Survival outcomes fifteen years after minimally invasive esophagectomy

Abstract

Background and objectives

Minimally invasive esophagectomy (MIO) offers a less traumatic resection for cancer patients resulting in improved quality of life. Concerns about the oncological efficacy of the procedure and potential impact on survival may have limited its wider adoption. This study reports survival outcomes fifteen years after patients underwent a total MIO for esophageal cancer.

Methods

A single-centre analysis of survival outcomes was conducted on all patients who underwent MIO between 2004 and 2010 and had completed at least 15-years follow-up. Actual overall survival (OS) and disease-free survival (DFS) were evaluated with the Kaplan–Meier method. The pattern of association of patient factors with survival was assessed with the Cox regression analysis.

Results

A total of 121 patients underwent resection, with 4 (3.3%) in-patient deaths. With a median follow-up time of 15.2 years, the median OS was 41 months. At 15-years there were 20 survivors (17.1%), with numbers at 10 and 5 years being 34 (29.1%) and 45 (38.5%) respectively. Median DFS was 27 months, with 19 (16.2%) patient disease free at 15 years, and 32 (27.4%) and 38 (32.5%) at 10 and 5 years respectively. Predictors of poor survival were the intracorporeal fashioning of the gastric conduit, perioperative blood transfusions, advanced disease stage and recurrence.

Conclusions

Long term survival outcomes following MIO support its increasing recognition as the standard of care as for curative resection in esophageal cancer.

Background

Surgical resection is fundamental to the curative management of esophageal cancer. Open esophagectomy involves multi-stage access through the abdomen, chest and neck. Variations of technique are in current use, but no one procedure is the standard of care. These are highly invasive procedures, associated with significant mortality, morbidity and diminished health-related quality of life [1, 2].

Minimally invasive esophagectomy (MIO) procedures, incorporating contemporary scientific and technological advances have increased in popularity over the last two decades. Comparative studies demonstrate that MIO is associated with reduced perioperative complications and enhanced quality of life. Survival and oncological outcomes in the short-term and up to 5 years appear to be equivalent to open surgery [3, 4] but data beyond this time is limited [5].

This study reports on the long-term survival outcomes of patients who underwent total MIO for cancer with curative intent between 2004 and 2010.

Materials and methods

Patients

Local institutional review board approval was obtained for the study. All methods were carried out in accordance with relevant guidelines and regulations. This was a single-centre survival analysis of patients who underwent MIO with curative intent, from April 2004 to January 2010. Data was collected from a prospectively maintained database and patients’ hospital records. Study period end was July 2023. Subjects with incomplete data were excluded. Patients' consent to the study was not obtained, as data were anonymised and no identifying details were used.

Intervention

Patients with a diagnosis of loco-regional and potentially curative esophageal or esophago-gastric cancer were discussed at the multi-disciplinary team meeting. MIO was offered to all suitable patients as an alternative to two-stage laparotomy / right thoracotomy (Ivor Lewis) open operation. Neo-adjuvant chemotherapy (two cycles of cisplatin and 5-fluorouracil) was offered to all the subjects, based on the recommendations of the OE02 trial, unless clinically contraindicated [6].

At the time of surgery, patients had a naso-jejunal feeding tube inserted endoscopically. In case the positioning was not satisfactorily achieved, a feeding jejunostomy was fashioned instead with the standard technique [7]. Three-stage total MIO comprised a thoracoscopic esophageal mobilisation and lymphadenectomy, followed by laparoscopic gastric mobilisation, lymphadenectomy, resection/conduit formation and finally a cervical anastomosis based on the technique described by Luketich and colleagues [8]. Subsequent modifications included the use of ischaemic pre-conditioning and extracorporeal gastric conduit formation as prospective randomised controlled trials [9].

Data

Clinical, surgical and pathological details were collected into a prospectively maintained database. Mortality was defined as any cause of death within 30 days from the index procedure. Complications were graded according to the Clavien-Dindo classification [10].

Tumour characteristics were described in accordance with the American Joint Committee on Cancer 6th and 7th Edition [11, 12]. The cancer site was defined by the distance between the incisors and the tumour epicentre and included the cervical esophagus (≤ 20 cm), the upper thoracic esophagus (20–24 cm), the mid thoracic esophagus (25–29 cm), lower thoracic esophagus (30–39 cm), gastro-esophageal junction (≥ 40 cm) [13]. The number of lymph nodes harvested was categorised as 0, 1–23, ≥ 24 [14]. Longitudinal resection margins were positive if cancer cells were within 1 mm from tissue edges, circumferential margins were not assessed as resection specimens were dissected into lymph node groups immediately after removal [15].

Survival outcomes were calculated from the date of the decision to treat, with overall survival (OS) the time to death from any cause, and disease-free survival (DFS) the time to first clinical evidence of primary cancer recurrence, and cancer-specific survival (CSS) the time to death from a cancer-related cause. Follow up was calculated from the date of completion of treatment.

Statistical analysis was conducted using Prism version 9.1.0 (GraphPad Holdings LLC, California) and R version 3.6.1 (R Foundation for Statistical Computing, Wien, Austria). Medians were evaluated with the Kruskall-Wallis test, differences in frequencies with the Chi-square and Fisher’s exact tests. OS and DFS estimates were assessed with the Kaplan–Meier method. The probability of survival of those who were alive at 5 years was computed with the conditional survival analysis. The Cox regression model was used to test the association of patient factors to OS and results were reported as hazard ratios (95% confidence interval [CI]). Significance was set at p < 0.05.

Results

A total of 132 patients underwent MIO. Of these, 11 (8.3%) had incomplete data and were excluded from analysis, leaving 121 subjects in the study. Demographic, clinical, and pathological data are summarised in Table 1.

Table 1.

Clinical, surgical and pathological characteristics

Parameter	N = 121
Median age (range)	67 (42 −84)
Male:Female	102:19
Tumour site (%)
Mid thoracic esophagus (25–29 cm)	3 (2.5)
Lower thoracic esophagus (30–39 cm)	86 (71.1)
Gastro-esophageal junction (≥ 40 cm)	32 (26.4)
Neoadjuvant chemotherapy (%)	96 (79.3)
Ischaemic conditioning (%)	62 (51.2)
Extracorporeal gastric conduit formation (%)	38 (31.4)
Estimated blood loss in ml, median (range)	500 (150–1500)
Operation duration in minutes, mean (range)	437.5 (280–745)
Postoperative adverse events (%)	33 (27.3)
Clavien-Dindo
Major
Death	4
Bleeding	2
Anastomotic leak	5
Gastric conduit necrosis	4
Anastomotic stricture	1
Chylothorax	4
Minor
Pneumonia	3
Atrial fibrillation	2
Transient vocal chords palsy	6
Raised temperature	1
Delirium	1
Pathology
Adenocarcinoma (%)	97 (80.2)
Well-differentiated	2
Moderately differentiated	55
Poorly-differentiated	40
Squamous cell carcinoma (%)	15 (12.4)
Well-differentiated	0
Moderately differentiated	7
Poorly-differentiated	8
High-grade dysplasia (%)	5 (4.1)
No cancer (complete response) (%)	4 (3.3)
Lymph node harvested, number (%)
0	1 (0.8)
1–23	87 (71.9)
≥ 24	33 (27.3)
TNM Stage (%)
0	9 (7.4)
1	15 (12.4)
2	46 (38.1)
3	51 (42.1)
4	0
Resection margin
Positive LRCa (%)	6 (5.4)

^aLongitudinal resection margin

Data from the 11 subjects who were excluded are summarised in Table 2.

Table 2.

Excluded patients’ characteristics

Patient	Sex	Age	Site	Neoada	IPb	Conduit	EBLc	Blood Txd	Op timee	Histf	AJCCg	N + h	LRMi	Clavien-Dindo	Complication type	Hospital death
1	M	67	n/aj	N	N	Intracorporeal	500	N	575	HGDl	0	0	Negative	0	None	N
2	M	70	lower	Y	N	Intracorporeal	500	N	382	SCCm	0	0	Negative	0	None	N
3	M	81	GEJk	N	N	Intracorporeal	500	N	364	ACAn	3	2	Negative	0	None	N
4	M	44	GEJk	Y	Y	Intracorporeal	700	N	415	ACAn	1	0	Negative	0	None	N
5	M	76	lower	Y	Y	Intracorporeal	500	N	341	ACAn	3	1	Negative	0	None	N
6	M	61	n/aj	N	N	Intracorporeal	450	N	406	HGDl	0	0	Negative	3a	RLNPo	N
7	M	64	lower	N	Y	Extracorporeal	500	N	500	ACAn	1	0	Negative	3a	RLNPo	N
8	F	57	mid	N	Y	Extracorporeal	400	N	480	ACAn	1	0	Negative	0	None	N
9	F	64	lower	N	Y	Extracorporeal	500	N	420	HGDl	1	0	Negative	3a	Chylothorax	N
10	M	65	lower	N	Y	Extracorporeal	n/aj	N	430	HGDl	0	0	Negative	2	Pneumonia	N
11	M	72	lower	Y	N	Extracorporeal	n/aj	N	690	SCCm	2	0	Negative	3a	Chylothorax	N

^aNeoadjuvant chemotherapy. ^bIschemic preconditioning. ^cEstimated blood loss. ^dBlood transfusion. ^eDuration of the operation in minutes. ^fHistology. ^gAmerican Joint Committee on Cancer. ^hNumber of positive lymph nodes. ⁱLongitudinal resection margin. ^jNot available. ^kEsophago-gastric junction. ^lHigh-grade dysplasia. ^mSquamous cell carcinoma. ⁿAdenocarcinoma. ^oRecurrent laryngeal nerve palsy

The median age at the time was 67 years (range 42–84) and there were 102 men (84.2%). Ninety-six (79.3%) had neo-adjuvant chemotherapy.

All the procedures were successfully completed as a total MIO. There were 4 in-hospital deaths on postoperative day 2,14, 37 and 46, respectively, giving a mortality rate of 3.3%. Causes of death were pulmonary embolism (1), acute respiratory distress syndrome (2), sepsis secondary to small bowel leak from the jejunostomy site (1). Complications occurred in 29 (23.9%) patients, 16 (13.2%) were graded as major and included bleeding (2), anastomotic leak (5), gastric conduit necrosis (4), anastomotic stricture (1), chylothorax (4).

Histopathology found 97 (80.2%) with adenocarcinoma, 15 (12.4%) with squamous cell carcinoma, 5 (4.1%) with high-grade dysplasia, and 4 (3.3%) with complete pathological response to chemotherapy with no cancer detected.

Stage III disease was most common with 51 (42.1%), followed by stage II with 46 (38.1%) then stage I 15 (12.4%) and 9 (7.4%) patients with stage 0.

Survival outcomes

The survival outcomes are summarised in Table 3.

Table 3.

Overall and Disease-Free Survivals

	Stage 0 n = 8	Stage 1 n = 15	Stage 2 n = 44	Stage 3 n = 50	Overall n = 117	P
Median OS in months (range)	184.5 (85–232)	110 (12–201)	43 (11–216)	30 (11–196)	41 (11–232)	0.00001*
Alive at 15 years (%)	5 (62.5)	4 (26.7)	9 (20.5)	2 (4)	20 (17.1)	0.0003**
Alive at 10 years (%)	7 (87.5)	7 (87.5)	16 (36.4)	4 (8)	34 (29.1)	0.00001**
Alive at 5 years (%)	8 (100)	10 (66.7)	20 (45.6)	7 (14)	45 (38.5)	0.0009***
Median DFS in months (range)	184.5 (85–232)	110 (12–201)	35 (8–208)	23.5 (4–196)	27 (4–208)	< 0.00001*
Disease-free at 15 years (%)	5 (62.5)	4 (26.7)	8 (18.2)	2	19 (16.2)	0.00001**
Disease-free at 10 years (%)	7 (87.5)	7 (46.7)	14 (31.8)	4 (8)	32 (27.4)	< 0.00001**
Disease-free at 5 years (%)	8 (100)	10 (66.7)	16 (36.4)	4 (8)	38 (32.5)	0.0002***

OS. Overall survival; DFS. Disease-free survival; * Kruskall-Wallis test; ** Chi-square test; *** Fisher’s exact test

At the end of study, the median follow-up for survivors was 15 years (181.5 months, interquartile range [IQR] 168–193). The median OS of the study cohort was 3 years (IQR 1–12, Fig. 1), the DFS was 2 years (IQR 1–12, Fig. 2).

Fig. 1.

Kaplan–Meier analysis of the overall survival. The shaded area indicates the 95% CI)

Fig. 2.

Kaplan–Meier analysis of the disease-free survival. The shaded area indicates the 95% CI)

After 15-years, there were 20 patients from this study group still alive (17.1%). The median age was 80, range 63- 93 years. One of these patients had a local recurrence in the neck after 20 months and was treated successfully with chemo-radiotherapy, thus representing a disease-free survival of 16.2% for the cohort. Two patients had non-related and treated cancers, one with basal cell cancer eyelid and one with prostate cancer.

After 10-years follow-up, there had been 34 survivors (29.1%), 32 of whom were disease-free (27.4%). Between 10 and 15 years, 13 of the 14 deaths were for causes unrelated to the primary cancer. Survival at 5-years was 45 alive (38.5%), with 38 (32.5%) disease- free. For those who were alive at 5 years, the probability of survival was 75% at 10 years, 53% at 15 years.

Disease recurrence was observed in 64 (54.7%) patients. Local disease occurred in 9 (7.7%), distant metastases in 46 (39.3%), and combined distant and local cancer deposits in another 9 (7.7%). Port-site recurrence was detected in 2 patients, one as isolated event, one in the context of widespread intrabdominal disease. There were five cases of recurrence at the anastomosis. The median survival after diagnosis of recurrence was 5.5 months.

Cancer-related death occurred in 63 (69.2%) subjects, while 15 (16.5%) died of other causes and in 13 (14.3%) cases, the cause of death was unknown. The median CSS was 32 months. Eighteen (15.4%) patients were diagnosed with cancer other than esophago-gastric (one had two different types of malignant disease); the incidence of a second primary cancer was 12.5% in Stage 0, 40% in Stage I, 18.2% in Stage II, 15.4% in Stage III (p = 0.0135) and overall, 6 (5.1%) succumbed to this.

Shorter OS and DFS estimates were observed in subjects with advanced disease stage (Figs. 3 and 4).

Fig. 3.

Kaplan–Meier analysis of the overall survival by disease stage

Fig. 4.

Kaplan–Meier analysis of the disease-free survival by disease stage

On Cox regression analysis, factors that predicted shorter OS were the formation of intracorporeal conduit (p = 0.035), blood transfusions (p = 0.015), with each unit of blood transfused increasing the hazard by 35%, cancer stage 2 (p = 0.038) and stage 3 (p = 0.017), disease recurrence (p < 0.001; Fig. 5).

Fig. 5.

Cox regression model of the association of patient factors to overall survival and Forest plot. BMI. Body Mass Index; Neoadj. Neoadjuvant; LRM. Longitudinal resection margin

Discussion

Survival after diagnosis of cancer is multi-factorial, and the most desired ultimate outcome. Advances in chemotherapy, medical optimisation, nutritional support and critical care techniques, have contributed to improved outcomes for esophageal and gastric cancer patients. [16, 17]

This study, with a follow up of 15-years is the first that reports the long-term survival after an exclusive group of patients that have undergone MIO. The survival rate was 17.1% and for those who were alive at 5 years, the probability of survival at 15 years was 53%. Our figures are consistent with recently published other longitudinal studies. Abou Chaar and colleagues reported a median survival rate at 15 years of 25.2% in 870 patients who underwent esophagectomy, the majority of which (96%) were open [18]. As with our series, the advanced tumour stage is associated with worse outcome and if patients can reach the 5-year point without disease-recurrence, then the likelihood is that most will then experience survival outcomes similar to the age and health matched general population. The intracorporeal formation of the gastric conduit was predictor of poor OS. We shifted to the extracorporeal fashioning technique, based on the evidence of reduced risk of conduit necrosis [19–22]. The negative impact of anastomotic complications after esophagectomy on patient’s survival had been demonstrated [23]; we hypothesize that the pre-ischaemic conditioning and the extracorporeal fashioning of the gastric conduit led to reduced anastomotic complications, thus enhancing patient’s OS. Similarly, in line with other published series [24], perioperative blood transfusions predicted poor OS. While it is accepted that the transfusion of more than 3 units of blood is a negative predictor [25], our analysis showed that each single unit increased the risk of not surviving by 35%. Blood transfusions are associated with the downregulation of the immune system, potentially leading to infectious diseases or secondary cancers [26]; nevertheless, it could also be hypothesized that the health conditions of subjects who required blood transfusions were poor and as such, were associated with reduced immune function and malnutrition. As expected, advanced disease stage was associated with shorter survival rates. Potential advances in early detection [27] and the increasing development of tailored chemotherapy [28] represent exciting advances in the management of esophageal cancer with the possibility of a greater proportion of patients with less than Stage III disease being available for surgery. Our results showed that 12.5% of survivors developed a secondary cancer. The onset of primary malignancies following esophagectomy for cancer had been described; in their long-term survival analysis, Kakuta et al. [29] reported 27.7% rate of secondary cancers. The authors suggested that the downregulation of the immune system secondary to surgery, blood transfusions, malnutrition and adjuvant treatment, could play a major role. However, we would argue that after 5 years from MIO, the risk of developing another cancer might potentially reflect that of the general population.

Providing peri-operative morbidity can be avoided or limited [30], MIO offers the real possibility of long-term survival.

Since the start of this study in 2004, there has been substantial progress in the acceptance of MIO techniques as an option for resection. Recent advances of Robotic Surgery have helped to overcome some of the technically challenging aspects of MIO [31, 32]. Limitations of the study are the retrospective design which could lead to the potential loss of follow-up data and the lack of long-term analysis of patients’ quality of life. However, we have previously reported a rapid restoration of health-related quality of life after MIO and proposed this to be an argument in favour of its use [33].

In conclusion, the long duration of follow-up of patients and consistency of the surgical procedure in this study provide evidence to support the continued integration of MIO as the standard of care for curative surgery in patients with esophageal cancer.

Author contributions

Conceptualization: S.W., D.D. Methodology: I.N, D.D, S.W. Data collection: A.O, G.C., D.D, D.V. Formal analysis and investigation: I.N., D.D. Writing—original draft preparation: A.O., G.C. Writing—review and editing: A.O., G.C., D.D. Supervision: A.M., D.V., S.W.

Data availability

The dataset analysed during the current study is not publicly available, in accordance to the Confidentiality Policy in force at the Royal Devon University Healthcare NHS Foundation Trust; this is a legal requirement under the Data Protection Act 2018 (United Kingdom). Data are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study had been registered and approved by the Royal Devon and Exeter NHS Foundation Trust Governance Board—reference number 24–1389. The Governance Board acted as an advisory panel not as an ethics committee. The study protocol was submitted to the Governance Board; given that the research was limited to secondary use of information previously collected in the course of normal care, and data were fully anonymised, the study did not fill the criteria to require review by the research ethics committee nor patients' consent form, as per the NHS Health Research Authority guidelines (https://www.hra-decisiontools.org.uk/ethics/docs/Algorithm%20-%20Does%20my%20project%20require%20REC%20review%20v2.0%2020200304.pdf). Hence, participants’ consent to the study was not obtained.

Data transparency

The dataset analysed during the current study is not publicly available, in accordance to the Confidentiality Policy in force at the Royal Devon University Healthcare NHS Foundation Trust; this is a legal requirement under the Data Protection Act 2018 (United Kingdom). Data are available from the corresponding author on reasonable request.

Competing interests

The authors declare no competing interests.