SEARCH

SEARCH BY CITATION

Keywords:

  • adenocarcinoma;
  • protection;
  • cervical cytology;
  • incidence

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

BACKGROUND

The incidence of adenocarcinoma of the cervix did not decline in Western countries in the 1970s and 1980s despite the availability and use of cervical cytologic screening. The impact of improved endocervical sampling and better recognition of the cytologic precursors to adenocarcinoma during the 1990s in reducing the risk of adenocarcinoma is currently unknown.

METHODS

Using records from a statewide registry, the authors compared the screening histories of 160 women with adenocarcinoma (33 microinvasive cases and 127 invasive cases) who were diagnosed between 1995 and 2001 with the screening histories of 640 control women in a matched case–control study.

RESULTS

A decreased risk of invasive adenocarcinoma was associated with a recent negative Papanicolaou (Pap) smear and at least one smear with an endocervical component since 1994. An increased risk of invasive adenocarcinoma was associated with a history of previous cervical abnormalities and greater than 5 years between Pap smears. There was no significant difference between cases and controls with regard to the number of negative Pap smears. There was little apparent difference in the screening histories of patients with microinvasive tumors and the controls, with the control women appearing to have no greater relative protection than the cases. With biennial screenings, the authors estimate that a 46% reduction in the cumulative incidence of invasive adenocarcinoma could be achieved, and with annual screening, a 65% reduction could be achieved.

CONCLUSIONS

The incidence of cervical adenocarcinoma should begin to decline in the current decade among screened women. Cancer (Cancer Cytopathol) 2003;99:336–41. © 2003 American Cancer Society.

Women who are diagnosed with squamous cervical carcinoma appear to have a reduced frequency of screening compared with healthy, age-matched control women, according to recent studies.1–3 The conclusions drawn from these studies is that if women with cervical carcinoma had been screened at the same frequency as control women, their tumors would have been prevented. It has been estimated that annual screening can prevent 93.5% of cases of squamous carcinoma of the cervix, biennial screening can prevent 92.5% of cases, and triennial screening can prevent 90.8% of cases.1

To our knowledge, four studies published to date have explored the relative protection conferred by Papanicolaou (Pap) smear screening against adenocarcinomas of the cervix diagnosed in the 1980s.4–7 To our knowledge, none of the studies demonstrated a substantial benefit. An increase in the absolute rate of adenocarcinoma among women ages 25 to 49 years was documented in many countries between 1973 and 1991.8 These two observations indicate little benefit from cytologic screening as performed in the 1970s and 1980s in preventing adenocarcinoma.

Limitations of the existing studies exploring relative protection against adenocarcinoma include the inaccuracies associated with self-reported screening histories; a relatively small number of cases such that limited benefit may not have been detected even if it was present; not distinguishing between normal and abnormal Pap smears; and not using the endocervical component of the smears as a variable in the analysis.4–7 Furthermore, the relevance of the existing studies, which mainly recruited cases from the 1980s, to cervical screening performed in the 1990s is questionable. During the last decade, cytologists have given greater emphasis to predicting the precursor lesion, adenocarcinoma in situ. In addition, improved sampling of the transformation zone (from which the majority of adenocarcinomas are believed to arise) was a strong focus for quality improvement in cervical screening during the 1990s.9

We undertook a case–control study to explore the relative protection conferred by negative Pap smears against adenocarcinoma of the cervix, using cases diagnosed between 1995–2001. This timeframe allows for the early evaluation of the benefit from cytologic screening in the 1990s.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

The sampling frame was the records of the Victorian Cervical Cytology Registry. This is a statewide registry for a region of Australia in which 1.5 million women are eligible for cervical screening. The registry compiles record-linked summaries of cytology and histology for individual women, drawing on information from all laboratories reporting cervical cytology or histology for women residents in the state. The registry operates under the Cancer Act and has functioned since late 1989. However, because the registry was built on the records of the public laboratory that had reported nearly all Pap smears from the inception of cervical screening in 1965 and had run a comprehensive followup program, its data approach a lifetime screening record for many women.

Cases were comprised of all women with an incident histologic diagnosis of adenocarcinoma of the cervix between January 1, 1995 and December 31, 2001 who were younger than 70 years at the time of the carcinoma diagnosis and who had at least 1 negative cervical smear prior to the exit date. The exit date for a case and her matched controls was 6 months prior to the carcinoma diagnosis date. Both microinvasive and fully invasive cases were included. Women with adenosquamous malignancy were excluded.

Controls were selected from a random sample of the registry database. The selection of the random women was independent of the number and timing of Pap smears for each woman. The screening history of each potential control woman was not known at the time of selection for the study. Four control women were selected for each case, and were age-matched to within 2 years of the case. This provided 90% power with which to detect a relative protection of 2 within 2 years of a negative smear, assuming 130 available cases (P=0.05), a prevalence of exposure among the controls of 55%, and a correlation coefficient (phi) between cases and their matched controls of 0.2.7, 10 Women with a history of adenocarcinoma of the cervix were ineligible to be controls.

The approach to selecting controls differed for microinvasive and invasive cases, with the former regarded as screen detected and the latter as symptomatic at diagnosis.11, 12 For invasive cases, control women were required to have had at least one negative cervical smear prior to the exit date, and, at the time of the cancer diagnosis in the matched case, evidence of being eligible to be diagnosed with adenocarcinoma of the cervix. This was comprised of evidence that the woman had a cervix at the time of the matched cancer case, namely cervical cytology or histology during the 2 years before the exit date or at any time after the exit date, or a hysterectomy after the exit date. For microinvasive cases, control women were required to have had a negative cervical smear within 6 months before or after the date of the carcinoma diagnosis in the matched case (this smear defined the woman as a control and was not counted in the number of preceding smears) and at least 1 negative smear prior to the exit date.

The number, timing, and endocervical status of negative cervical cytology reports issued for cases and controls was determined. Negative cytology was defined as a report of no abnormal cells or a report of minor reactive or inflammatory change. Neither cases nor controls were excluded because of a history of cervical abnormality, defined as abnormal cytology (atypia or worse) or histology occurring greater than 18 months before the exit date. The median time interval between negative cytology reports was determined for each case and control.

The endocervical status of cervical cytology specimens has been reported in Victoria since 1987. Three time periods were defined (1987–1989, 1990–1993, and 1994–2001), corresponding to periods during which the average proportion of cervical smears with an endocervical component varied (56%, 76%, and 83%, respectively). Australia formally adopted the Bethesda definition of an endocervical component in 1996.13

The Victorian Cervical Cytology Registry does not record whether the cervical cytology was a conventional smear or a liquid-based preparation. Most of the cytology in this study would have been conventional smears. Time trends in the use of liquid-based cytology are not available for the state of Victoria, but communication from laboratories suggests that current usage is not more than 10%.

Conditional logistic regression was used to estimate the relative protection (inverse of the odds ratio) of select variables. A stepwise technique was used, with variables selected for inclusion in the model on the basis of a significant change in the log likelihood. Interactions were investigated by fitting interaction terms. Subgroup analyses were undertaken to investigate the relative protection in 3 age groups (age younger than 35 years, ages 35 to younger than 50 years, and age 50 years or older). The percentage reduction in the cumulative rate of adenocarcinoma among women ages 20 to 70 years who were screened at different time intervals was calculated.1 All analyses were conducted using Stata software (Stata Corporation, College Station, TX).14

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

A total of 160 cases (33 microinvasive cases and 127 invasive cases) were eligible for the study, of whom 32 women (20%) were younger than 35 years of age at the time of carcinoma diagnosis, 93 women (58%) were ages 35 to younger than 50 years, and 35 women (22%) were age 50 years or older.

A history of cervical abnormality was known for 27% of cases and 15% of controls. Controls had a greater number of previous negative smears (5.28 vs. 4.53; P < 0.01) with a range of 1–23 previous negative smears for the whole sample. Twenty-nine cases (18%) and 74 controls (12%) had had only 1 negative smear before the exit date. There was no difference in the proportion of previous negative smears with an endocervical component between cases and controls (57% for cases and 60% for controls). However, controls had a significantly greater number of smears with an endocervical component during the years 1994–2001 (1.40 vs. 1.13; P < 0.01), the closest time period to the exit date.

Conditional logistic regression showed that less than 1 year since the last negative smear and having 1 or more smears with an endocervical component between 1994–2001 were protective against invasive adenocarcinoma (Table 1). Both these variables offered an approximately threefold increase in relative protection against invasive cervical adenocarcinoma.

Table 1. Relative Protection Against Invasive Adenocarcinoma for All Age Groups
Risk factorNo. of casesNo. of controlsUnadjusted relative protection95% CIAdjusted relative protectiona95% CI
  • 95% CI: 95% confidence interval.

  • a

    Adjusted for all other variables.

  • b

    Excludes the most recent negative smear.

Time since last negative smear (yrs)      
 3+32511.00 1.00 
 2–< 319501.710.85–3.441.190.55–2.60
 1–< 2431592.271.30–3.971.310.67–2.55
 <1332484.752.66–8.492.751.36–5.56
No. of negative smears      
 123611.00   
 222591.060.53–2.11  
 320801.560.77–3.14  
 410532.100.92–4.81  
 59522.290.96–5.43  
 6+432031.901.02–3.52  
History of cervical abnormality      
 No914341.00 1.00 
 Yes36740.420.27–0.670.260.15–0.45
Median time between previous negative smears (yrs)b      
 0–3713441.00 1.00 
 > 3–515741.000.55–1.821.340.69–2.62
 > 5, or no previous smears41900.450.29–0.710.480.27–0.85
No. of negative smears with endocervical component since 1994      
 0541081.00 1.00 
 1351883.381.99–5.762.971.60–5.54
 2+382124.412.47–7.903.061.39–6.75
No. of negative smears with endocervical component 1990–1993      
 0471661.00   
 1441831.180.74–1.87  
 2+361591.260.77–2.05  
No. of negative smears with endocervical component before 1990      
 01083991.00   
 119861.190.69–2.05  
 2+3232.030.59–6.91  

An increased risk of invasive adenocarcinoma was associated with a history of cervical abnormality and with having more than 5 years between smears (Table 1). The median time between negative smears and the number of negative smears were highly negatively correlated (P < 0.01), with time between negative smears being a stronger predictor of invasive adenocarcinoma. A test for a trend in increasing relative protection by time since last negative smear was significant (P < 0.01).

The pattern of relative protection varied among the three age groups. Among women age younger than 35 years, a test for a trend in increasing protection by time since last negative smear was significant (P < 0.01), but no clear correlation with time since last negative smear was evident for the other 2 age groups. Women in the older 2 age groups had a significantly increased risk of invasive adenocarcinoma if they had a history of cervical abnormality, more than 5 years between smears, or only 1 negative smear prior to the exit date. Having at least one negative smear with an endocervical component since 1994 was associated with a decreased risk.

There was little apparent difference between microinvasive cases and controls in the variables studied, a finding that is consistent with the screening histories of the two groups of women being similar and control women having no greater relative protection than cases (Table 2).

Table 2. Relative Protection against Microinvasive Adenocarcinoma for All Age Groups
Risk factorNo. of casesNo. of controlsUnadjusted relative protection95% CIAdjusted relative protectiona95% CI
  • 95% CI: 95% confidence interval.

  • a

    Adjusted for all other variables.

  • b

    Excludes the most recent negative smear.

Time since last negative smear (yrs)      
 3+12211.00 1.00 
 1 to < 314893.931.51–10.244.871.41–16.76
 < 17221.740.56–5.372.540.50–12.93
No. of negative smears      
 16131.00   
 2 to 413551.930.62–6.01  
 5+ or more14642.140.67–6.82  
History of cervical abnormality      
 No261081.00 1.00 
 Yes7240.830.32–2.110.820.29–2.35
Median time between previous negative smearsb (yrs)      
 0–322991.00 1.00 
 > 3–53110.800.21–3.131.040.25–4.32
 > 5, or no previous smears8220.620.25–1.550.780.27–2.26
No. of negative smears with endocervical component since 1994      
 015521.00 1.00 
 1+18801.400.57–3.430.620.17–2.22

Table 3 shows estimates of the percentage reduction in the cumulative incidence of invasive adenocarcinoma among women ages 20–70 years with different frequencies of screening. With annual screening, we estimate a 65% reduction in the cumulative incidence of adenocarcinoma could be achieved; with biennial screening, we estimate a 46% reduction.

Table 3. Percent Reduction in the Cumulative Incidence of Invasive Adenocarcinoma in Women Ages 20–70 Years with Different Frequencies of Screening
Interval between screenings (yr)Relative protection95% CIReduction in cumulative incidence (%)
  • 95% CI: 95% confidence interval.

  • a

    The information for an interval of 1 to 6 years between screenings is based on those who have had 2 or more negative smear reports.

  • b

    Includes those who have had only one negative smear report.

1a2.851.56–5.2365
2a1.380.77–2.4946
3a1.120.54–2.3334
4a1.530.52–4.5134
5a0.890.21–3.7825
6a0.810.19–3.4217
7b1.00  

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

The results of the current study indicate that cervical screening offered some protection against invasive adenocarcinoma diagnosed between 1995 and 2001. We were encouraged that our most optimistic estimates of relative protection were in women aged younger than 35 years, given that the incidence rate is very low before 25 years of age, has a sharp increase between 25–35 years of age, and declines thereafter, and that a year-of-birth cohort effect has been described, with women born around 1955 having 3 times the risk experienced by women born around 1935.8

We found little difference in the screening history between microinvasive cases and their controls; these findings are compatible with microinvasive adenocarcinoma being a screen-detected disease.11, 12

The strengths of the current study include its population base, a confirmed screening history for both cases and controls, and the ability to include endocervical status of the negative smears as a variable in the analysis. On a more cautious note, our study was small (160 cases), particularly compared with the International Agency for Research on Cancer (IARC) study of relative protection against squamous carcinoma that had 1381 cases of microinvasive and invasive carcinoma.1 Given the rarity of adenocarcinoma, a multicenter study will be needed to confirm and give greater precision to our findings.1

Recent time trends in the incidence of adenocarcinoma of the cervix in Australia lend support to our finding that cytologic screening as performed in the 1990s offered some protection. The age-standardized incidence averaged 2.3 cases per 100,000 women during 1987–1989.15 By 1997–1999, the average age-standardized incidence had fallen to 1.6 cases per 100,000 women, a 30% decrease.16

Ultimately, the best evidence of the protective effect of regular participation in screening will be demonstrated when the incidence rates for invasive adenocarcinoma decline in a majority of countries in which screening is widespread. In monitoring the incidence rates, it may be important to exclude microinvasive adenocarcinoma because, if this is a screen-detected disease as the current study findings suggest, a paradoxical increase in the adenocarcinoma rates may be evident in the short to medium term.

Finally, adenocarcinoma of the cervix is rare. We suggest that screening policies should not be based on preventing this infrequent type of cervical carcinoma, but rather they should focus predominantly on reducing the burden of squamous cervical carcinoma. Nevertheless, any benefit from cytologic screening in reducing the incidence of adenocarcinoma will be welcomed by women and health practitioners alike.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES
  • 1
    International Agency for Research on Cancer (IARC). Working Group on Evaluation of Cervical Cancer Screening Programmes. Screening for squamous cervical cancer: duration of low risk after negative results of cervical cytology and its implication for screening policies. BMJ. 1986; 293: 659664.
  • 2
    Parazzini F, Negri E, La Vecchia C, Bocciolone L. Screening practices and invasive cervical cancer risk in different age strata. Gynecol Oncol. 1990; 38: 7680.
  • 3
    Sasieni PD, Cuzick J, Lynch-Farmery E, and The National Co-ordinating Network for Cervical Screening Working Group. Estimating the efficacy of screening by auditing smear histories on women with and without cervical cancer. Br J Cancer. 1996; 73: 10011005.
  • 4
    Brinton LA, Tashima KT, Lehamn HF, et al. Epidemiology of cervical cancer by cell type. Cancer Res. 1987; 47: 17061711.
  • 5
    Herrero R, Brinton LA, Reeves WC, et al. Screening for cervical cancer in Latin America: a case-control study. Int J Epidemiol. 1992; 21: 10501056.
  • 6
    Makino H, Sato S, Yajima A, Komatsu S, Fukao A. Evaluation of the effectiveness of cervical cancer screening: a case-control study in Miyagi, Japan. Tohoku J Exp Med. 1995; 175: 171178.
  • 7
    Mitchell H, Medley G, Giles G. Cervical cytology reported as negative and risk of adenocarcinoma of the cervix: no strong evidence of benefit. Br J Cancer. 1995; 71: 894897.
  • 8
    Vizcaino AP, Moreno V, Bosch FX, Munoz N, Barros-Dios XM, Parkin DM. International trends in the incidence of cervical cancer: I. adenocarcinoma and adenosquamous cell carcinomas. Int J Cancer. 1998; 75: 536545.
  • 9
    Lee KR, Flynn CE. Early invasive adenocarcinoma of the cervix. Cancer. 2000; 89: 10481055.
  • 10
    Dupont WD. Power calculations for matched case-control studies. Biometrics. 1988; 44: 11571168.
  • 11
    Sasco AJ, Day NE, Walter SD. Case-control studies for the evaluation of screening. J Chronic Dis. 1986; 39: 399405.
  • 12
    Moss SM. Case-control studies of screening. Int J Epidemiol. 1991; 20: 16.
  • 13
    Kurman R, Solomon D. The Bethesda system for reporting cervical/vaginal cytologic diagnoses. New York: Springer-Verlag, 1994: 6.
  • 14
    StatCorp. Stata statistical software. Release 7.0. College Station, TX: Stata Corporation, 2001.
  • 15
    Australian Institute of Health and Welfare (AIHW). Cervical Screening in Australia 1998–1999. AIHW Cat. No. 11. Canberra: Australian Institute of Health and Welfare, 2002: 63.
  • 16
    Australian Institute of Health and Welfare (AIHW). Cervical Screening in Australia 1999–2000. Canberra: Australian Institute of Health and Welfare, 2003: 61. Available from URL: http://www.aihw.gov.au/publications/index.cfm?type=detail&id=8478 [accessed May 9, 2003].