Fax: +46-8-33 15 87
Improved survival in both histologic types of oesophageal cancer in Sweden
Article first published online: 18 APR 2002
Copyright © 2002 Wiley-Liss, Inc.
International Journal of Cancer
Volume 99, Issue 5, pages 751–754, 10 June 2002
How to Cite
Sundelöf, M., Ye, W., Dickman, P. W. and Lagergren, J. (2002), Improved survival in both histologic types of oesophageal cancer in Sweden. Int. J. Cancer, 99: 751–754. doi: 10.1002/ijc.10420
- Issue published online: 14 MAY 2002
- Article first published online: 18 APR 2002
- Manuscript Accepted: 28 FEB 2002
- Manuscript Revised: 27 FEB 2002
- Manuscript Received: 15 NOV 2001
- Swedish Cancer Society
- oesophageal cancer;
The prognosis among patients diagnosed with oesophageal cancer is poor with an overall 5-year survival close to 5% in most countries. Improved diagnostic and surgical strategies might influence the survival, however. We investigated the observed and relative survival among all patients in Sweden diagnosed with oesophageal adenocarcinoma (n = 1,441) or squamous cell carcinoma (n = 6395) from 1961–1996 with follow-up to December 1997. Observed survival rates were calculated by the life-table method. Relative survival rates were computed as the ratio of the observed to the expected survival. The expected survival was inferred from the survival among the entire Swedish population in the same age, sex and calendar year strata. The 5-year observed survival rate for adenocarcinoma increased from a stable figure close to 4% during the entire period 1961–1989 to 10.5% during 1990–1996. Similarly, the 5-year relative survival rate was stable around 5% during 1961–1989, but during 1990–1996 the survival was increased to 13.7%. For squamous cell carcinoma, the survival improved slightly by each decade, starting with 3.8% 5-year observed survival in 1961–1969 to 7.0% during 1990–1996. Similarly, the 5-year relative survival improved from 5.0% to 8.9% during the study period. In conclusion, the survival rates for both oesophageal adenocarcinoma and squamous cell carcinoma have increased significantly during the 1990s compared to those in the previous 3 decades (p < 0.001). © 2002 Wiley-Liss, Inc.
Oesophageal cancer is a highly lethal malignancy. The previously reported 5-year survival rate of close to 5% for both histologic types, i.e., adenocarcinoma and squamous cell carcinoma, is worse than that of most other malignancies.1 Surgical resection is the only established curative treatment for oesophageal cancer. Several randomised clinical trials trying adjuvant therapy have been conducted, but the results have generally been disappointing. There is currently no evidence of an increased long-term survival as a result of the use of adjuvant chemotherapy or radiotherapy.2, 3 In recent decades there has been a rapid increase in the incidence of oesophageal adenocarcinoma in Western countries, whereas the incidence of squamous cell carcinoma of the oesophagus has been stable or decreasing.4, 5 Most oesophageal adenocarcinomas develop from Barrett′s oesophagus, a columnar cell metaplasia of the native distal oesophageal squamous cell mucosa, easily detectable at endoscopy and verified by histology.6 The strong association with oesophageal adenocarcinoma has prompted calls for introduction of endoscopic surveillance programs for early detection of adenocarcinoma among persons with Barrett′s oesophagus,7 although cost-effectiveness should be taken into account.8 Improved survival of oesophageal adenocarcinoma as a result of surveillance programs has been reported in clinical settings.9 Improvements in diagnostic procedures and improvements in surgical techniques might have influenced the long-term survival among patients with oesophageal cancer. In our study, we investigate whether the survival rates in oesophageal adenocarcinoma or squamous cell carcinoma have changed during recent decades in Sweden.
MATERIAL AND METHODS
The Swedish Cancer Register was established in 1958 and since then physicians throughout Sweden are obliged to report all cancer cases to the Cancer Register. Additionally, pathologists and cytologists report all cancer diagnoses based on biopsies, cytologic specimens, surgically resected tissues and autopsies. Therefore, most cases have been doubly verified. The Swedish Cancer Register has been estimated to be 98% complete.10 The percentage of oesophageal cancers that were histologically verified in the Register exceeded 90% for each year under study and was close to 100% during the last 20 years.11–13 The records in the Register are deemed to be incomplete during the initial period in 1958–1960, however. We therefore limited our analysis to start from 1961. The Cancer Register used the 7th version of International Classification of Diseases (ICD-7) for coding diagnoses during the entire study period. Data available in the Cancer Register include the National Registration Numbers, a 10-digit unique personal identifier assigned to all Swedish residents, gender, date of diagnosis, age at diagnosis, basis of diagnosis, index for cancers found only in autopsy, index for first or secondary cancers and pathology. For our study, oesophageal cancer (ICD-7 code 150) was further divided by histology into adenocarcinoma (PAD = 096) and squamous cell carcinoma (PAD = 146). Cases found incidentally at autopsy, or with other or unknown histologic type of their oesophageal tumour, were excluded from the analyses. To avoid influence of the survival rates due to the occurrence of other cancers, we only included primary oesophageal cancers, i.e., patients without any previous cancer diagnoses recorded in the Cancer Register. The vital status of the cases was determined by linkage to the nationwide Swedish Death Register and any emigration of the cohort members was assessed by linkage to the nationwide Swedish Emigration Register. We also linked the cohort to the Total Population Register, which annually registers all residents living in Sweden. The National Registration Numbers were used to ensure correct matching in all record linkages. All registers used are continuously updated. A record, which could not be found in any of the above-mentioned registers, was deemed to be an invalid National Registration Number and was thus deleted. Oesophageal cancer patients were followed up until death, emigration or the end of observation (December 31, 1997), whichever occurred first.
Observed and relative survival rates were analysed per 5-year age groups in 4 calendar periods of diagnosis according to histologic type of oesophageal cancer. The 4 calendar periods were: 1961–1969, 1970–1979, 1980–1989 and 1990–1996. Observed survival rates were estimated using the life-table method.14 Relative survival rates were computed as the ratio of the observed to the expected survival, the latter inferred from the survival among the entire Swedish population in the same age, sex and calendar year strata. Tests of the equality of survival between patients diagnosed in different periods were performed using the maximum likelihood tests described by Hakulinen et al.15
To investigate independent effects of patient characteristics along with calendar periods on survival, we also fitted relative survival models as proposed by Hakulinen and Tenkanen.16 Model parameters were estimated by the SAS GENMOD procedure17 by using grouped life-table data. Variables included in the model were sex, age at diagnosis in 4 groups (<60, 60–69, 70–69 and 70+) and the calendar year at diagnosis in 4 groups (1961–1969, 1970–1979, 1980–1989 and 1990–1996). The results were expressed as relative hazards of dying of adenocarcinoma or squamous cell carcinoma of the oesophagus with 95% confidence intervals (CI). Standard errors of the estimates were adjusted for overdispersion.
We identified a total of 10,820 records with oesophageal cancer through the Swedish Cancer Register during the period 1961–1996. We excluded 480 cases found first at autopsy, 918 secondary oesophageal cancer cases and 70 invalid records. Among the remaining 9,352 patients with oesophageal cancers for final analyses, 1,441 and 6,395 patients were diagnosed with adenocarcinoma and squamous cell carcinoma, respectively. There was a strong male predominance among patients with adenocarcinoma, 80% being males, whereas among patients with squamous cell carcinoma, 69% were males. The median age at diagnosis among adenocarcinoma patients increased from 67.5 in 1961–1969 to 70.0 in 1990–1996, whereas the median age among patients with squamous cell carcinoma was stable and close to 69 years during the study period (Table I).
|Histologic type of oesophageal cancer||Period of diagnosis||No. of patients||Mean age at diagnosis||Observed survival rate (%) with 95% confidence interval||Relative survival rate (%) with 95% confidence interval|
|1 year||3 year||5 year||1 year||3 year||5 year|
|Adeonocarcinoma||1961–1969||214||67.5||29.9 (26.8–33.0)||6.5 (2.6–10.2)||5.1 (3.6–6.7)||31.2 (24.7–37.7)||7.5 (3.6–11.4)||6.5 (2.7–10.3)|
|1970–1979||331||68.9||24.2 (21.8–26.5)||5.4 (2.5–8.3)||2.7 (1.8–3.7)||25.4 (20.5–30.3)||6.4 (3.5–9.3))||3.6 (1.3–5.9)|
|1980–1989||455||69.4||24.6 (22.6–26.6)||6.0 (3.4–8.6)||4.5 (3.5–5.4)||25.9 (21.0–30.8)||7.0 (4.4–9.6)||5.8 (3.3–8.3)|
|1990–1996||441||70.0||31.8 (29.5–34.0)||14.2 (10.1–18.3)||10.5 (8.7–12.2)||33.3 (28.7–38.0)||16.5 (12.4–20.6)||13.7 (9.1–18.3)|
|Squamous cell carcinoma||1961–1969||1,383||69.0||23.6 (22.8–24.4)||6.2 (5.6–7.7)||3.8 (3.3–4.4)||24.8 (22.4–27.3)||7.2 (6.7–8.7)||5.0 (3.7–6.4)|
|1970–1979||1,749||68.9||23.5 (22.5–24.5)||7.2 (5.8–8.6)||5.0 (4.5–5.6)||24.6 (22.5–26.7)||8.3 (6.9–9.7)||6.5 (5.2–7.8)|
|1980–1989||1,929||69.5||25.5 (24.5–26.5)||8.6 (7.1–10.1)||6.2 (5.6–6.7)||26.7 (24.6–28.8)||9.8 (8.3–11.3)||7.9 (6.5–9.3)|
|1990–1996||1,481||69.6||30.6 (29.4–31.9)||10.3 (8.3–12.3)||7.0 (6.2–7.8)||31.9 (29.3–34.6)||11.8 (9.8–13.8)||8.9 (6.8–10.9)|
Survival in adenocarcinoma of the oesophagus
Among patients with adenocarcinoma of the oesophagus, both the observed and the relative survival increased during the recent decade compared to all previous decades studied (Table I). The 1-year relative survival was 31.2% during 1961–1969, decreased to about 25–26% during 1970–1989 and increased to 33.3% during 1991–1996. The 5-year observed survival was lowest during 1970–1979 (2.7%) and increased to 10.5% in 1990–1996. The 5-year relative survival was stable around 5% during the first 3 decades and increased to 13.7% during the most recent study period 1990–1996 (Table I and Fig. 1). The improvement in survival during the entire follow-up duration for patients diagnosed during 1990–1996 was statistically significant (p < 0.001) compared to that during 1961–1989. Generally, the 3-year survival rates did not differ much from the 5-year survival rates (Table I). A multiple regression analysis showed that there was no significant difference in the 5-year relative survival between sexes. Patients diagnosed at ages older than 80 years had a significantly increased mortality compared to those diagnosed at ages younger than 60 (relative hazard = 1.82, 95% CI 1.33–2.48). There was a 19% deficit in relative hazard of dying of oesophageal adenocarcinoma in the first 5-year period after diagnosis among those diagnosed during the 1990s compared to those diagnosed in the 1960s after controlling for effects of sex and age at diagnosis (Table II).
|Adenocarcinoma||Squamous cell carcinoma|
|Relative hazard||95% CI||Relative hazard||95% CI|
|Period of diagnosis|
|Age at diagnosis|
Survival in squamous cell carcinoma of the oesophagus
Among patients with a diagnosis of oesophageal squamous cell carcinoma, the long-term survival improved slightly and gradually for each decade during the study period (Table I and Fig. 1). The 5-year observed survival increased from 3.8% to 7.0% and the 5-year relative survival increased from 5.0% to 8.9% during the 4 decades of observation. The comparison of the 5-year survival between the time periods 1990–1996 and 1961–1989 revealed a statistically significant improvement (p < 0.001). The 3-year survival rates did not differ importantly from the 5-year rates (Table I). Multiple regression analysis showed that in the first 5-year follow-up period after diagnosis of oesophageal squamous cell carcinoma, women had a significantly decreased mortality compared to men. Patients who were diagnosed at ages older than 80 years had a significantly increased mortality compared to those diagnosed at ages younger than 60 (relative hazard = 1.59, 95% CI 1.20–2.11). After controlling for effects of gender and age at diagnosis, relative hazard decreased with the more recent calendar period (Table II).
Our study reveals a statistically significant improvement in the long-term survival of both histologic types of oesophageal cancer in Sweden during the recent decade.
Strengths of our study include the nationwide and population-based design, the length of the observation, the completeness of follow-up and the precision due to the large number of cases. Weaknesses include the lack of data of the tumour stage. Therefore, the stage-specific survival was not possible to analyse. Furthermore, we did not have data of the given treatment. Hence, we could not make a distinction between patients who were treated with a palliative intention and those who were treated with a radical oesophageal resection. To avoid influence of the survival rates due to the occurrence of other cancers, we only included patients without any previous cancer diagnoses recorded in the Cancer Register. Since the Cancer Register started in 1958, we had no data concerning previous cancers before this year and we might therefore have included some patients with a previous cancer in our study. This source of error should have affected patients during the early 1960s more than later diagnosed cases. Since we could identify all cancers at least 3 years before the start of the inclusion, any increased mortality due to a previous cancer is likely to have been of minor importance, however.
We report the observed survival as well as the relative survival in our study. Although the observed survival is of interest in clinical practise, in any comparison between different time periods it is more accurate to study the relative survival rates, since it takes into account the generally increasing survival noted in the Swedish population. The small difference between the relative survival at 3 years and 5 years after the diagnosis indicates that most cancer recurrences in oesophageal cancer occur shortly after the treatment has been completed and it underlines the aggressiveness of this tumour. Patients who survive 3 years after the diagnosis was first confirmed are likely to have been cured from their oesophageal cancer.
A population-based study on the prognosis of oesophageal cancer patients in Sweden during 1961–1987 did not show any clear improvement of the survival rates.18 A slightly improved survival among patients with a diagnosis of oesophageal adenocarcinoma who were diagnosed in the 1980s compared to previous decades has been observed in the United States, however.19 We cannot give an answer to which factor(s) is responsible for the improved survival, but we can only speculate. Previous studies indicate that an earlier discovery of the oesophageal cancer is an important predictor of long-term survival.9 The general introduction of endoscopy as a diagnostic tool during the 1980s could be one explanation of earlier tumour detection due to the improved diagnostic accuracy. Such improvement might partly explain the improved survival that we found, but such effect should have been obvious for adenocarcinomas already during the 1980s, which was not found in our study. The introduction of endoscopic surveillance programs for Barrett′s oesophagus might explain the more pronounced improvement in survival among patients with adenocarcinoma compared to patients with squamous cell carcinoma during the 1990s. Furthermore, an improved surgical treatment with reduced operative mortality and more radical resection margins might contribute to the benefits in survival seen for both histologic types of oesophageal cancer during recent years. Adjuvant chemotherapy or radiotherapy might have influenced the results slightly. A very limited number of hospitals in Sweden used such therapy during the study period, however. Another hypothesis is that an improved awareness of early disease symptoms in the general population might have contributed to our results.
In conclusion, our nationwide survival study during a period of nearly 4 decades reveals that the prognosis of oesophageal cancer is poor, but during the recent decade the prognosis has significantly improved for both adenocarcinoma and squamous cell carcinoma. The reasons for the better survival rates are unknown. Further population-based studies of the survival in oesophageal cancer, including detailed data of the treatment and the tumour stage, are warranted. Such studies might help identify therapeutic strategies that could further improve the chances of cure among patients with oesophageal cancer.
- 1IARC. Survival of cancer patients in Europe: the EUROCARE II study. Lyon: International Agency for Research on Cancer, 1999.
- 10Completeness of the Swedish Cancer Registry. Non-notified cancer cases recorded on death certificates in 1978. Acta Radiol 1984;23: 305–13., .
- 11National Board of Health and Welfare. Cancer incidence in Sweden 1975. Stockholm: Centre for Epidemiology, National Board of Health and Welfare, 1979.
- 12National Board of Health and Welfare. Cancer incidence in Sweden 1985. Stockholm: Centre for Epidemiology, National Board of Health and Welfare, 1988.
- 13National Board of Health and Welfare. Cancer incidence in Sweden 1995. Stockholm: Centre for Epidemiology, National Board of Health and Welfare, 1997.
- 17SAS Inc. S.I., SAS/STAT software: changes and enhancements through release 6.12. Cary, NC: SAS Inc., 1997.