Extensive PAN lymphadenectomy means the first-tier and second-tier lymph nodes plus PANs are removed, with node clearance in a wide range and in large numbers. Given the high rate of approximately 20% of PANs with micrometastasis[7], which are not completely detectable by the current preoperative imaging examinations, prophylactic PAND measures were taken to prevent the relapse of local lymph nodes. However, many studies did not show survival benefit after prophylactic D2 + PAND[30,40]. Among these known studies, a large Japanese prospective randomized trial (JCOG 9501) investigated the efficacy of prophylactic PAND for curable patients with AGC, which was not considered justified for patients with AGC without improved survival. The study also proved that extensive PAN dissection performed safely by specialized surgeons did not increase the incidence of major surgical complications (anastomotic leak, pancreatic fistula, abdominal abscess, and pneumonia) and acquired a low 0.8% rate of hospital mortality. However, the operation time was prolonged, and the blood loss was increased with the D2 dissection[30]. However, some drawbacks existed in the JCOG9501 trials, in which patients with macroscopically involved metastasis were excluded. Nevertheless, whether harboring pathological micrometastasis was uncertain between the two groups, and the numbers of patients with pathological micrometastasis were not be balanced.

With the exception of the JCOG9501 study on the role of prophylactic PAND, studies on therapeutic PAND have also been performed by many surgeons with varying outcomes, as shown in Table 2. The surgical results of patients with PAN metastasis were disappointing[54], but many surgeons thought that the potential benefit of D2 plus PAND over standard D2 alone should not be ignored. For example, Hu et al[44] demonstrated that an improved survival was accomplished after extensive nodal dissection. In addition, many studies revealed that the advantages of patients who have experienced PAND with survival benefits were from related clinicopathological factors. Roviello et al[21] showed that the high probability of survival was closely dependent on pT and pN staging. The 5-year survival rate particularly worsened with the growth of pN stage under the premise of node-positive patients and worsened with increasing infiltration depth (pT staging) under the premise of node-negative patients. Better survival can be achieved by D2 + PAND for patients without node invasion (pN0) and those whose tumor was limited to subserosal invasion (pT2). However, other studies indicated that patients with tumor diameter measuring 50–100 mm and with pN2 staging might benefit from D3 dissection[55]. A Chinese study initiated by Zhang et al[53] demonstrated that D2 + PAND may be beneficial in patients with T3/T4 tumors with 1–3 PAN clinical involvement as a therapeutic method. Moreover, studies demonstrated that PAND may be beneficial in patients with a small number (< 3 or 4) of PAN[6,56] and total lymph node (< 11) metastases[15] or patients with < 15 total positive lymph nodes but macroscopic type, except type 4, on the basis of R0 resection[31]. With more nodes dissected, some specialists speculated that the improved survival of PAND may benefit from accurate staging information provided by extensive surgery[32,57,58]. Combined with these clinical factors, screening the best indications for PAND and then performing the operation by well-trained surgeons are both necessary.

Apart from this situation, for patients with proximal GC invading the esophagus, the left thoracoabdominal approach was compared with an abdominal–transhiatal (TH) approach, and TH was selected as the better method, which was recorded in the JCOG9502 trial[59]. Overall, the TH group who underwent a total gastrectomy with D2 and additional dissection of the left upper PANs showed better survival and less morbidity than those who underwent the left thoracoabdominal approach accompanied by a thorough lower mediastinal lymphadenectomy. Thus, the fact that some western surgeons do not advocate PAND based on partial outcomes is not reasonable, because the accurate value of therapeutic PAND in PAN-positive patients with curative purpose remains undetermined. Therefore, several rigorous large-scale trials are needed to further verify whether differences exist in survival between these two lymphadenectomies (D2 vs D2 + PAND) performed by experienced surgeons.

Systemic chemotherapy is regarded as the standard treatment for systemic macro- or micro- metastases involving the PAN region and beyond the PAN area. Nevertheless, systemic chemotherapy alone is unlikely to have a meaningful or lasting benefit in unresectable tumors, such as those with PAN involvement. However, adjuvant chemotherapy partly helps improve surgical survival[60].

Recently, neoadjuvant chemotherapy has gained considerable attention in the treatment of patients with distant metastasis. The following three studies, JCOG 0001, 0405, and 1002 were implemented to investigate the utility and the efficacy of neoadjuvant chemotherapy followed by gastrectomy with D2 + PAND. Among these studies, due to the same eligibility criteria but different regimens of preoperative chemotherapy followed by surgery, patients in JCOG 0001 received two or three cycles of irinotecan (70 mg/m² on days 1 and 15) and CDDP (80 mg/m² on day 1) therapy. In addition, patients in JCOG 0405 received two or three cycles of CDDP (60 mg/m² on day 8) and S-1 (40 mg/m² twice daily from day 1 to day 21 followed by a 1-week rest period) (CS) chemotherapy. JCOG 0001[61] showed a good 3-year survival of 27.0%, but the study was terminated due to three treatment-related deaths among the 55 enrolled patients. JCOG 0405[33] showed an excellent response rate of 64.7% and a 3-year survival of 58.8% with no treatment-related death. Since then, CS chemotherapy has been considered the current standard for patients with ELM, in which ELM was defined as PAN metastasis (no. 16a2/16b1) or bulky lymph nodes (one larger than 3 cm or two larger than 1.5 cm) along the celiac, splenic, common hepatic, or proper hepatic arteries, or both. Triplet therapy with the addition of docetaxel to CS (DCS) was then introduced as neoadjuvant chemotherapy for local patients with AGC and ELM (JCOG1002). In JCOG 1002[62,63], with the same eligibility criteria as the above two trials, patients received two or three 28-day cycles of docetaxel (40 mg/m² on day 1), CDDP (60 mg/m² on day 1), and S-1 (40 mg/m² twice daily for 2 weeks) (DCS) therapy. However, this regimen achieved a high rate of R0 resection and a 5-year survival of 54.9% (95% confidence interval: 40.3%-67.3%), with an insufficient pathological response rate of 50.0% (26/52).

Controversy still exists regarding the best regimen of chemotherapy for patients with PAN metastasis (Table 3). Notwithstanding the JCOG 1002 study of DCS that failed to show superiority over CS[62,63], many studies were carried out to explore its benefits. A triplet therapy of docetaxel added to CDDP and S-1 (DCS) showed longer survival of patients with PAN metastasis when compared with 5-fluorouracil. Likewise, many other studies incorporated other regimens of different chemotherapeutics. A phase II trial performed in China, which adopted capecitabine and oxaliplatin as preoperative chemotherapy delivered for a maximum of six cycles, introduced the concept of conversion therapy to treat PAN metastasis in patients with AGC. The results demonstrated a good response rate and a sufficient R0 resection rate, with acceptable toxicities[64].

Many trials have investigated the outcome of different schemes of preoperative chemotherapy, showing different survival effects. Currently, as the tentative standard chemotherapy in Japan[33], CS together with subsequent radical surgery is still considered the de facto standard treatment for patients with AGC and ELM. Thus, further investigations on appropriate regimens and suitable durations of perioperative chemotherapy should be used in clinical practice for better survival.

According to the fundamental role of surgery in the treatment of GC, radiotherapy is rarely used, and almost no research is available on radiotherapy alone used for AGC and is always a part of a comprehensive treatment in combination with other palliative interventions. Radiotherapy is commonly combined with chemotherapy before surgery or implemented concurrently with or subsequently to adjuvant chemotherapy after surgery in patients with GC[65-68].

Nonetheless, treatment measures including radiotherapy aimed at curing patients with PAN involvement are rare. An individualized comprehensive treatment including neoadjuvant chemotherapy, subsequent surgery, and radiotherapy for patients with AGC and PAN metastasis with a high response rate of 76.1% for positive PANs and without treatment-related death is beneficial[68].

The overall survival rate varies greatly between studies due to different treatments. Within the scope of an 8.1%-51% incidence of PAN metastasis, the 5-year survival of patients ranges from 43.7% to 70.3%, as listed in Tables 2 and 3.

Considering the rare application of radiotherapy and other unconventional treatments, discussion on the adverse effects of chemotherapy and surgery is included. Many trials on the impact of surgery plus extensive lymphadenectomy on prognosis were carried out to explore its benefit on survival. In terms of interim/short-term outcomes, studies pointed out that extended lymphadenectomy could influence the function of adjacent abdominal organs and induce high postoperative morbidity and mortality[30]. In addition, reduced risk can be achieved by preserving the spleen and/or pancreas[17,69]. For instance, Kunisaki et al[13] indicated that pancreatic fistula and respiratory complications were significantly higher in patients with D2 + PAND as compared with standard D2. Conversely, no differences in surgical morbidity between D2 and D3 lymphadenectomy were found by several studies performed by experienced surgeons[14,17,18,20,44,70,71]. In addition, several European studies also reported no association between postoperative mortality and extended lymphadenectomy[20,57,72]. Abdominal abscess, anastomotic leakage, pancreatic fistula, abdominal abscess[17], and pneumonia[21,30,73,74] were reported as the most common complications after extensive surgery observed in studies. In addition, these morbidities were highly related to the American Society of Anesthesiologists’ class II/III vs I, perioperative blood transfusions, low albumin serum levels, and age (> 75 years). The degree of radical surgery was regarded as an independent predictor of mortality by Marrelli et al[17]. Other less common complications, such as diarrhea, orthostatic hypotonia, and lymphocele or lymphorrea, were serious, and measures should be taken to reduce these complications[30,73]. However, many studies found that PAND could increase the operation time and blood loss, required greater blood transfusion[19,20,30,53,74], had high relaparotomy[19,20,30,74], and could prolong hospital stay[55] with no harmful effect on quality of life.

At the same time, the side effects of chemotherapy combined with surgery during chemotherapy were monitored by numerous trials. Furthermore, neoadjuvant chemotherapy subsequent to surgery can lead to different adverse events during chemotherapy due to the use of different chemotherapy regimens. Common adverse events but different incidences were recorded in three JCOG trials (JCOG 0001, 0405, and 1002) with similar inclusion criteria but different preoperative chemotherapy regimens[33,61,62]. Among these adverse events, grade 3 or 4 toxicity during chemotherapy included leucopenia (31% vs 4% vs 18.9%), neutropenia (55% vs 19% vs 39.6%), anemia (24% vs 13% vs 7.5%), febrile neutropenia (16% vs 2% vs 5.7%), nausea (36% vs 4% vs 1.9%), diarrhea (5% vs 2% vs 7.5%), thrombocytopenia (4% vs 1.9%), anorexia (10% vs 9.4%) for 0405 and 1002, vomiting (13%) for 0001, anorexia (10%) for 0405, hyponatremia (15.1%), hypokalemia (5.7%), and upper respiratory tract infection (1.9%) for 1002 (Table 4). Two chemotherapy-related deaths (4%) among all 55 patients in 0001 were observed, and no chemotherapy-related deaths were noted in the other two trials. Only grade 3 adverse events without grade 4 toxicities were stated by Wang et al[64], and gastrointestinal issues and leukocytopenia were the most common. These conditions were the result of preoperative capecitabine and oxaliplatin chemotherapy followed by D2 gastrectomy without PAND for AGC patients with PAN involvement[64]. The outcomes from a retrospective study, showed that neutropenia (25.0%), leucopenia (18.8%), febrile neutropenia (6.3%), and diarrhea (6.3%) were the most common grade 3/4 toxicities, and no treatment-related deaths were observed[60].

Relapse rates remain high even with extensive lymph node dissection; thus, there is a need for other adjuvant treatments[72,75]. Lee’s research[54] concluded that more than 70% of patients with positive PANs relapsed within 11 months after surgery. Among the seven patients in the trial, two developed recurrences in local regions, and the other five patients developed distant metastases to the lung, bone, and left supraclavicular lymph node. Following D2 and neoadjuvant chemotherapy, no one had PAN recurrences. However, different to the above study, peritoneal metastasis followed by extra-regional nodal recurrence were identified as the two most frequent sites of relapse[37,73,76]. Among these sites, more than one site including the peritoneum, lymph nodes, liver and other areas[37] were involved at the time of first recurrence in the JCOG 9501 study. Moreover, a multi-institutional study by Kunisaki et al[55] indicated that recurrences in the surgical resection nodal area was significantly lower following D3 dissection, even with a similar overall nodal recurrence rate between D2 and D3 dissection. This condition may be the result of significant differences in the distribution of recurrent lymph nodes between D2 and D3 patients due to the numbers of lymph nodes in the second and third tiers, hepatic hilar region, and mediastinal or cervical regional lymph nodes. Metastasis may also recur in other regions such as the right supraclavicular lymph node[77,78]. However, after preoperative chemotherapy followed by surgery, the relapse-free survival rate can be as high as 70% at 2 years as shown by Oyama et al[60].