SEARCH

SEARCH BY CITATION

An analysis of temporal trends for esophageal cancer patients diagnosed in 1961–1996 in Sweden showed improvement in relative survival rates (RSRs), but data were not available on stage at diagnosis or on cancer-directed treatment.1 The present report considers temporal changes in RSRs by histology for esophageal cancer patients in the United States. The National Cancer Institute's Surveillance, Epidemiology and End Results (SEER) program of high-quality cancer registries, established in 1973 and regarded as generally representative of the entire United States.2 Data were obtained for the 9 SEER areas (Connecticut; Detroit, MI; Hawaii; Iowa; New Mexico; San Francisco–Oakland, CA; Seattle, WA; Utah; and, since 1975, Atlanta, GA). Patients with esophageal cancer as the first or only reportable cancer were identified, excluding those ascertained only by death certificate or autopsy.

Using a computer program that includes mortality rates for the general U.S. population (by age, sex and race),3 RSRs at 1, 3 and 5 years after diagnosis in 1975–1979, 1980–1984, 1985–1989, 1990–1994 and 1995–1998 were obtained for squamous cell carcinoma and adenocarcinoma of the esophagus (defined by ICD-O-2 morphology codes).4 Confidence limits on RSRs were estimated as ± twice the SE (p = 0.05).3 RSRs by histologic group are not routinely included in SEER reports,2 and a SEER monograph on histology included data for cancers diagnosed only through 1986.4, 5 Another SEER study was limited to esophageal adenocarcinoma, showing slight increases in RSRs in the 1980s vs. earlier years.6

RSRs increased over time for both squamous cell and adenocarcinoma (Table I), beginning earlier than in Sweden.1 The overall increases from 1975–1979 to 1995–1998 were similar for the 2 histologic groups (e.g., ratios of 3-year RSRs were 2.2 for squamous cell carcinoma and 2.1 for adenocarcinoma), but the larger sample sizes for squamous cell carcinoma in the earlier years resulted in statistical significance for more comparisons than for adenocarcinoma. Age-standardized incidence rates for adenocarcinoma of the esophagus in the SEER program have surpassed the declining rates for squamous cell carcinoma.7 For squamous cell carcinoma patients, RSRs were statistically significantly higher in 1980–1984 vs. 1975–1979 and in 1985–1989 vs. 1980–1984 (except for the 5-year RSR); thereafter, increases were smaller (Table I). RSRs were slightly higher for adenocarcinoma than for squamous cell carcinoma, but differences were not statistically significant (except for 1-year RSRs in 1995–1998).

Table I. RSRs for Esophageal Cancer by Histologic Group, SEER Program (9 Registries)
Years of diagnosisNumberRSR (%) (95% confidence interval)
1 year3 years5 years
  • 1

    Follow-up (i.e., through the end of 1999) was too limited.

  • *

    p < 0.05 for comparison with RSR in 1975–1979 (see text).

Squamous cell carcinoma    
 1975–19792,46328.0 (26.2–29.9)7.4 (6.3–8.5)4.5 (3.6–5.5)
 1980–19842,51132.5 (31.1–34.4)*10.1 (8.8–11.4)*7.2 (6.1–8.4)*
 1985–19892,48837.2 (35.2–39.2)*13.6 (12.1–15.0)*9.1 (7.8–10.4)*
 1990–19942,27838.8 (36.7–40.9)*16.0 (14.3–17.6)*11.8 (10.3–13.4)*
 1995–19981,58340.6 (38.0–43.3)*16.3 (14.1–18.5)*1
Adenocarcinoma    
 1975–197940730.5 (25.8–35.1)9.6 (6.5–12.8)5.7 (3.1–8.2)
 1980–198456834.5 (30.4–38.6)12.2 (9.2–15.1)8.2 (5.7–10.8)
 1985–19891,02738.1 (35.0–41.2)13.2 (11.0–15.5)10.2 (8.1–12.3)
 1990–19941,53943.3 (40.7–45.9)*18.3 (16.2–20.5)*13.6 (11.6–15.6)*
 1995–19981,65345.6 (43.1–48.1)*20.4 (18.0–22.8)*1

Increases in RSRs were evident for both males and females. For example, 3-year RSRs increased from 1975–1979 to 1995–1998 from 6.4% to 15.2% in males and from 9.8% to 18.5% for females for squamous cell carcinoma and from 10.3% to 20.7% in males and from 6.6% to 18.6% in females for adenocarcinoma (data not shown).

Increases in RSRs could reflect improvements in detection resulting from the use of endoscopy (starting in the 1980s).1 Data on SEER historical summary stage (local, regional, distant or unknown) were available for all years of diagnosis. Using the data in Tables I and II, the proportion of esophageal cancers diagnosed at an early (localized) stage decreased slightly for squamous cell carcinoma (33.0% in 1975–1979 to 28.1% in 1995–1998) but increased slightly for adenocarcinoma (19.9% to 25.3%). However, increases in RSRs were evident within the local-stage group for both squamous cell carcinoma and adenocarcinoma (Table II). Differences in RSRs between squamous cell and adenocarcinoma patients were larger within local stage (Table II) than for all stages combined (Table I). The survival difference by histology may have been influenced by the detection of superficially invasive adenocarcinomas during surveillance of Barrett's esophagus patients,5, 8 but the differences in RSRs were evident even in 1975–1979 (i.e., prior to the spread of endoscopy).

Table II. RSRs for Local-Stage Carcinoma of the Esophagus in U.S. SEER Program (9 Registries)
Years of diagnosisNumberRSR (%) (95% confidence interval)
1 year3 years5 years
  • 1

    Follow-up (i.e., through the end of 1999) was too limited.

  • *

    p < 0.05 for comparison with RSR in 1975–1979 (see text).

Squamous cell carcinoma    
 1975–197982737.8 (34.3–41.3)11.4 (9.0–13.7)7.8 (5.7–9.9)
 1980–198474846.1 (42.3–49.9)*18.6 (15.5–21.6)*14.2 (11.3–17.0)*
 1985–198967955.2 (51.2–59.2)25.8 (22.2–29.4)*16.8 (13.5–20.0)*
 1990–199463853.4 (49.4–57.7)*27.4 (23.5–31.2)*20.9 (17.2–24.5)*
 1995–199844557.6 (52.7–62.5)*25.9 (21.0–30.8)*1
Adenocarcinoma    
 1975–19798151.2 (39.7–62.7)26.2 (15.7–36.8)15.0 (6.1–23.9)
 1980–198410753.4 (43.4–63.4)28.4 (19.0–37.9)25.2 (15.6–34.8)
 1985–198919964.6 (57.4–71.8)31.7 (24.4–39.0)28.7 (21.2–36.2)
 1990–199436766.7 (61.4–71.9)43.0 (37.3–48.8)*35.0 (29.2–40.9)*
 1995–199841969.5 (64.7–74.4)*44.2 (38.1–50.2)*1

Improvements in treatment of local- and regional-stage cancers could have resulted in temporal increases in RSRs. However, RSRs increased over time among local- and regional-stage cancer patients who had neither cancer-directed surgery nor radiotherapy coded in the SEER database, though little or no improvement was evident from 1990–1994 to 1995–1998 (Table III).

Table III. RSRs Among Patients Diagnosed with Local- or Regional-Stage Esophageal Cancer: Patients Who Received neither Cancer-Directed Surgery nor Radiation Therapy
Years of diagnosisNumberRSR (%) (95% confidence interval)
1 year3 years5 years
  • 1

    Limited follow-up.

  • *

    p < 0.05, for comparison with RSR in 1975–1979 (see text).

Squamous cell carcinoma    
 1975–1979112632.0 (29.1–34.8)8.6 (6.6–10.1)5.8 (4.2–7.3)
 1980–1984102537.0 (33.9–40.1)12.6 (10.4–14.8)*8.8 (6.8–10.8)
 1985–1989100843.9 (40.6–47.1)*18.3 (15.6–20.9)*11.9 (9.7–14.2)*
 1990–1994104446.1 (42.9–49.3)*20.9 (18.2–23.6)*15.8 (13.2–18.4)*
 1995–199876347.1 (43.3–50.8)*19.3 (15.9–22.7)*1
Adenocarcinoma    
 1975–197917243.9 (36.1–51.7)17.5 (11.3–23.6)9.1 (4.3–14.0)
 1980–198421749.7 (42.6–56.7)23.7 (17.5–29.9)18.5 (12.5–24.4)
 1985–198939950.0 (44.8–55.2)20.8 (16.4–25.1)16.2 (12.0–20.4)
 1990–199460757.0 (52.8–61.2)*28.8 (24.8–32.8)*21.4 (17.5–25.2)*
 1995–199863056.8 (52.7–61.0)*30.3 (25.8–34.8)*1

Among patients with radical surgery (i.e., partial or total esophagectomy plus partial or total removal of other organs), coded in the SEER database only for diagnoses in 1988 and later, samples were small but improvement in survival was evident (albeit not statistically significant) for adenocarcinoma and not for squamous cell carcinoma (Table IV). Any reductions in mortality (after esophagectomy) due to improvements in perioperative care during the last 2 decades9 should have affected RSRs for both histologic types. However, the possibility that increasingly earlier detection of adenocarcinomas, due to endoscopic screening of the distal esophagus for Barrett's metaplasia and gastroesophageal reflux disease,9 may have increased the effectiveness of surgery requires further examination.

Table IV. RSRs Among Patients Diagnosed with Esophageal Cancer who Received Radical Surgery
Years of diagnosisNumberRSR (%) (95% confidence interval)
1 year3 years5 years
  • 1

    Radical surgery was not coded in the SEER database until (year of diagnosis) 1988.

  • 2

    Limited follow-up.

Squamous cell carcinoma    
 1988–198916860.3 (48.0–72.5)25.4 (14.3–36.6)18.6 (8.3–28.9)
 1990–199418060.3 (52.8–67.9)28.9 (21.7–36.0)21.7 (15.0–28.4)
 1995–19986959.1 (48.7–71.3)26.4 (14.6–38.2)2
Adenocarcinoma   
 1988–198919355.2 (44.6–65.8)24.6 (15.2–34.1)23.1 (13.3–32.8)
 1990–199424067.6 (61.3–73.9)35.5 (28.9–42.1)27.3 (20.9–33.7)
 1995–199814671.4 (63.6–79.2)39.3 (30.5–48.2)2

Any impact of increasing use of chemotherapy could not be assessed because chemotherapy is underreported to SEER registries and is not included on public-use data files.3 Conflicting results have been reported from clinical trials of chemotherapy in local–regional operable esophageal cancer, and additional trials are needed.9, 10

Another hypothesis for the temporal increase in RSRs involves temporal changes in prognostic factors other than stage at diagnosis among esophageal cancer patients. Such changes would need to have occurred in both male and female patients. A large effect of the declining prevalence of smoking in the general population (and presumably among cancer patients), which would affect the risk of second primary cancer and (hence) survival among esophageal cancer patients, is unlikely because smoking is more strongly associated with squamous cell carcinoma than with adenocarcinoma of the esophagus,7 whereas rising RSRs were evident in both histologic groups (Tables I–III.

Although improvements in RSRs are evident in the SEER data, the highest 3-year RSR was still only 44% (Table II). U.S. mortality rates for esophageal cancer have continued to rise slightly (through 1999), which is consistent with the rising incidence rates (mainly in white males) in SEER data.2

REFERENCES

  1. Top of page
  2. REFERENCES