Should quality of life influence the decision for esophagectomy in high grade dysplasia?
D. Provenzale (Durham)
There is an increased risk of esophageal adenocarcinoma in patients with Barrett's esophagus. Patients with Barrett's esophagus develop adenocarcinoma of the esophagus at a rate of 1/75 patient-years, an incidence that is 30-125 times that of the general population [1-7]. Patients with high grade dysplasia are at greatest risk.
Our previously published decision model suggested that surveillance of patients with Barrett's esophagus is effective; increasing life expectancy and preventing cancer deaths .
The objective of this study was to examine the effectiveness of the alternative management strategies for Barrett's patients diagnosed with high grade dysplasia, and to determine their impact on quality of life.
We modified our previously published decision model  to compare the alternative indications for esophagectomy in patients diagnosed with high grade dysplasia.
We used a Markov model  and constructed a computer cohort simulation of 10,000, hypothetical men with Barrett's esophagus. A Markov model is a mathematical model for estimating life expectancy that can be used for determining prognosis in medical applications. It is particularly useful for simulating clinical situations that involve an ongoing risk over time (such as the risk for developing adenocarcinoma in patients with Barrett's esophagus), because it considers the consequences of ongoing risk, namely, that the times that events will occur are uncertain, and that an event may occur more than once . For example, certain members of the hypothetical cohort with high grade dysplasia might develop cancer within a year, while others might not develop cancer for several years.
Structure of the analysis
We evaluated the following strategies:
- A: immediate esophagectomy (for those who are surgical candidates);
- B1-5: surveillance every 3, 6, 12, 18 and 24 months with esophagectomy for cancer;
- C: no surveillance.
Endoscopy is performed for dysphagia, and esophagectomy is performed if cancer is diagnosed.
Our simulation begins at the point at which an endoscopic biopsy demonstrates Barrett's esophagus and high grade dysplasia. Over time, patients with high grade dysplasia may remain in this health state, they may develop cancer or they may die. Our model also includes long-term post-surgical states for the prognosis of patients following surgery, an inoperable state for patients with cancer who are inoperable due to other illnesses , and a dead state. The likelihood of entering each of these states, each year, depends on annual transition probabilities derived from published literature [5, 6, 12] Because the current therapy for patients with high grade dysplasia is controversial, patients detected to have high grade dysplasia in strategy A undergo esophagectomy, while those in strategies B1-B5 undergo endoscopic surveillance with biopsy every three months and undergo esophagectomy only if carcinoma is diagnosed.
The analysis continues until all members of the cohort have died, either from esophageal cancer or from other causes, based on age, gender and race-specific mortality rates taken from life tables for the general population . The simulation permits calculation of the average life expectancy, and the cumulative incidence of cancer.
All parameters used in the model are shown in Table I. Estimates were obtained from published reports. When published data were not available, we consulted experts in the field to obtain estimates [11, 13-20].
Table II shows the baseline results using the values in Table I.
Cumulative incidence of cancer
Our model predicts that 96% of patients with high grade dysplasia who do not undergo surveillance will develop cancer over their lifetime. Endoscopic surveillance and esophagectomy for cancer decreases the lifetime cumulative incidence to 49-81% (a 16-49% reduction). The shortest surveillance interval results in the greatest reduction in cancer incidence.
For reference, 55 year old men from the general population have an average life expectancy of 24.5 years. For patients with high grade dysplasia who do not undergo surveillance, our model predicts an average life expectancy of 5.3 years. Surveillance with esophagectomy for cancer adds 5.4-9.7 years to life expectancy compared to no surveillance. The optimal surveillance interval is every 3 months. For those who are surgical candidates, immediate esophagectomy for high grade dysplasia extends life expectancy by an additional 5.5-9.8 years.
Quality of life
These life expectancy gains, however, are associated with procedures and symptoms which some patients might prefer to avoid. Endoscopic procedures may be inconvenient and uncomfortable, and esophagectomy may be associated with life long symptoms. In order to determine the impact of quality of life after esophagectomy on the decision for immediate esophagectomy versus continued surveillance, we performed sensitivity analysis. In sensitivity analysis, the quality of life with an esophagectomy is varied over a broad range to determine its impact on the preferred strategy. In our analysis we varied quality of life from 0.00 (equivalent to being dead) to 1.00 (equivalent to a state of perfect health). The results are shown in Figure 1 and suggest that if the quality of life with an esophagectomy is greater than 0.28, immediate esophagectomy is the preferred strategy for high grade dysplasia patients who are surgical candidates, providing an increase in quality-adjusted life expectancy of up to 15.2 years compared to no surveillance and of up to 9.8 years compared to surveillance with esophagectomy for cancer. If the quality of life is between 0.01 and 0.28, then continued surveillance is preferred providing an increase in quality-adjusted life expectancy of approximately 3.9 years compared to immediate esophagectomy and of up to 1.5 years compared to no surveillance. If the quality of life with an esophagectomy is less than 0.01, then no surveillance is preferred.
In order to evaluate the quality of life of Barrett's patients who undergo esophagectomy, we performed a pilot study of patients with a history of Barrett's esophagus who had undergone esophagectomy for high grade dysplasia or cancer at Duke University and the Durham Veterans Affairs Medical Center . We used the time trade-off technique  to measure quality of life. The time trade-off is a form of utility assessment, in which a set of dimensions is described in a scenario, and a score is assigned to reflect the overall assessment. Health state utilities are numeric values assigned to a particular health state. These values reflect the quality of the health states and allow morbidity and mortality estimates to be combined into a single weighted measure, a "QALY" or quality-adjusted life year gained . Preference values are derived implicitly, based on individual responses to decision situations, e.g.: "Would you rather live 10 years with an esophagectomy or 5 years in excellent health?" The time trade-off is a valid, reliable measure of quality of life that has been administered to diverse populations including post-colectomy patients with a history of ulcerative colitis , and patients on long-term home TPN . Measured values are reported as reliable and stable . Our results suggested that the quality of life post-esophagectomy is excellent. The mean (range) for quality of life was 0.9 (0.6, 1.0). We incorporated this value for quality of life into our decision model. Using a quality-adjustment for esophagectomy of 0.9, the analysis suggests that for those who are surgical candidates, immediate esophagectomy will increase quality-adjusted life expectancy by up to 13.3 years compared to no surveillance and by up to 8.5 years compared to surveillance strategies with esophagectomy performed for the development of cancer. For those who choose continued surveillance as a management strategy, surveillance every 3 months would increase quality-adjusted life expectancy, on average, by approximately 8.5 years compared to no surveillance.
Figure 1. Preferred strategy based on quality of life.
Quality of life is a critical component in the decision for immediate esophagectomy or continued surveillance among Barrett's patients with high grade dysplasia. Our results suggest that either immediate esophagectomy (for surgical candidates), or frequent surveillance (every 3 months), should increase both length and quality of life. Our decision analysis suggests that the benefits of immediate esophagectomy in preventing cancer outweigh the morbidity and chronic symptoms associated with esophagectomy for most values of quality of life. Although the gains are not as great, surveillance will also increase both length and quality of life.
Our study of quality of life post esophagectomy was the first to use the time trade-off in this group. The results suggest that quality of life is excellent post-esophagectomy. To provide a prospective for these results, the quality of life of patients who have been treated for benign esophageal disease including achalasia and gastroesophageal reflux ranges from 0.74-0.96, and the quality of life of patients who have undergone colectomy for ulcerative colitis is 0.96 (mean). Thus, the quality of life post-esophagectomy compares favorably with other post-surgical groups.
Our results provide a guide for decision making with patients with high grade dysplasia. Clinicians and patients can use these results to determine the expected outcomes with alternative management strategies. Those who make health policy can use these results to develop programs that optimize both length and quality of life of Barrett's patients with high grade dysplasia.
15. Ellis FH, Gibb SP, Watkins E. Esophagogastrectomy - A safe, widely applicable and expeditious form of palliation for patients with carcinoma of the esophagus and cardia. Ann Surg 1983;198:531-540.
23. Provenzale D, Shearin M, Tillinger W, Phillips B, Bollinger R, Koruda M, Drossman D, Wong JB. The effect of patient preferences on policy recommendations for prophylactic colectomy or surveillance in patients with ulcerative colitis. Gastroenterology 1994;106(4)A:22.
26. Guanrei Y, Songliang Q, He H, et al. Natural history of early esophageal squamous carcinoma and early adenocarcinoma of the gastric cardia in the People's Republic of China. Endoscopy 1988;20:95-98.