total pancreatectomy (TP) for PDAC of the pancreatic head or neck between January 2005 and March 2015 at the Department of Visceral, Thoracic and Vascular Surgery, University Hospital Carl Gustav Carus, Technische Universität Dresden, Germany, were screened for inclusion. Although PALN resection was performed routinely, the resected lymph nodes could not be unequivocally allocated to the PALN in all cases during pathological workup. Only cases with unequivocally labelled PALN were included for further analysis. The patients were divided into two subgroups: patients with a PALN-positive status (histopathological tumour invasion of at least one PALN, PALN+) and patients without PALN metastasis (PALN−). For survival analysis, PALN-negative cases were subdivided according to the pN stage. Patients who underwent palliative surgery for unresectable or metastatic PDAC in the same observation period were likewise considered for comparative survival analysis. Tumour recurrence was assessed by regular follow-up examinations in the outpatient clinic, telephone interviews, or by contacting the respective primary physician of the patients. Tumour progression was assumed, if imaging modalities, e.g. sonography, computed tomography (CT) or magnetic resonance imaging (MRI) scans, or the clinical and laboratory examination, e.g. carbohydrate angigen 19-9 (CA 19-9) were indicative. Borderline resectability of the tumours was assessed using the National Comprehensive Cancer Network (NCCN) definition based on the preoperative CT scans 20 . The experimental protocol of the study was approved by the local Ethics Committee of the TU Dresden (decision number EK70022017). All methods were carried out in accordance with relevant guidelines and informed consent was obtained from the all included patients.
Operative technique. The basic operative technique of the PPPD and cPD was recently described 21 . After an extended Kocher manoeuvre, the PALN were dissected on the ventral aspect and between the vena cava and aorta, beginning at the right side of the coeliac trunk and origin of the superior mesenteric artery to the origin of the inferior mesenteric artery.
Pathological assessment. Para-aortic lymph node samples were examined macroscopically, dissected and all palpable nodes were submitted for histological examination. Histological analysis was carried out using haematoxylin & eosin slides without serial sectioning or immunohistochemical analyses for single tumour cells. A metastasis was diagnosed when tumour infiltration with desmoplastic reaction was visible.
Radiological assessment and patient matching. Radiological prediction of postoperative, histopathological PALN tumour invasion was investigated using preoperative CT imaging studies. Therefore, the 14 patients with PALN metastases were matched 1:1 to the patients within the PALN− subgroup, taking into consideration the following variables: age, body mass index (BMI), TNM stage, type of operation. An experienced radiologist, who was blinded with respect to the histopathological outcome, retrospectively analysed the available preoperative CT studies of the cases for assessment of PALN status. The para-aortic anatomical region of the Ln16b1 station was examined on venous-phase sequences. Lymph nodes >1 cm and those showing an inhomogeneous structure and/or irregular borders were considered suspicious for tumour invasion.