Objective
While the last 3 decades have seen numerous advances in the treatment of cervical cancer, it remains unclear if population-level survival has improved. We examined relative survival, the ratio of survival in cervical cancer patients to matched controls over time.
Study Design
Patients with cervical cancer diagnosed from 1983 through 2009 and recorded in the Surveillance, Epidemiology, and End Results database were examined. Survival models were adjusted for age, race, stage, year of diagnosis, and time since diagnosis. Changes in stage-specific relative survival for patients with cervical cancer compared to the general population matched by age, race, and calendar year were examined over time.
Results
A total of 46,932 patients were identified. For women with stage I tumors, the excess hazard ratio for women diagnosed in 2009 was 0.91 (95% confidence interval [CI], 0.86–0.95) compared to 2000, 0.81 (95% CI, 0.73–0.91) compared to 1990, and 0.75 (95% CI, 0.64–0.88) compared to 1983. For patients with stage III tumors, the excess hazard ratios for patients diagnosed in 2009 (relative to those diagnosed in 2000, 1990, and 1983) were 0.83 (95% CI, 0.80–0.87), 0.68 (95% CI, 0.62–0.75), and 0.59 (95% CI, 0.52–0.68). Similar trends in improved survival over time were noted for women with stage II tumors. There were no statistically significant improvements in relative survival over time for women with stage IV tumors.
Conclusion
Relative survival has improved over time for women with stage I-III cervical cancer, but has changed little for those with metastatic disease.
Cervical cancer remains a major cause of cancer-related mortality worldwide. Over the last 3 decades, a number of treatment advances for cervical cancer have been demonstrated in clinical trials. For women with early-stage disease, newer surgical options are available and techniques for the delivery of radiation have improved. For advanced-stage disease, improved survival for the use of combination chemotherapy and radiotherapy resulted in a paradigm shift for patients with stage II-IV disease in 1999. However, data describing how survival has changed over time remain limited.
Quantifying changes in survival for cancer patients is of great importance as therapeutic advances that have shown efficacy in clinical trials are of little practical value if these treatments cannot be translated into clinical practice. However, examining secular trends in survival for cancer is methodologically challenging. First, as general medical care has improved over time, it is difficult to ascertain if improved survival for cancer patients is due to improved cancer treatment or due to greater longevity in the population as a whole. Second, measuring cancer-specific survival is inherently difficult as data from death certificates are often inaccurate and may not reflect cancer-associated mortality in patients who die from complications and the sequelae of cancer.
To quantify secular trends in survival for cancer patients, relative survival, the ratio of the observed survival rate for cancer patients to the expected survival rate of matched patients from the general population, has been described. Relative survival is a useful metric that controls for changes in survival in the general population and describes excess mortality in cancer patients over time. We performed a population-based analysis to examine secular changes in survival for women with cervical cancer treated in the United States from 1983 through 2009.
Materials and Methods
Data source
Data from the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) database were used for analysis. The SEER program comprehensively collects data on all newly diagnosed cancer patients from a number of registries located throughout the United States. We included patients with invasive cervical cancer diagnosed from January 1983 through December 2009 with follow-up through Dec. 31, 2011.
Data from the SEER 18 registries including San Francisco-Oakland, CA; Connecticut; Detroit, MI (metropolitan); Hawaii; Iowa; New Mexico; Utah; Seattle, WA (Puget Sound); Atlanta, GA (metropolitan); San Jose-Monterey, CA; Los Angeles, CA; Alaska Natives; rural Georgia; greater California; Kentucky; Louisiana; New Jersey; and greater Georgia were utilized. Louisiana cases diagnosed for July through December 2005 were excluded due to the impact of Hurricanes Katrina and Rita on the registry’s ability to report data. Non-white and non-black women were excluded from the analyses since reliable population-level estimates of survival were required. Patients with unknown stage were excluded. The Columbia University Institutional Review Board deemed the study exempt.
Staging
Staging was based on the derived sixth edition of the American Joint Committee on Cancer (AJCC) staging for patients diagnosed from 2004 through 2009, and the SEER modified third edition of AJCC staging for those women diagnosed from 1988 through 2003. Prior to 1988, AJCC staging was not recorded in SEER. For those women diagnosed prior to 1988, we constructed AJCC staging through the use of 4-digit extent of disease codes for patients treated in 1983 through 1987.
Statistical analysis
The primary analysis focused on overall survival defined as the time from diagnosis of cervical cancer until death from any cause. Relative survival, the ratio of the observed survival rate for cancer patients to the expected survival rate of matched patients from the general population, was then estimated. Patients in the general population were matched to those with cervical cancer based on age, race, and calendar year using the Ederer II method calculated using SEER*Stat software. Expected survival life tables were used to derive survival estimates for the controls. The expected life tables provide survival by age, race, and calendar year. Estimates were derived from interpolating the US Decennial Life Tables from the National Center for Health Statistics (NCHS) into individual years for the years 1970 through 2000 and from the US Annual Life Tables from NCHS for the years 2001 through 2009. Although the effect of deaths due to cervical cancer is also included in the life tables, this does not affect the estimated survival of the populations. After matching, relative survival models were developed using annual intervals in the framework of generalized linear models with a Poisson error structure. When modeling relative survival, the model is an additive hazards model where the total hazard is the sum of the baseline hazard from the control and the excess hazard associated with a diagnosis of cervical cancer. The exponentiated parameter estimates are interpreted as excess hazard ratios (HR).
We fit separate models for patients with stage I, II, III, and IV neoplasms. Each of the models included age at the time of diagnosis, race, year of diagnosis, and time since diagnosis. Age at diagnosis was defined as a categorical variable, <50 years, 50-59 years, 60-69 years, 70-79 years, and ≥80 years, consistent with SEER reporting methodology. Year of diagnosis was included as a linear function. Time since diagnosis was a time-varying covariate updated in 1-year increments at the end of the yearly interval. Time since diagnosis was included as a piecewise linear function of year allowing changes in slope at 2, 5, and 10 years after the time of diagnosis to allow rapid change in excess hazards. Data on each covariate were updated with each new interval. Goodness of fit of the relative survival models was examined using deviance statistics.
All analyses were performed using software (SAS, version 9.4; SAS Institute Inc, Cary, NC). Statistical tests were 2-sided, and a P value of < .05 was considered statistically significant.
Results
We identified 46,932 women with invasive cervical cancer diagnosed from 1983 through 2009. Within the cohort, 26,337 (56.1%) women had stage I tumors; 7091 (15.1%), stage II; 8090 (17.2%), stage III, and 5414 (11.5%), stage IV malignancies ( Table 1 ) . For all stages, women <50 years of age made up the greatest percentage of patients. Women with higher stage tumors were older, with 36.3% of stage IV patients <50 years of age compared to 68.4% for patients with stage I neoplasms. The percentage of black women increased with stage from 13.2% of patients with stage I tumors to 17.9% of stage III and 18.3% of stage IV patients.
| Characteristic | Stage I | Stage II | Stage III | Stage IV | ||||
|---|---|---|---|---|---|---|---|---|
| n | % | n | % | n | % | n | % | |
| No. | 26,337 | (56.1) | 7091 | (15.1) | 8090 | (17.2) | 5414 | (11.5) |
| Age at diagnosis, y | ||||||||
| <50 | 18,023 | (68.4) | 3045 | (42.9) | 3943 | (48.7) | 1965 | (36.3) |
| 50–59 | 3770 | (14.3) | 1579 | (22.3) | 1747 | (21.6) | 1276 | (23.6) |
| 60–69 | 2534 | (9.6) | 1194 | (16.8) | 1135 | (14.0) | 1044 | (19.3) |
| 70–79 | 1388 | (5.3) | 797 | (11.2) | 799 | (9.9) | 709 | (13.1) |
| ≥80 | 622 | (2.4) | 476 | (6.7) | 466 | (5.8) | 420 | (7.8) |
| Race | ||||||||
| Black | 3485 | (13.2) | 1262 | (17.8) | 1448 | (17.9) | 990 | (18.3) |
| White | 22,852 | (86.8) | 5829 | (82.2) | 6642 | (82.1) | 4424 | (81.7) |
| Year of diagnosis | ||||||||
| 1983 through 1990 | 4671 | (17.7) | 1272 | (17.9) | 840 | (10.4) | 669 | (12.4) |
| 1991 through 2000 | 8982 | (34.1) | 2214 | (31.2) | 2418 | (29.9) | 1564 | (28.9) |
| 2001 through 2009 | 12,684 | (48.2) | 3605 | (50.8) | 4832 | (59.7) | 3181 | (58.8) |
For all stages of disease, black women had higher excess HR than white women ( Table 2 ). The excess HR for black women relative to white women was 1.70 (95% confidence interval [CI], 1.53–1.88) for stage I, 1.22 (95% CI, 1.11–1.34) for stage II, 1.31 (95% CI, 1.20–1.43) for stage III, and 1.23 (95% CI, 1.10–1.37) for stage IV tumors ( P < .0001). For all stages we noted reductions in excess mortality over time. For women with stage I tumors (HR, 0.99; 95% CI, 0.98–1.00), the excess HR for women diagnosed in 2009 was 0.91 (95% CI, 0.86–0.95) compared to those diagnosed in 2000, 0.81 (95% CI, 0.73–0.91) compared to women diagnosed in 1990, and 0.75 (95% CI, 0.64–0.88) in comparison to patients diagnosed in 1983.
| Characteristic | Stage I | Stage II | Stage III | Stage IV | ||||
|---|---|---|---|---|---|---|---|---|
| Excess HR | P value | Excess HR | P value | Excess HR | P value | Excess HR | P value | |
| Black, relative to white | 1.70 (1.53–1.88) | < .001 | 1.22 (1.11–1.34) | < .001 | 1.31 (1.20–1.43) | <.001 | 1.23 (1.10–1.37) | < .001 |
| Age at diagnosis, y | ||||||||
| <50, relative to 50–59 | 0.54 (0.49–0.61) | < .001 | 1.01 (0.91–1.11) | .88 | 0.74 (0.68–0.80) | < .001 | 0.86 (0.77–0.96) | .007 |
| 60–69, relative to 50–59 | 1.27 (1.09–1.48) | .002 | 1.03 (0.91–1.17) | .67 | 1.04 (0.93–1.17) | .48 | 1.11 (0.97–1.26) | .12 |
| 70–79, relative to 50–59 | 1.42 (1.16–1.75) | < .001 | 1.27 (1.10–1.47) | .001 | 1.40 (1.23–1.59) | < .001 | 1.34 (1.16–1.55) | < .001 |
| ≥80, relative to 50–59 | 2.65 (2.00–3.51) | < .001 | 1.73 (1.42–2.11) | < .001 | 1.57 (1.31–1.88) | < .001 | 1.36 (1.11–1.66) | .003 |
| Year of diagnosis a | ||||||||
| 2009, relative to 2005 | 0.96 (0.93–0.98) | < .001 | 0.91 (0.90–0.93) | < .001 | 0.92 (0.90–0.94) | < .001 | 0.99 (0.96–1.01) | .24 |
| 2009, relative to 2000 | 0.91 (0.86–0.95) | < .001 | 0.82 (0.78–0.86) | < .001 | 0.83 (0.80–0.87) | < .001 | 0.97 (0.91–1.02) | .24 |
| 2009, relative to 1995 | 0.86 (0.79–0.93) | < .001 | 0.73 (0.68–0.79) | < .001 | 0.75 (0.70–0.81) | < .001 | 0.95 (0.87–1.04) | .24 |
| 2009, relative to 1990 | 0.81 (0.73–0.91) | < .001 | 0.65 (0.59–0.72) | < .001 | 0.68 (0.62–0.75) | < .001 | 0.93 (0.83–1.05) | .24 |
| 2009, relative to 1983 | 0.75 (0.64–0.88) | < .001 | 0.56 (0.49–0.64) | < .001 | 0.59 (0.52–0.68) | < .001 | 0.91 (0.77–1.07) | .24 |
| Time since diagnosis b | ||||||||
| 1 y | 3.56 (3.20–3.97) | < .001 | 3.39 (3.13–3.68) | < .001 | 2.67 (2.52–2.82) | < .001 | 2.60 (2.45–2.76) | < .001 |
| 5 y | 4.32 (3.46–5.38) | < .001 | 3.00 (2.52–3.57) | < .001 | 1.25 (1.08–1.44) | < .001 | 0.59 (0.47–0.75) | < .001 |
| 10 y | 2.49 (1.93–3.22) | < .001 | 0.95 (0.73–1.24) | < .001 | 0.45 (0.35–0.59) | < .001 | 0.23 (0.13–0.41) | < .001 |
a P values for year of diagnosis are same since year of diagnosis is included as continuous
b P values for time since diagnosis are for changes in slope at 2, 5, and 10 y after time of diagnosis.
Similar trends were noted for women with both stage II and III tumors; there was a consistent decline in the excess hazard of death over time: stage II (HR, 0.98; 95% CI, 0.97–0.98) and stage III (HR, 0.98; 95% CI, 0.98–0.99). For patients with stage III tumors, this translated to an excess HR of 0.83 (95% CI, 0.80–0.87) for patients diagnosed in 2009 relative to 2000 and 0.59 (95% CI, 0.52–0.68) relative to women diagnosed in 1983. We noted modest reductions in excess mortality over time, although these reductions did not reach statistical significance for women with stage IV neoplasms (HR, 1.00; 95% CI, 0.99–1.00).
Table 3 displays the cumulative relative survival over time stratified by age and years since diagnosis. For most age groups, we noted that the decrease in relative survival was more pronounced over time. These absolute reductions in relative survival were greater for women with more advanced-stage tumors. For example, among women <50 years of age with stage I tumors, relative survival was 0.98 (95% CI, 0.96–0.99) at 1 year and decreased to 0.93 (95% CI, 0.89–0.95) at 10 years. The corresponding relative survival values for women <50 years of age with stage IV neoplasms were 0.43 (95% CI, 0.22–0.62) at 1 year and 0.05 (95% CI, 0–0.20) at 10 years.
| Time | Stage I | Stage II | Stage III | Stage IV | ||||
|---|---|---|---|---|---|---|---|---|
| Observed survival | Relative survival | Observed survival | Relative survival | Observed survival | Relative survival | Observed survival | Relative survival | |
| DIAGNOSED AT AGE <50 Y | ||||||||
| Years since diagnosis | ||||||||
| 1 | 0.98 (0.96–0.99) | 0.98 (0.96–0.99) | 0.91 (0.80–0.96) | 0.91 (0.80–0.96) | 0.81 (0.68–0.89) | 0.81 (0.68–0.89) | 0.43 (0.22–0.62) | 0.43 (0.22–0.62) |
| 5 | 0.93 (0.90–0.95) | 0.94 (0.91–0.96) | 0.60 (0.46–0.71) | 0.60 (0.46–0.72) | 0.53 (0.39–0.65) | 0.53 (0.39–0.65) | 0.14 (0.04–0.32) | 0.14 (0.04–0.32) |
| 10 | 0.92 (0.88–0.94) | 0.93 (0.89–0.95) | 0.50 (0.37–0.63) | 0.51 (0.37–0.64) | 0.42 (0.29–0.54) | 0.43 (0.30–0.55) | 0.05 (0.00–0.20) | 0.05 (0.00–0.20) |
| DIAGNOSED AT AGE 50–59 Y | ||||||||
| Years since diagnosis | ||||||||
| 1 | 1.00 (─ b ) | 1.00 a (─ b ) | 0.87 (0.65–0.96) | 0.87 (0.65–0.96) | 0.73 (0.49–0.87) | 0.73 (0.49–0.87) | 0.40 (0.12–0.67) | 0.40 (0.12–0.67) |
| 5 | 0.89 (0.80–0.94) | 0.92 (0.81–0.97) | 0.61 (0.38–0.77) | 0.63 (0.39–0.79) | 0.32 (0.14–0.51) | 0.33 (0.15–0.52) | 0.30 (0.07–0.58) | 0.30 (0.07–0.59) |
| 10 | 0.78 (0.67–0.86) | 0.85 (0.71–0.92) | 0.39 (0.20–0.58) | 0.42 (0.21–0.62) | 0.14 (0.03–0.31) | 0.15 (0.04–0.33) | 0.10 (0.01–0.36) | 0.11 (0.01–0.37) |
| DIAGNOSED AT AGE 60–69 Y | ||||||||
| Years since diagnosis | ||||||||
| 1 | 0.93 (0.84–0.98) | 0.95 (0.83–0.98) | 0.80 (0.50–0.93) | 0.81 (0.50–0.94) | 0.57 (0.34–0.75) | 0.58 (0.34–0.76) | 0.44 (0.22–0.65) | 0.45 (0.22–0.66) |
| 5 | 0.85 (0.74–0.92) | 0.90 (0.75–0.96) | 0.33 (0.12–0.56) | 0.35 (0.13–0.59) | 0.29 (0.12–0.48) | 0.31 (0.12–0.52) | 0.06 (0.00–0.22) | 0.06 (0.00–0.24) |
| 10 | 0.65 (0.52–0.76) | 0.79 (0.61–0.89) | 0.27 (0.08–0.50) | 0.30 (0.09–0.55) | 0.19 (0.06–0.38) | 0.23 (0.07–0.45) | 0.06 (0.00–0.22) | 0.06 (0.00–0.24) |
| DIAGNOSED AT AGE 70–79 Y | ||||||||
| Years since diagnosis | ||||||||
| 1 | 1.00 (─ b ) | 1.00 a (─ b ) | 0.89 (0.62–0.97) | 0.92 (0.56–0.99) | 0.53 (0.18–0.79) | 0.55 (0.18–0.81) | 0.07 (0.01–0.28) | 0.07 (0.01–0.28) |
| 5 | 0.83 (0.60–0.93) | 0.96 (0.11–1.00) | 0.22 (0.07–0.43) | 0.27 (0.08–0.51) | 0.27 (0.04–0.58) | 0.28 (0.04–0.61) | 0.00 (─ b ) | 0.00 (─ b ) |
| 10 | 0.52 (0.31–0.70) | 0.86 (0.15–0.99) | 0.00 (─ b ) | 0.00 (─ b ) | 0.13 (0.01–0.44) | 0.21 (0.01–0.61) | 0.00 (─ b ) | 0.00 (─ b ) |
| DIAGNOSED AT AGE ≥80 Y | ||||||||
| Years since diagnosis | ||||||||
| 1 | 0.90 (0.47–0.99) | 1.00 a (─ b ) | 0.90 (0.45–0.98) | 0.96 (0.00–1.00) | 0.67 (0.28–0.88) | 0.71 (0.27–0.92) | 0.33 (0.05–0.68) | 0.37 (0.05–0.73) |
| 5 | 0.50 (0.18–0.75) | 0.82 (0.01–0.99) | 0.34 (0.08–0.63) | 0.52 (0.07–0.85) | 0.11 (0.01–0.39) | 0.13 (0.01–0.43) | 0.00 (─ b ) | 0.00 (─ b ) |
| 10 | 0.00 (─ b ) | 0.00 (─ b ) | 0.11 (0.01–0.39) | 0.25 (0.01–0.69) | 0.00 (─ b ) | 0.00 (─ b ) | 0.00 (─ b ) | 0.00 (─ b ) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
