| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
ORIGINAL RESEARCH |
From the Royal Adelaide Hospital, Adelaide, South Australia; Institute of Medical and Veterinary Science, Adelaide, South Australia; Epidemiology Branch, Department of Human Services, Adelaide, South Australia; and Anti-Cancer Foundation of South Australia, Unley, South Australia, Australia.
Address reprint requests to: David M. Roder, DDSc, MPH, Anti-Cancer Foundation of South Australia, GPO Box 929, Unley, South Australia 5061; E-mail: roder{at}chariot.net.au.
 |
ABSTRACT |
|---|
TOP
 ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
METHODS: Data on cervical cancers, diagnosed in 1984–2000, were obtained from the gynecologic oncology registry of hospitals of the University of Adelaide. Comparisons were made of disease-specific survival, age at diagnosis, diagnostic period, stage, grade, and primary course of treatment.
RESULTS: The study included 544 squamous cell carcinomas, 43 adenosquamous carcinomas, five clear cell cancers, 136 other adenocarcinomas, and 19 cancers of "other" histological type. Overall survival was 72.2% at 5 years from diagnosis, decreasing to 67.5% at 15 years. Survival was lower for older ages, higher grades, and higher International Federation of Gynecology and Obstetrics stages, although equivalent for stages IIA and IIB. Unadjusted survivals varied by histological type (P = .001), with lower survivals suggested for adenosquamous and clear cell lesions and "other" histological types than for squamous cell carcinomas and other adenocarcinomas. After adjusting for age, stage, grade, and diagnostic period, adenocarcinomas had a higher case fatality than squamous cell lesions (relative risk 2.08, 95% confidence limit 1.35, 3.21), whereas the elevation in relative risk was lower and not statistically significant for a combined adenosquamous and clear cell category at 1.25 (0.69, 2.24). For stage II, both adenocarcinomas and the adenosquamous and clear cell group had lower survivals than squamous cell cancers.
CONCLUSION: Relative to squamous cell carcinomas, adenocarcinomas and potentially adenosquamous cancers are becoming more common. This has implications for screening, treatment, and prognosis.
Australian data, available nationally since the 1980s, show an annual decrease in age-standardized (world population) incidence of cervical cancer of about 4% and a corresponding decrease in mortality of 7%.1 South Australian data show similar trends.2
Population-based data for South Australia also reveal differences in time trend by histological type.2 Although the age-standardized incidence of squamous cell carcinomas decreased by 48% between 1977–1984 and 1995–1999, there was no reduction for glandular lesions (adeno- and adenosquamous cell carcinomas). On the contrary, in women under 50 years, the incidence increased by 40%.2 Glandular lesions have become more common, relative to squamous cell lesions,2 and similar trends have been reported nationally,1 and for other populations, especially in younger women.3–7
Epidemiological features of cervical cancers vary by histological type. Compared with squamous cell carcinomas, glandular lesions have occurred more commonly in younger women and in higher socioeconomic areas.2 Although findings have been inconsistent, the prognosis for glandular carcinomas has been reported to be less favorable than for squamous cell lesions, and differences in management have been advocated.6,8,9
Using data from the oncology registry of the Gynecologic Oncology Unit of hospitals of the University of Adelaide for the 1984–2000 diagnostic period, survivals of women diagnosed with cervical cancer were analyzed with special reference to histological type and its influence on prognostic characteristics, management, and outcomes. The data included clinical details for this purpose that were not available through the state’s population-based registry.2
|
MATERIALS AND METHODS |
|---|
|
TOP
ABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Cancers were divided into squamous cell, glandular, and "other" histological types (mostly sarcomas and epithelial cancers not otherwise specified). Glandular lesions were further subdivided into adenosquamous, clear cell, and other adenocarcinomas. Because there were too few clear cell cancers for separate analysis (n = 5), it was decided on a prognostic basis to include them with the adenosquamous category.
Data items and data collection methods used by the oncology registry have been described in annual reports.2,9 Items include: International Classification of Disease for Oncology (3rd edition) histological codes,17 and International Federation of Gynecology and Obstetrics stage and grade.18 Registry operations and data standards criteria were used to classify patients according to whether they received surgery, radiotherapy, and chemotherapy, respectively, as part of their primary course of treatment.19 Patients were similarly classified according to whether they received specified combinations of treatments.
Hospitals were provided with the following information on their patients by the State Cancer Registry: date of birth, date of diagnosis, date of death, and cause of death.9 The Registry obtained the death data from official state death records, the National Death Index at the Australian Institute of Health and Welfare, and from interstate cancer registries.9 Under-reporting was checked periodically through active follow-up, and with deaths reported independently, and was found to be minimal, with little effect on calculated survivals.9,20 Further details of registry operations are reported elsewhere.2,9 The collection of State Cancer Registry data is mandated by State law.
Three sets of analyses were performed, using STATA 6.0 software21 as follows.
Disease-specific survival was calculated, using the Kaplan-Meier product-limit estimate.22 The relative risk of case fatality from cervical cancer (the converse of disease-specific survival) was investigated by histological type, using multivariable Cox proportional hazards regression.23 Censoring occurred at death from other causes, or on December 31, 2000, whichever came first, both for the Kaplan-Meier and regression analyses.
All patient and tumor variables were entered into the multivariable analysis, with backwards elimination of those where the fit of the model was not reduced (ie, P > .05 for change in 
2 goodness of fit). The exception was diagnostic epoch (1984–1993 and 1994–2000) and histological category, which were retained to gain an indication of trend. Assumptions underlying the analysis, such as proportionality and an absence of colinearity, were tested and found to be met.23
The potential for clustering of observations by hospital was tested by comparing the robust variance estimator with the corresponding variance estimator (unclustered). Little or no clustering was apparent, and an independence of observations was assumed.
Age at diagnosis, diagnostic period, stage, and grade were compared by histological type, using the Kruskal-Wallis analysis of variance for the four histological groups and the Mann-Whitney U test when comparisons were limited to two histological categories.23
Multiple logistic regression then was used to indicate predictors of each glandular subcategory, compared with squamous cell lesions.23 All predictor variables were entered into the regression analysis, with backwards elimination, as described for the Cox proportional hazards regression. Again, assumptions underlying the model, such as a lack of colinearity, were checked and found to be met. Model calibration was checked by comparing observed and expected numbers of glandular lesions across quantiles of estimated probabilities, as described by Hosmer and Lemeshow.24
The provision or nonprovision of surgery, radiotherapy, and chemotherapy, respectively, was analyzed using the Mann-Whitney U test for all patient and tumor variables apart from histological type, where the Pearson 2 was employed (substituting Fisher exact test where cell size requirements were not met).23
The relative odds of provision of individual treatments also were analyzed by histological type, using multiple logistic regression. The predictors were histological type (dummy variables) and other variables that remained after backwards elimination of factors that did not contribute to the goodness of fit (P > .05).25 Three separate analyses were undertaken for surgery, radio-therapy, and chemotherapy, as the dependent variable, respectively.
|
RESULTS |
|---|
|
TOP
ABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Data ascertainment was complete for all characteristics except International Federation of Gynecology and Obstetrics stage and grade, where the level of completeness was 99.9% and 90.0%, respectively. The cancers comprised 544 squamous cell carcinomas, 43 adenosquamous carcinomas, five clear cell cancers, 136 other adenocarcinomas, and 19 cancers of "other" histological type.
The overall survival of 72.2% from cervical cancer at 5 years from diagnosis decreased to 68.2% at 10 years and 67.5% at 15 years (Table 1
). Survival decreased with age at diagnosis (P < .001), the 5-year figure ranging from 84.1% for patients under 40 years to 37.2% for those aged 80 years or more. Survival did not vary in these univariate analyses between the 1984–1988, 1989–1993, and 1994–2000 diagnostic periods (P = .831).
|
), with the proportion surviving 5 years from diagnosis decreasing from 100% for stage 1A to 11.6% for stage IV. Survival was similar, however, for stages IIA and IIB. Grade was predictive of outcome, with higher grades having lower survivals (Table 1).
Survival varied by histological category (P = .001), with poorer outcomes suggested for the combined adenosquamous and clear cell group and for "other" cancers, than for adenocarcinomas or squamous cell lesions (Table 1).
The predictors of a high case fatality were: an age at diagnosis of 80 years or more; higher stage (although fatality was not higher for stage IIB than IIA); high grade; and an adenocarcinoma as opposed to a squamous cell lesion (Table 2). Although the relative risk (95% confidence limit) was 2.08 (1.35, 3.21) for adenocarcinomas compared with squamous cell lesions, it was lower at 1.25 (0.69, 2.24) for the combined adenosquamous and clear cell category and not statistically significant (P = .463).
|
Although the diagnostic period was not statistically significant (P = .136), case fatality tended to be lower for 1994–2000 than 1984–1993, with a relative risk of 0.77 (0.54, 1.09). Stratified analyses pointed to a similar secular decline by histological category.
Stage-specific analyses (using the four main International Federation of Gynecology and Obstetrics categories) were limited in statistical power because of small patient numbers. Nonetheless, glandular lesions were found to have elevated case fatalities for stage II, the relative risk being 2.59 (1.40, 4.78) for adenocarcinomas (n = 28), and 3.89 (1.82, 8.29) for the combined adenosquamous and clear cell category (n = 10), when compared with the squamous cell lesions (n = 166). Similar results applied, when adjusting for age, substage, grade, and diagnostic period in a multivariable proportional hazards regression. A statistically significant difference in case fatality by histological type did not apply for stages I, III, or IV, neither when the analysis was unadjusted, nor when there was adjustment for age, substage (for stage I), and grade (P 
.100).
Differences in distributions of age, diagnostic period, stage, and grade were evident by histological category (Table 3). Compared with squamous cell lesions, adenocarcinomas affected younger patients (P = .003), were diagnosed more recently (P = .002), and had lower stages (P < .001) and grades (P < .001).
|
).
Results indicate that, compared with squamous cell lesions, adenocarcinomas tended to be diagnosed more recently, and to be stage IB and lower in grade (Table 4). The adenosquamous and clear cell group also tended more to be stage IB, and to be diagnosed recently, but statistical significance was not achieved (P .091). Unlike adenocarcinomas, however, they presented higher grades than the squamous cell lesions (P = .012) (Table 4).
|
). Unlike stage, grade was not a surgical criterion, but its association with surgery was influenced indirectly by its positive association with stage (P < .001).
|
). Multiple logistic regression confirmed that after adjusting for age, stage, and diagnostic period, surgery was provided more frequently for adenocarcinomas and (possibly) the combined adenosquamous and clear cell category than for squamous cell cancers. The relative odds (95% confidence limits) of surgery were: for adenocarcinomas, 2.77 (1.32, 5.81); for the adenosquamous and clear cell group, 2.69 (0.92, 7.86); and for "other" histological types, 3.81 (0.93, 15.73).
Overall, 53.9% of patients had radiotherapy as part of the primary course of care. The proportion was larger for older age groups (P < .001) and also varied by stage (P < .001), increasing from stage IA to IIB, but then decreasing in the higher stages (Table 5). As for surgery, grade was not a criterion for radiotherapy. The association of higher grade with radiotherapy (P < .001) was influenced indirectly by its positive association with stage.
Treatment by radiotherapy also varied by histological type (P < .001), with the proportion receiving this treatment being highest for squamous cell lesions (61%) and lowest for adenocarcinomas (26%) (Table 5). Multiple logistic regression confirmed that after adjusting for age, stage, and diagnostic period, adenocarcinomas were less likely to receive this treatment than squamous cell lesions, the relative odds being 0.15 (0.08, 0.29). The relative odds of radiotherapy also tended to lower for the combined adenosquamous and clear cell category at 0.50 (0.22, 1.11), whereas a difference was not suggested in exposure to this treatment between "other" histological types and squamous cell carcinomas (P = .474).
Overall, 15.1% of patients had chemotherapy as part of their primary care. The proportion varied by stage (P < .001) and grade (P < .001), with fewer stage I and lower grade lesions receiving this treatment (Table 5).
There was also a difference in the proportion exposed to chemotherapy by histological type (P = .002), the proportion being highest for the adenosquamous and clear cell category (31%) and lowest for the squamous cell carcinomas (13%) (Table 5).
Multiple logistic regression indicated that after adjusting for age, stage, and diagnostic period, the proportion having chemotherapy was higher for the adenosquamous and clear cell group and (possibly) adenocarcinomas than for squamous cell lesions, the relative odds being 3.03 (1.35, 6.81) and 1.82 (0.98, 3.36), respectively. Meanwhile, exposure to chemotherapy was not found to differ between the "other" histological types and the squamous cell carcinomas (P = .906).
|
DISCUSSION |
|---|
|
TOP
ABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Unadjusted analyses pointed to lower survivals for the combined category of adenosquamous and clear cell lesions than for squamous cell cancers and adenocarcinomas. This was affected by their higher grades. After adjusting for grade, stage, age, and diagnostic period, the elevated relative risk for these cancers was not statistically significant, when compared with squamous cell lesions, except in stage II. By comparison, adenocarcinomas presented with lower grades and stages than squamous cell cancers. After adjusting for these factors, age, and diagnostic period, they had lower survivals than squamous cell carcinomas.
Although it is possible that criteria for histological assessment may have changed over the 17-year period of this study, tissue biopsies were examined by the one pathologist group, and there was no indication from interaction terms that differences in survival by histological category varied across the three diagnostic periods. Similar results applied when the diagnostic period was expressed as calendar years. Despite these observations, a review of all histopathological specimens would be desirable to seek additional assurance that the results were not influenced by measurement variation.
Adenocarcinomas and the combined adenosquamous and clear cell category were treated differently from squamous cell lesions in this case series, as observed previously.9 In general, more of these patients had a surgical intervention as part of the primary course of care, fewer had radiotherapy, and more had chemotherapy, which reflects differences in recommended protocols.
The similarity in patient survivals in this study for stage IIA and IIB cervical cancers was unexpected and might reflect an increased potential for random error from the comparatively small number of stage IIA patients available for analysis. Alternatively, it is possible that patterns of selective referral, or the exposure of a higher proportion of stage IIB than IIA patients to radio-therapy may have had an influence.
Tumor grade was predictive of outcome in this study, with high-grade lesions having a higher case fatality. This is at odds with some earlier results for squamous cell carcinomas of the cervix.26 Meanwhile, previous results for glandular lesions have been variable.27,28 In this investigation, the one pathologist group undertook the assessments of grade throughout the study period. It seems possible to us that their assessments of grade may have been influenced by accompanying features, such as degree of keratinization, pattern of invasion, and host lymphoid response.29
Limited evidence is available on the effectiveness of cytology screening for glandular lesions.5,30,31 Additional research on this topic and on the treatment of glandular cancers is becoming more imperative as the frequency of these lesions increases. With changes in cell type, the elevation in incidence of cervical cancer now seen in older women and lower socioeconomic groups probably would diminish.1,2,21,32 This would have implications for the targeting of screening programs.
Glandular lesions have negative public health qualities, in that they impact on women’s lives at an earlier age (the median age at diagnosis in South Australia was 48 years in 1977–2000 compared with 50 years for squamous cell lesions).2 Because the median age at death from glandular lesions is 2 years younger than for squamous cell carcinomas (63 years compared with 65 years during 1977–2000),2 glandular lesions are likely to cause more loss of life years per patient.
|
Footnotes |
|---|
The authors are indebted to the Adelaide Gynecologic Oncology Group, both medical specialists and trainees, for supporting the data collection for this study and to registry staff, Ms. Margaret Colbeck and Ms. Helen Reissenweber. Past staff of the registry, who also are acknowledged for their contributions, include Ms. Juliet Bradley, Ms. Nancy Olszewski, Ms. Heather Hall, and Ms. Melinda Myers.
Received March 12, 2002. Received in revised form May 28, 2002. Accepted July 18, 2002.
|
REFERENCES |
|---|
|
TOP
ABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2. South Australian Cancer Registry. Epidemiology of cancer in South Australia. Incidence, mortality and survival, 1977 to 2000. Incidence and mortality, 2000. Adelaide, Australia: Openbook Publishing House, 2001:83–90.
3. Liu S, Semenciw R, Mao Y. Cervical cancer: The increasing incidence of adenocarcinoma and adenosquamous carcinoma in younger women. CMAJ 2001;164:1151–2.
4. Bergstrom R, Sparen P, Adami HO. Trends in cancer of the cervix uteri in Sweden following cytological screening. Br J Cancer 1999;81:159–66.[Medline]
5. Zheng T, Holford TR, Ma Z, Chen Y, Liu W, Ward BA, et al. The continuing increase in adenocarcinoma of the uterine cervix: A birth cohort phenomenon. Int J Epidemiol 1996;25:252–8.
6. Wharton V. Neoplasms of the cervix. In: Holland JF, Frei E, Bast RC, Kufe DW, Morton DL, Weichselbaum RR, eds. Cancer medicine. Vol II. 4th ed. Toronto: Williams and Wilkins, 1995:2227–61.
7. Sasieni P, Adams J. Changing rates of adenocarcinoma and adenosquamous carcinoma of the cervix in England. Lancet 2001;357:1490–3.[Medline]
8. Kjaer SK, Brinton LA. Adenocarcinomas of the uterine cervix: The epidemiology of an increasing problem. Epidemiol Rev 1993;15:486–98.
9. South Australian Cancer Registry. Epidemiology of cancer in South Australia. Incidence, mortality and survival, 1977–1999. Incidence and mortality, 1999. Adelaide, Australia: Openbook Publishing House, 2000:19–269.
10. Cannistra SA, Niloff JM. Cancer of the uterine cervix. N Engl J Med 1996;334:1030–8.
11. Whitney CW, Sause W, Bundy BN, Malfetano JH, Hannigan EV, Fowler WC, et al. Randomized comparison of fluorouracil plus cisplatin versus hydroxyurea as an adjunct to radiation therapy in stage IIB–IVA carcinoma of the cervix with negative para-aortic lymph nodes: A Gynecologic Oncology Group and Southwest Oncology Group study. J Clin Oncol 1999;17:1339–48.
12. Keys HM, Bundy BN, Stehman FB, Muderspach LI, Chafe WE, Suggs CL, et al. Cisplatin, radiation, and adjuvant hysterectomy compared with radiation and adjuvant hysterectomy for bulky stage IB cervical carcinoma. N Engl J Med 1999;340:1154–61.
13. Rose PG, Bundy BN, Watkins EB, Thigpen JT, Deppe G, Maiman MA, et al. Concurrent cisplatin-based radiotherapy and chemotherapy for locally advanced cervical cancer. N Engl J Med 1999;340:1144–53.
14. Morris M, Eifel PJ, Lu J, Grigsby PW, Levenback C, Stevens RE, et al. Pelvic radiation with concurrent chemotherapy compared with pelvic and para-aortic radiation for high-risk cervical cancer. N Engl J Med 1999;340: 1137–43.
15. Thomas GM. Improved treatment for cervical cancer—concurrent chemotherapy and radiotherapy. N Engl J Med 1999;340:1198–200.
16. Zhang M, Cai S, Shi D. The comparison of radiosensitivity between uterine carcinoma and adenocarcinoma. Chung Hua Fu Chan Ko Tsa Chih 1998;33:611–3.
17. Fritz A, Percy C, Jack A, Kanagaratnam S, Sobin L, Parkin DM, et al, eds. International classification of disease for oncology. 3rd ed. Geneva: World Health Organization, 2000.
18. UICC International Union Against Cancer. TNM classification of malignant tumours. 4th ed. Berlin: Springer Verlag, 1987.
19. Hahn Johnston C, Richards L. Standards of the Commission on Cancer, Volume 2. Registry operations and data standards (ROADS). Chicago: Commission on Cancer, 1998.
20. Bonett A, Roder D, Esterman A. Cancer case-survival rates for South Australia: A comparison with US rates and a preliminary investigation of time trends. Med J Aust 1988;148:556–9.[Medline]
21. StataCorp. Stata statistical software: Release 6.0. College Station, Texas: StataCorp.
22. South Australian Cancer Registry. Epidemiology of cancer in South Australia. Incidence, mortality and survival, 1977 to 1996. Incidence and mortality, 1996. Adelaide, Australia: Openbook Publishing House, 1997:11–9, 21–8.
23. Armitage P, Berry G. Statistical methods in medical research. Oxford, United Kingdom: Blackwell Scientific Publications, 1987:387, 411–6, 428–32, 436–7.
24. Hosmer DW, Lemeshow S. Applied logistic regression. New York: John Wiley and Sons, 1989:132–3.
25. South Australian Cancer Registry. Epidemiology of cancer in South Australia. Incidence mortality and survival, 1977 to 1993. Incidence and mortality 1993. Adelaide, Australia: Openbook Publishing House, 1994.
26. Zaino RJ, Ward S, Delgado G, Bundy B, Gore H, Fetter G, et al. Histological predictors of the behaviour of surgically treated stage 1B squamous cell carcinoma of the cervix. A Gynaecological Oncology Group Study. Cancer 1992;69: 1750–8.[Medline]
27. Ishikawa H, Nakanishi T, Inoue T, Kuzuya K. Prognostic factors of adenocarcinoma of the uterine cervix. Gynecol Oncol 1999;73:42–6.[Medline]
28. Costa MJ, McIlnay KR, Trelford J. Cervical carcinoma with glandular differentiation; histological evaluation predicts disease recurrence in clinical stage I or II patients. Hum Pathol 1995;26:829–37.[Medline]
29. Kristensen GB, Abeler VM, Risberg B, Trope C, Byrne M. Tumor size, depth of invasion, and grading of the invasive tumor front are the main prognostic factors in early squamous cell cervical carcinoma. Gynecol Oncol 1999;74: 245–51.[Medline]
30. Australian Health Ministers’ Advisory Council, Cervical Cancer Screening Evaluation Steering Committee. Cervical cancer screening in Australia: Options for change. Australian Institute of Health prevention program evaluation series no. 2. Canberra, Australia: Australian Government Publishing Service, 1990.
31. Mitchell H, Medley G, Gordon I, Giles G. Cervical cytology reported as negative and risk of adenocarcinoma of the cervix: No strong evidence of benefit. Br J Cancer 1995; 71:894–7.[Medline]
32. South Australian Cancer Registry. Epidemiology of cancer in South Australia. Incidence, mortality and survival, 1977 to 1997. Incidence and mortality, 1997. Adelaide, Australia: Openbook Publishing House, 1998:77–90.
This article has been cited by other articles:
|
|
 |
|
  M. Sindos, N. Pisal, T. Freeman-Wang, A. Singer, and M. Davy Cervical Cancer: Effect of Glandular Cell Type on Prognosis, Treatment, and Survival Obstet. Gynecol., June 1, 2003; 101(6): 1354 - 1355. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
