A refined procedure for esophageal resection using a full minimally invasive approach

We conducted a retrospective review of consecutive cases of Kaiser Permanente Northern California (KPNC) patients who underwent this streamlined MIE procedure and perioperative-care program (described below) and detailed in the accompanying document (Additional file 1). KPNC is an integrated health plan serving 4.4 million members through a network of 21 medical centers in Northern California. The demographics of KPNC approximate the underlying demographics of the geographic region [48]. All study procedures were approved by the KPNC Institutional Review Board.

Eligible patients were KPNC members referred for esophagectomy who underwent this MIE procedure at the KPNC Oakland Medical Center by one of five board-certified attending thoracic surgeons between January 2013 and August 2018; additional eligibility criteria included a surgical indication of primary esophageal cancer, a requirement that the procedure was elective, that the surgery consisted of the MIE only (i.e., not combined with another procedure during the same surgery), and that the patient had 90 days of postoperative follow-up at the time of data analysis (patients who died within 90 days of surgery were not excluded). All patients meeting these eligibility criteria were included and full ascertainment of all relevant patients was achieved by a search of the electronic medical record (EMR), operating-room records, and a list maintained by the first author. All patients who underwent the procedure in the specified interval were included in the final cohort, including those with extremely advanced disease for whom surgery was performed at the patient's request and were not expected to survive. Postoperative follow-up for surgical outcomes (including complications and mortality) was carried out for 90 days postoperatively on all patients; mortality follow-up was carried out to a maximum of 84.0 months (median: 29.0 months; survival curves were truncated when fewer than 10 patients remained at risk); data on mortality were obtained from the EMR and state and federal mortality databases.

Perioperative care was guided by strict adherence to ERAS protocols, which included patient education with preoperative nutritionist consultation, mandatory perioperative medicine clinic consultation, carbohydrate loading with clear liquids 2-to-4 h preoperatively, avoidance of perioperative use of sedating medications, and use of nonopioid multimodal analgesia. Postoperatively, the urinary catheter and nasogastric tube were removed on the first day and a clear-liquid diet started and early ambulation instituted. Detailed descriptions of all aspects of the KPNC ERAS protocols are provided elsewhere [49, 50]. Most patients are discharged on postoperative day 3 (patients are eligible for discharge on day 3 if they have no leak on the esophagram, absence of tachycardia or fever, are able to sustain hydration and nutrition with over 2 L of fluid orally per day, their pain is adequately treated with oral medication, and they appear clinically well). Patients call or text the surgeon daily for 10 days after discharge after which they are seen in clinic; physician assistants assist in monitoring discharged patients.

The procedure described in this paper is classified as “IVL-LV” in the taxonomy of a recent consensus statement [51]. Briefly, the MIE in this study was performed in the following manner: The abdominal portion was conducted through a 15 mm, 10 mm, three mm ports and a 5 Nathanson liver retractor. The camera was operated by a surgical trainee. Both surgeon and assist employed two instruments. The hiatus was mobilized and Penrose drain placed, the greater curvature mobilized, the left gastric divided with a stapler and the stomach partially tubularized with three thick tissue stapler loads. Jejunostomy tube placement, the pyloroplasty, and the Kocherizaton of the duodenum were not done. To avoid trauma to the gastric conduit it was never grasped by instruments. The thoracic portion was conducted through four 10 mm ports and a 40–50 mm utility port. The pleura was incised caudal to cephalad and mobilized off the spine. The Penrose was retrieved at the hiatus and the esophagus and periesophageal tissue mobilized. A longitudinal incision was made in the upper esophagus to allow insertion of the anvil, the esophagus transected, and the anvil secured with two endoloop sutures. A gastrostomy was made on the lessor curvature and the stapler inserted. An anastomosis was created as proximal as possible and the tubularization complete leaving a 2 cm bridge. The pleura was closed with interrupted 2.0 silk sutures cranial to caudal.

Patient-related data included demographics and clinical comorbidities (i.e., specific comorbidities and the Charlson-Quan Comorbidity Index summary score [52]), American Society of Anesthesiologists (ASA) preoperative risk score (range 1–5) [53], Eastern Cooperative Oncology Group (ECOG) performance-status score (range 0–5) [54], and neoadjuvant therapy received. Tumor-related variables included cell type, primary esophageal site, clinical and pathologic stage (consistent with the American Joint Committee on Cancer (AJCC)/Union for International Cancer Control (UICC) staging system, 8th edition, 2017 [55]) and regional lymph-node status. Variables related to the surgical procedure included both full operating-room time and “skin-to-skin” time (time between first incision and final closure), estimated blood loss, jejunostomy status, pleural-closure status and whether an intra-pyloric botulinum injection was used. Outcome variables included LOS for the overall hospitalization and for the ICU specifically as well as 30-day incidence of hospital readmission, emergency-department visits, whether the patient was discharged with a feeding tube in place, reoperation, and all-cause and cancer-specific mortality. Perioperative complications and post-discharge complications were all reviewed by the first author (SA) and categorized by the U.S. National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) system, v5.0 [56].

Data for this study were obtained by extraction of relevant variables from the EMR database supplemented by manual chart review for those variables not available in discrete electronic form. Some cancer-related variables were obtained from a KPNC Cancer Registry maintained by the KPNC Division of Research for research purposes and for reporting to the Surveillance, Epidemiology, and End Results (SEER) program of the U.S. National Cancer Institute [57]. Data domains extracted from the EMR included demographics, clinical comorbidities, overall hospital ICU length of stay, readmission rates, and mortality. Variables requiring manual chart review included clinical and pathologic staging, lymph-node involvement, specific procedure-related variables, operative times from anesthesia records, and procedure-related complications. Manual chart data extraction was conducted by SA, AA, and BH.

Continuous variables were summarized as means with Wald 95% confidence intervals and/or medians with the associated interquartile ranges (IQRs). Categorical variables were summarized with percentages and 95% confidence intervals (95% CI) were calculated with the exact Clopper-Pearson method [58]. Survival data for both total and cancer-specific mortality were summarized with Kaplan–Meier failure-time plots [59]. Comparisons of surgeries between the first and second halves of the cohort were conducted with nonparametric Wilcoxon rank-sum tests for continuous variables and Fisher's exact tests for categorical variables.

One hundred fifty-two KPNC patients underwent this MIE procedure-and-care program between January 2013 and August 2018; of these, 10 patients were excluded for the following reasons: seven cases were complex procedures in which the MIE was combined with a second procedure during the same surgery (e.g., adrenalectomy, colon transposition, or lung lobectomy), two were unplanned salvage procedures, and one case was performed for recurrent esophageal cancer. The final analytic cohort was comprised of the remaining 142 patients (Table 1). No patient had a planned elective esophagectomy at this medical facility by any other procedure (e.g., open esophagectomy). All patients underwent the procedure for esophageal cancer, nearly all of whom had disease in the lower third of the esophagus or at the gastroesophageal junction (90.8%) and had adenocarcinoma histology (89.4%). Eighty-five cancers (59.9%) were clinical stage III and 32 (22.5%) were clinical stage IVA. The mean age of all patients was 64.5 years and the majority (85.2%) were men. One hundred twenty-five (88.0%) received some form of neoadjuvant therapy.

Of the 142 procedures, 1 required conversion to an open laparotomy due to intraoperative technical difficulties. The median total patient operating room time was 290 min (IQR: 255 to 348 min; mean: 306.5 min) and the median “skin-to-skin” time was 237 min (IQR: 210 to 290 min; mean: 254.6 min) (Table 2). One hundred fifteen patients (81.0%, 95% CI: 73.6% to 87.0%) did not have a jejunostomy placed during or after surgery.

The median hospital LOS was 3 days (IQR: 2 to 4 days); the mean was 3.6 days (95% CI: 3.1 to 4.0 days). Only 8 patients (5.6%, 95% CI: 2.5% to 10.8%) required postoperative care in the ICU (all were unplanned) and, of these, only 1 patient required care for more than 4 days (Table 3).

Twenty-two patients (15.5%, 95% CI: 10.0% to 22.5%) suffered a total of 25 complications within 30 days following the surgical procedure (Table 3); 14 patients (9.9%) experienced a complication with a CTCAE severity score of ≥ 3. The most common complications were pneumonia, atrial fibrillation, and anastomotic leaks (the frequencies of all complications are detailed in Table 3). In addition to these complications, 21 patients (14.8%) had a postoperative anastomotic stricture (i.e., that required dilation any time during the full extended follow-up period).

Within the 30 days after discharge, 14 patients (9.9%, 95% CI: 5.5% to 16.0%) were readmitted to the hospital. Thirty-six patients (25.4%, 95% CI: 18.4% to 33.3%) were evaluated in the emergency department, and 6 patients (4.2%, 95% CI: 1.6% to 9.0%) were taken back to the operating room (three for anastomotic leak, one for gastric conduit loss, one for empyema, and one for tracheostomy and jejunostomy to treat respiratory failure). No patient died intraoperatively or prior to discharge (0%, 95% CI: 0% to 2.6%); 1 patient (0.7%, 95% CI: 0.02% to 3.9%) died within 30 days post-procedure; 3 additional patients died between 31 and 90 days postoperatively (full 90-day mortality: 2.8%, 95% CI: 0.8% to 7.1%). The Kaplan–Meier product-limit estimate of the longer-term total mortality experience of this cohort is shown in Fig. 1 (total median survival was 4.6 years); cancer-specific survival is shown in Fig. 2.

Comparing surgeries performed in the earlier half of the cohort (prior to January 20, 2016) to the latter half, we found a reduction in the median hospital length of stay (4 days vs. 2 days, p < 0.001), median skin-to-skin procedure time (259 min vs. 229 min, p = 0.01), median number of lymph nodes excised (18 nodes vs. 14 nodes, p = 0.002), and risk of readmission within 30 days of discharge (11 (15.5%) vs. 3 (4.2%), p = 0.046).