Esophageal cancer is one of the deadliest cancers with rapidly rising incidence. In our previous report, tumor recurrence after curative esophagectomy developed in 42.9% of patients in the median of only 10 months. As high as 66.2% of recurrences happened within 1 year after operation . The reported prognostic factors of PRS included liver recurrence, shorter disease-free interval and palliative therapy. Distant recurrence and more than 3 recurrent locations were associated with worse PRS, as demonstrated in K. Parry et al’s study . In current study on patients with oligo-recurrence after esophagectomy, we identified pathological T3 stage and mediastinal recurrence as the prognostic factors of PRS.
In general, patients with multiple recurrent sites had a worse survival compared with those with less involved sites . Combined failure pattern of simultaneous locoregional and distant recurrences also had inferior outcome . Indeed, patients with recurrent esophageal disease deserved multimodality therapy for better outcome. Furthermore, based on the rising publications showing evidence that oligo-recurrence may represent tumors with more favorable biology, early identification and aggressive treatment for oligometastatic recurrence might improve survival [8–12]. However, the role of surgery in patients with isolated oligo-recurrence has not yet been defined.
In Ghaly et al’s study , no pronounced difference was found in disease-free survival or in PRS between patients with oligo-recurrence treated with operation, with or without chemo- and radiotherapy, and patients who received definitive chemoradiotherapy without resection. In contrast, the study by Depypere et al  demonstrated prolonged survival in patients with isolated locoregional recurrence or single solid organ metastasis, especially if surgery was offered. Surgical resection (+/- systemic therapy) of solitary recurrent lesion was a considerable therapeutic option for well-selected patients with recurrent esophageal carcinoma. Similarily, Ohkura et al.  reported a significantly better OS rate in the patients who underwent resection of oligo-recurrences than in those who did not. However, they found no significant difference in survival between the patients with shorter and longer DFI (< 12 months vs. > 12 months), indicating that a shorter DFI should not be an unfavorable prognostic factor to hinder surgeons from choosing surgical resection. In line with these reports, we observed that surgical resection for oligo-recurrence may lead to better OS and PRS, albeit without statistical significance. However, in subgroup analysis, we found that surgery, compared with non-surgical treatments, may confer survival benefit to patients without comorbidities, highlighting the importance of patient selection in deciding treatment modalities.
Cervical and mediastinal lymph node failures have been reported to be one of the most common types of recurrence after esophagectomy in patients with thoracic esophageal squamous cell carcinoma [3, 13]. In Ni’s study , lymph node recurrence above the diaphragm and single region lymph node recurrence exhibited better OS than those at the subphrenic region and multiple regions, respectively. They also identified original pathological stage and salvage treatment regimen as independent prognostic factors. Whereas chemoradiotherapy could offer a safe and effective treatment for patients with lymph node recurrences, especially with a single region failure [14, 15], many reports suggested that salvage cervical lymphadenectomy as the main treatment could achieve locoregional disease control and prolong survival in patients with cervical LN recurrence after curative esophagectomy . Focusing on lymph node recurrence, Nakamura et al. have shown significantly better survival in the lymphadenectomy and chemoradiotherapy groups than in the patients who received chemotherapy or best supportive care for lymph node recurrence after curative esophagectomy . However, there was no statistically significant difference in survival between the surgical lymphadenectomy and chemoradiotherapy groups. Of note, 11 of 12 patients with cervical lymph node recurrence received surgical resection, whereas less than one third of patients with paraesophageal or paratracheal lymph node recurrences had surgical resection. These results were compatible with our findings that only one of 11 patients with isolated mediastinal lymph node recurrence received surgery. Mediastinal lymph node recurrence was also a significant prognostic factor of both OS and PRS. Therefore, the possibility and benefit of salvage surgical resection for mediastinal LN recurrence remains unclear and needs more data to clarify.
Limited reports on oligo-recurrence to distant solid organ have made it difficult to collect sufficient cases for analysis. The role of surgery in these patients thus remains unknown. For example, only 30% of the patients with oligo-recurrence had surgery in Nobel’s study . However, several reports have recommended pulmonary metastasectomy as an acceptable and effective treatment for solitary pulmonary metastasis [18–21]. For example, Kobayashi et al. analyzed 23 patients who underwent 30 curative pulmonary metastasectomies at a single institution . In their report, the unfavorable prognostic factors included history of extrapulmonary metastases before pulmonary metastasectomy, poorly differentiated primary esophageal carcinoma, and short disease-free interval. Their results also recommended that pulmonary resection for lung metastases from esophageal carcinoma should be considered in carefully selected patients, and repeated metastasectomy was encouraged in suitable patients.
On the other hand, Nobel et al. have reported significantly worse outcome in patients with liver and brain oligo-recurrence from esophageal cancer when compared with lung oligo-recurrence, which showed a more indolent course . In our series, patients with isolated lung metastasis had better OS (HR: 0.57, p = 0.086) and PRS (HR: 0.60, p = 0.125) compared with patients with oligo-recurrences in other sites. Moreover, 17 of 21 patients had pulmonary resection, compared with 5 of 27 patients in Nobel’s cohort . On the contrary, patients with brain oligo-recurrence had worst OS (HR: 3.83, p = 0.031) and PRS (HR: 2.69, p = 0.105) compared with patients with oligo-recurrences in other sites, which are compatible with the findings in Nobel’s cohort . Whether patients with esophageal cancer should be screened or surveyed for brain metastases remains unclear [22–24]. Although a previous study  has reported that half of brain oligo-recurrences occurred within 12 months of esophagectomy and all were diagnosed because of symptomatic disease, which led to the suggestion of brain surveillance imaging for high risk patients, there was report showing low incidence of brain metastasis in patients with esophageal carcinoma, which made it unnecessary for baseline screening or surveillance, even the prognosis was poor .
There are several limitations in our study. First, this is a single-institution study. The inherent bias of retrospective nature and relatively small sample size may limit the power of statistical significance. Second, due to the lack of strict definition for “resectable” and guidelines for therapeutic approach in each recurrence site, selection bias for treatment modalities and surgical intervention are inevitable. Third, routine surveillance for bone and brain was not performed in our practice, which might miss early diagnosis of oligo-recurrences in these sites. Finally, since only patients with single site of recurrence were selected, the role of surgery in combination with other aggressive local control or systemic therapy in patients with limited or multiple sites of recurrence was beyond the scope of our study. External validation study with more cases is needed to confirm our observations.