Risk and outcomes of metachronous gastric cancer following endoscopic resection of early gastric cancer: some answers, more questions

The last 20 years have seen significant reductions in the worldwide incidence of and mortality from gastric cancer, largely due to improvements in sanitation, greater refrigeration with concomitant reduction in the consumption of salted, smoked, and chemically-preserved foods, identification and eradication of Helicobacter pylori infection, and endoscopic screening [1].

Endoscopic submucosal dissection (ESD) has proven to be an alternative and effective management option to open gastrectomy for early gastric cancer (EGC), reducing surgical morbidity and mortality, and contributing to improved survival, with 5-year survival rates of up to 97 % [2] [3] [4]. Leaving the native stomach in place, however, raises the need to understand the risks and outcomes of metachronous gastric cancer (MGC) and to clarify to what extent post-ESD surveillance is required. The risk of MGC after endoscopic resection is 5.8 %, with a hazard ratio of 6.7 compared with gastrectomy, over a mean of 81 months [5]. Some of these issues are enlightened by Abe and colleagues [6] in this issue of Endoscopy.

In this retrospective analysis of 1526 consecutive patients from Japan who underwent curative ESD of EGC between 1999 and 2006, the investigators measured cumulative incidence, risk factors, and treatment outcomes of MGC as well as disease-specific survival at 5, 7, and 10 years. MGC was defined as a gastric cancer located in a different position from the index EGC more than 1 year after the initial ESD. A total of 238 patients (15.6 %) developed MGCs during a median follow-up of 82 months, of which 215 (90.3 %) underwent endoscopic resection (183 curatively), 14 had surgery, three had chemotherapy, and six were observed. Seven patients (0.3 %) died of their recurrence (most more than 5 years after their initial ESD), while 153 of the patients (10 %) died of other causes. The 5, 7, and 10-year cumulative incidence of MGC was 9.5 %, 13.1 %, and 22.7 %, respectively. Male sex and multiple initial EGCs were risk factors for MGC. Disease-specific survival was 99.2 % at 5 years, 98.6 % at 7 years, and 92.5 % at 10 years [6].

These findings support several observations for consideration. First, MGCs following ESD for EGC occur with a fairly high incidence that may increase more than linearly over time. Second, most MGCs may be “curatively” resected endoscopically. Third, death from MGCs is uncommon, with death due to other causes occurring 20 times as often. This observation seems to support the effectiveness of endoscopic surveillance, although the intensity of surveillance required and the degree of risk reduction from it are uncertain. Last, men with multifocal initial EGCs are at highest risk of MGC, suggesting the potential for risk stratification and tailoring of endoscopic surveillance.

How well do the data support these observations? What implications do they have for clinical practice and, in countries where gastric cancer remains highly prevalent, for guideline formation and health policy?

An understanding of the study’s strengths and limitations is needed to evaluate the validity of these observations. Although the study was retrospective, consecutive patients were identified, and despite two different sources of follow-up (the National Cancer Center Hospital and referring endoscopists), the degree of follow-up appears reasonably complete. While the “majority” of subjects underwent annual or biannual endoscopic surveillance, neither the proportion that received “any” surveillance nor how much surveillance was received by any subject is known. This limitation and the lack of a control group that did not undergo surveillance precludes knowing both the extent to which surveillance may reduce mortality from MGC and how much surveillance is needed for a particular degree of mortality reduction.

The investigators diligently examined several baseline features as possible risk factors for MGC and found no association with tumor location, size, macroscopic type, and histologic type. However, a second study limitation is the fact that the prevalence of H. pylori infection at index ESD was known for only 27 % of the cohort. While previous evidence suggests that the presence of H. pylori increases the risk of MGC and its eradication reduces this risk [7] [8], we cannot know whether H. pylori status at baseline and/or its eradication in the current cohort may have been useful for risk stratification.

Whether H. pylori status is associated with risk for MGC independently of male sex and initial EGC multifocality is unclear; these variables (and perhaps others such as age) require further testing in an independent cohort. The multisite prospective study of 10 000 patients undergoing ESD for EGC described in the discussion section should be very useful for this purpose, and may identify a strategy for tailoring surveillance endoscopy based on the risk of MGC and perhaps the number and severity of comorbid conditions.

Two related “big picture” issues require consideration to help understand the potential implications of this study. One issue is the potential for and degree of overdiagnosis, which occurs when a disease that is identified either incidentally or as a result of screening would otherwise not evolve to cause morbidity or mortality [9]. Overdiagnosis – and the potential for overtreatment and oversurveillance – may occur when screening for any disease, and requires careful consideration to both quantify its magnitude and to minimize its (negative) impact on patient outcomes.

Cancer overdiagnosis occurs when non-progressive or very slow-growing cancers are detected that would not otherwise manifest clinically or result in death. What is typically seen when examining population-based statistics is an increase in incidence or early stage disease, or both, along with no or lesser reductions in cancer mortality [9]. This phenomenon has been well-described for breast and prostate cancers, where screening has identified more (primarily early stage) disease, with either no or very modest reductions in cancer-specific mortality. Overdiagnosis and subsequent overtreatment with its unfavorable balance between benefits and harms requires consideration in any cancer screening and surveillance. Is overdiagnosis – or more precisely oversurveillance – occurring after ESD of EGC?

Gastric cancer meets the two prerequisites for overdiagnosis: a silent reservoir of disease and a way to detect it (particularly through screening) [10]. However, both incidence and mortality have fallen concurrently and to similar extents, making overdiagnosis of the index EGC less likely. Our lack of understanding of the dwell time for EGC precludes knowing whether at least some lesions are “nonprogressive cancers,” increasing the potential for overdiagnosis, an issue of particular relevance for surveillance endoscopy.

What about the potential for and degree of endoscopic oversurveillance for MGC? While the large difference in magnitude between MGC incidence and mortality makes overdiagnosis possible, the difference could also represent the effectiveness of surveillance. In the Abe study, the “number needed to treat” of surveillance endoscopy to diagnose one patient with MGC was only 6.4; the “number needed to scope” to prevent one death from MGC cannot be determined because of the absence of a control group. Furthermore, the true endoscopic effort required in the study is not clear because the surveillance frequency was not specified.

Related to overdiagnosis/oversurveillance, the second issue is our understanding of whether and to what extent surveillance for MGC adds to the quality and quantity of life. Initial ESD likely adds to both by obviating the need for open gastrectomy, but what about subsequent endoscopic surveillance? And which statistical metrics are best for assessing a survival benefit?

For cancer screening, the most favored statistic is cancer-specific mortality [11]. All-cause mortality may be an important secondary statistic, particularly if treatment for the condition has risks and adverse effects that could decrease survival. In general, survival statistics, including 5-year survival and survival duration, are subject to lead-time bias, where “survival” is prolonged because of early (preclinical) detection. Therefore, cancer screening may result in prolonged survival, but have no effect on cancer-specific mortality. While screening for an index gastric cancer could result in prolonged survival – due either to lead-time bias or to real improvements in early diagnosis and treatment – surveillance for MGC would appear to be less subject to lead-time bias since a “zero-time” for all subjects is identifiable, specifically the date of endoscopic cure of the index gastric cancer. In this case, cancer-specific survival and survival trends over time are valid measures.

In the study by Abe and colleagues, cancer-specific survival was high. While the use of survival statistics may be valid, the reason for the high survival rate is not clear, as it could be due either to the effectiveness of endoscopic surveillance or to overdiagnosis of MGCs that would not ever have manifested clinically. While the high MGC incidence and low MGC mortality are suggestive of overdiagnosis, understanding the extent to which endoscopic surveillance results in overdiagnosis versus prolongs life requires either a randomized trial or a carefully controlled cohort study in which some persons do not receive surveillance.

In conclusion, the study by Abe and colleagues shows a high endoscopic incidence of MGC following ESD of EGC, but a high rate of survival in the setting of endoscopic surveillance of some degree. Further study is needed to understand how absence and/or eradication of H. pylori affects the risk of MGC, whether clinically meaningful risk stratification is possible, and how much endoscopic surveillance is enough to achieve high disease-specific survival.

• References

• 1 Fock KM. Review article; the epidemiology and prevention of gastric cancer. Aliment Pharmacol Ther 2014; 40: 250-260
• 2 Choi MK, Kim GH, Park doY et al. Long-term outcomes of endoscopic mucosal dissection for early gastric cancer: a single-center experience. Surg Endosc 2013; 27: 4250-4258
• 3 Oka S, Tanaka S, Kaneko I et al. Advantage of endoscopic mucosal dissection compared with EMR for early gastric cancer. Gastrointest Endosc 2006; 64: 877-883
• 4 Chung IK, Lee JH, Lee SH et al. Therapeutic outcomes in 1000 cases of endoscopic mucosal dissection for early gastric neoplasms: Korean ESD study Group multicenter study. Gastrointest Endosc 2009; 69: 1228-1235
• 5 Choi KS, Jung HY, Choi KD et al. EMR versus gastrectomy for intramucosal gastric cancer: comparison of long-term outcomes. Gastrointest Endosc 2011; 73: 942-948
• 6 Abe S, Oda I, Suzuki H et al. Long-term surveillance and treatment outcomes of metachronous gastric cancer occurring after curative endoscopic submucosal dissection. Endoscopy 2015; 47: 1113-1118
• 7 Pasechnikov V, Chukov S, Federov E et al. Gastric cancer: prevention screening and early diagnosis. World J Gastroenterol 2014; 20: 13842-13862
• 8 Fukase K, Kato M, Kikuchi S et al. Effect of eradication of Helicobacter pylori on incidence of metachronous gastric carcinoma after endoscopic resection of early gastric cancer: an open-label, randomized controlled trial. Lancet 2008; 372: 392-397
• 9 Esserman LJ, Thompson Jr IM, Reid B. Overdiagnosis and overtreatment of cancer: an opportunity for improvement. JAMA 2013; 310: 797-798
• 10 Welch HG, Black WC. Overdiagnosis in cancer. J Natl Cancer Inst 2010; 102: 605-613
• 11 Dickman PW, Adami H-O. Interpreting trends in patient cancer survival. J Intern Med 2006; 260: 103-117