Screen-detected invasive cervical carcinoma and its clinical significance during the introduction of organized screening

Authors


Dr A Herbert, Department of Histopathology, Southampton University Hospitals Trust, Southampton, UK. Email amanda.herbert@kcl.ac.uk

Abstract

Objective  To assess the clinical significance of screen-detected cancer compared with symptomatic cervical cancer during the introduction of organized screening.

Study design  Screen-detected and symptomatic cancers were analysed according to International Federation of Obstetrics and Gynecology stage and histological type of cancer and age of women during four 3-year periods.

Setting  Southampton and South West Hampshire (SSWH).

Population  Women resident in SSWH registered with invasive cervical carcinoma between 1985 and 1996.

Methods  Screen-detected and symptomatic cancers were defined on the basis of clinical information obtained at the time of diagnosis and recorded together with date of diagnosis, age group, histology and cytology results. Data were held for analysis on an anonymous spreadsheet.

Main outcome measures  Incidence of screen-detected and symptomatic cancers per 100 000 women; association between route to diagnosis, age group, stage of cancer and screening history during four 3-year periods.

Results  Of 382 cancers, 256 were symptomatic, 78 screen-detected fully invasive and 48 screen-detected microinvasive. Incidence of symptomatic cancers fell from 13.0 to 6.4 per 100 000 total female population (P = 0.0005), and there was a trend towards screen-detected cancers as a proportion of all cancers (P = 0.002). By the latest period of the study (1994–96), the likelihood of cancers being screen-detected was strongly associated with younger age groups (P = 0.001). Screen-detected fully invasive cancers were significantly less likely (P = 0.001) to be diagnosed at stage II or above (16.6%) compared with symptomatic cancers (63.3%). Women with screen-detected microinvasive cancers were significantly more likely (P < 0.001) to have been screened within 5 years (83.3%) than screen-detected fully invasive cancers (60.2%) or symptomatic cancers (33.6%).

Conclusion  Screen-detected cancers were more likely to be diagnosed at an earlier stage, in younger women and in women who had previously been screened and became relatively more frequent as symptomatic cancers declined.

Introduction

Since the introduction of the NHS Cervical Screening Programme (NHSCSP) in 1988, the incidence of cervical cancer has fallen from 18.4 in 1988 to 8.8 in 2005,1 despite an increasing risk in women born since 1940.2,3 Before 1988, screening had been largely opportunistic in younger age groups and quality control was known to be poor.4

A national audit of invasive cervical cancer reported that cytological screening was less effective at preventing frankly invasive cancer in women younger than over the age of 40 years.5 Microinvasive carcinomas were excluded from that analysis as they were likely to be screen-detected; however, the authors found that it was not possible to identify screen-detected fully invasive cancers from routine data in centrally held records.5 A national audit is now under way to investigate all cases and stages of invasive cervical cancer,6 but the criteria for defining screen-detected cancers have yet to be established.

This study has been carried out to investigate the significance of the route to diagnosis of cervical cancer since a preliminary report of the same data set, led by one of the authors, indicated that screen-detected cancers in general (microinvasive and fully invasive) were more likely to be stage I than symptomatic cancers and were associated with a better 5-year survival.7 This finding applied to adenocarcinomas as well as squamous cell carcinomas.8 There are few studies on screen-detected cervical cancers as such, although cancers in previously screened women have been shown to be diagnosed at an earlier stage than those in unscreened women.9,10

We revisited the data collected in the Southampton and South West Hampshire (SSWH) study because we believe that it may be relevant to the national audit of invasive cervical cancer that is now under way. We intend to use these data as a baseline for the audits of cervical cancers that are currently in progress in Southampton and Guy’s & St Thomas’ NHS Foundation Trust as part of the NHSCSP audit. Our findings may also be relevant to programmes currently being introduced in Europe in populations like SSWH that have already been partially screened opportunistically.

In this study, registrations of invasive cancer in four consecutive 3-year periods are described (from 1985–87 to 1994–96), which represent screening before, during and after the implementation of the NHSCSP in SSWH. The incidence of cervical cancer in SSWH declined more or less in parallel with that of England as a whole during the period of the study.8,11 We have compared symptomatic, screen-detected fully invasive and screen-detected microinvasive cancers with respect to the 3-year period of the study, age group, stage, histological diagnosis and screening history. The complete details of screening histories are presented in further detail in a second paper.12

Population, subjects and methods

Study area and population

The study population comprised the female population in SSWH. That area was estimated by the Office for Population Censuses and Surveys to have an average total female population of 214 731 in 1986, 217 631 in 1989, 221 026 in 1992 and 224 249 in 1995, the mid-points of the four periods of the study.11 Women aged between 20 and 64 years were eligible for cervical screening, and a centralized call and recall system was implemented during the second period (1988–90), although screening uptake had increased steadily during the previous decade (unpublished observations).

Data collection

Invasive cervical cancers in women residents of SSWH were retrieved from the computerized histology records at Southampton General Hospital, local cancer registry data, Wessex Radiotherapy and Oncology Units and requested from other NHS hospitals in adjacent districts. All women were validated and checked for duplication and miscoding. The details of data collation have been outlined in a previous paper.7

Study group

The study group composed of 382 women residents of SSWH in whom invasive cervical cancer was diagnosed between 1985 and 1996. Cancers were grouped according to the age of the woman at the time of cancer diagnosis into four age bands: 20–34, 35–39, 50–64 and 65 years and older. Women 65 years and older were not invited for screening. For purposes of analysis, the cancers were grouped into four consecutive 3-year periods depending upon the time of cancer diagnosis: 1985–87, 1988–90, 1991–93 and 1994–96.

Histological type and stage of cancer

The histological type of carcinoma was recorded as squamous cell carcinoma, adenocarcinoma, adenosquamous carcinoma, undifferentiated carcinoma or type unknown. Stage was recorded for all except three cancers, which were only known to be ‘invasive’. Stage IA1 squamous carcinomas were classified as those with a depth of invasion less than 3 mm and lateral extent less than 7 mm using the revised International Federation of Obstetrics and Gynecology (FIGO) classification. Until 1996, stage IA2 cancers were not identified separately and any women in the study were included with IB cancers. Although there are no accepted criteria for stage IA adenocarcinoma, the same criteria for depth and extent of invasion were used.

Route to diagnosis

Women with invasive cancer were categorized as ‘screen-detected’ if the cancer was found as a direct result of an abnormal smear reported during a routine screening test and the woman was asymptomatic. This included microinvasive and fully invasive cancers. ‘Symptomatic’ cancers were found as a result of investigation of symptoms. Route to diagnosis, which was initially assessed from clinical data on cytology request forms, histology reports and referral letters from the general practitioners, was confirmed, in cases of doubt, by contacting the general practitioner by letter. Two women were difficult to classify, but as they consistently claimed to be symptom free at the time of diagnosis, they were included in the screen-detected group. The diagnosis was made as a direct result of an invitation for a cervical smear in these women. One was called for screening, but a smear was not taken because of vaginismus: a mass was noticed on examination and cervical cancer subsequently diagnosed. One was found to have a procidentia when called for a routine smear, which was negative. Hysterectomy revealed microinvasive carcinoma.

In 17 women with screen-detected cancers, minor gynaecological symptoms like ‘heavy periods’ or ‘discharge’ were noted on direct questioning after the diagnosis had been made. Some degree of cervical abnormality, ranging from ‘erosion’ or ‘cervicitis’ to clinical suspicion of cancer, was noted in 24 women at the time of smear taking.

Incidence of cancer

Incidence has been calculated for each 3-year period by calculating one-third of the total number of cancers per 100 000 women of all ages in the population at the mid-point of each time period. Similarly, incidence has been calculated for symptomatic, screen-detected fully invasive and screen-detected microinvasive cancers, respectively.

Screening histories

Women were classified as no previous cytology record, not screened within 5.5 years prior to cancer diagnosis or screened within 5.5 years prior to cancer diagnosis (described in this study as interval cancers). That interval was chosen because, until 1993, screening coverage was calculated at an interval of 5.5 years to take account of delays between invitation and reporting.13 The screening history excluded tests carried out within 6 months of diagnosis, which were either abnormal smears on the basis of which diagnosis was made (screen-detected cases) or smears taken during investigation of symptoms (symptomatic cases).

Statistical analysis

Chi-square test was used as a test of inference to compare difference between proportions. Chi-square for trend was used to test significance of trend of proportions across study years. Statcalc calculator in the EPI INFO 6 software was used for calculation of these tests.

Results

Incidence and trends according to route to diagnosis

Of the 382 carcinomas, 256 (67.0%) were symptomatic, 78 (20.4%) screen-detected fully invasive and 48 (12.6%) screen-detected microinvasive carcinomas (Table 1). There were 108 carcinomas registered in 1985–87, 102 in 1988–90, 102 in 1991–93 and 70 in 1994–96. Incidence of all cancers per 100 000 women fell from 16.8 in 1985–87 to 10.4 in 1994–96, but this fall took place almost entirely in the fourth period of the study (Table 1 and Figure 1). Incidence of symptomatic cancers fell steadily from 13.0 to 6.4 per 100 000, which was highly significant (P = 0.0005). As the numbers of symptomatic cancers fell, there was a significantly increased trend in the proportion of screen-detected cancers compared with all cancers (chi-square for trend 9.046, P = 0.002) during the study period. This trend was significant for screen-detected microinvasive cancers (chi-square for trend 21.198, P < 0.001) but not for screen-detected fully invasive cancers alone (chi-square for trend 0.077, P = 0.781).

Table 1.  Incidence per 100 000 total female population of symptomatic and screen-detected cancers according to 3-year time period and numbers of women according to age band
 1985–871988–901991–931994–96
  1. (n) = total 1985–87 to 1994–96.

Total female population at mid-point of 3-year period214 731217 631221 026224 249
Symptomatic (256)84715843
20–34 years (30)101073
35–49 years (67)26141611
50–64 years (55)23121010
65+ years (104)25352519
Incidence per 100 000, all ages13.010.98.76.4
Screen-detected fully invasive (78)21242013
20–34 years (13)3334
35–49 years (33)13785
50–64 years (25)41272
65+ years (7)1222
Incidence per 100 000, all ages3.33.73.01.9
Screen-detected microinvasive (48)372414
20–34 years (9)1125
35–49 years (27)14166
50–64 years (10)1153
65+ years (2)0110
Incidence per 100 000, all ages0.51.13.62.1
All cancers (382)10810210270
20–34 years (52)14141212
35–49 years (127)40254022
50–64 years (90)28252215
65+ years (113)26382821
Incidence per 100 000, all ages16.815.615.410.4
Figure 1.

Trends in incidence of symptomatic and screen-detected cervical cancer (1985–1996).

Route to diagnosis according to age band

Fifty-two cancers were seen in women aged 20–34 years, 127 aged 35–49 years, 90 aged 50–64 years and 113 aged 65 years and older. There was no fall in numbers of cervical cancers in women aged 20–34 (Table 1), but there was a significant trend towards screen-detected cancers in that age band (chi-square for trend 5.835, P = 0.015). Women aged 35–49 years, who comprised the largest group overall and 52.8% (142/269) in age bands eligible for screening, showed an increase in screen-detected microinvasive cancers in the third period of the study (Figure 1). In women aged 50–64 years, there was a steady decline in the total numbers, but an increase in screen-detected fully invasive cancers was noted in the second period of the study. In women above the age of 65 years, 92% of cancers were symptomatic, and this did not change significantly throughout the time period of the study. By the last period of the study (1994–96), there was a strong association between the likelihood of cancers being screen-detected and the younger groups. Screen-detected cancers composed 75.0% (9/12) of cancers in women aged 20–34 years, 50.0% (11/22) in women aged 35–49 years, 33.0% (5/15) in women aged 50–64 years and 9.5% (2/21) in women aged 65 years and older (P < 0.001).

Route to diagnosis and FIGO stage

Of the 382 women with carcinomas, 54 were stage IA1 (microinvasive), 153 stage IB, 84 stage II, 62 stage III and 26 stage IV. The stage was unknown in three women, although they were known to be fully invasive cancers (Table 2). Cases of stage II and above were symptomatic in 160 (83.3%) of 172 women. Although similar proportions of stage IB cases were symptomatic (57.5%) and screen-detected fully invasive cancers (42.4%), significantly more symptomatic cancers were stage II or above (162/256; 63.3%) compared with screen-detected fully invasive cancers (13/78; 16.7%) (P < 0.001).

Table 2.  Route to diagnosis of invasive cervical cancer according to FIGO stage, histological type and screening history
 Total cancersSymptomaticScreen-detected fully invasiveScreen-detected microinvasive
FIGO stage
IA546048
IB15388650
II8474100
III626110
IV262510
Invasive3210
Total382 (100)256 (67.0)78 (20.4)48 (12.6)
Histological type
Squamous cell carcinoma/carcinoma, not otherwise specified2892005237
Adeno/adenosquamous carcinoma93562611
Total382 (100)256 (67.0)78 (20.4)48 (12.6)
Screening history
No previous cytology148132133
Screened more than 5.5 years before diagnosis6138185
Screened within 5.5 years173864740
Total382 (100)256 (67.0)78 (20.4)48 (12.6)

Route to diagnosis and histological type

There were 281 women with squamous carcinomas and 78 adenocarcinomas. Eight carcinomas of either undifferentiated or unknown histological type were included with the squamous carcinomas for purposes of analysis, while 15 adenosquamous carcinomas were combined with the adenocarcinomas. Of squamous carcinomas, 69.2% were symptomatic, while 30.7% were screen-detected. While 60.2% of adenocarcinomas were symptomatic, 39.8% were screen-detected. There was no significant difference in the percentages of adenocarcinomas among symptomatic cancers (21.9%), screen-detected fully invasive cancers (33.3%) and screen-detected microinvasive cancers (22.9%).

Route to diagnosis and screening histories

There was no record of previous cytology in 148 women (38.7%), among whom 132 (89.2%) had symptomatic cancers. Sixty-one women (15.9%) had been screened more than 5.5 years before the diagnosis of cancer, and 173 (45.2%) had been screened in the preceding 5.5 years (excluding referral or symptomatic tests within 6 months of cancer diagnosis).

Although women who had been screened within 5.5 years were equally likely to have screen-detected (49.7%) as symptomatic (49.3%) cancers, there was a strong association between the route to diagnosis and the likelihood of the woman having been screened within 5.5 years; thus, 86 of 256 (33.6%) of symptomatic cancers, 47 of 78 (60.2%) of the screen-detected fully invasive cancers and 40 of 48 (83.3%) of screen-detected microinvasive cancers were in women who had been previously screened (interval cancers) (P < 0.001).

Discussion

This study shows the clinical relevance of route to diagnosis of cervical cancers in several ways. First, screen-detected cancers represented an increasing proportion of cancers, while the incidence of symptomatic cancers steadily declined. Screen-detected cancers were also more likely to be diagnosed at an early stage even when fully invasive. In addition, as organised screening became established, screen-detected cancers were increasingly associated with younger age groups and were more likely to be seen in previously screened women.

Route to diagnosis is not obvious from the central screening records used in England, which is in agreement with Sasieni et al.5 Although it might be feasible to decide which women were screen-detected from the date of invitation, many women have tests ‘opportunistically’ either before or after the due date. For instance, in SSWH, 74.7% of eligible women were screened within 3.5 years, by their own request or invited by their general practitioners, even though the official recall interval at that time was 5 years.14 The NHSCSP defines screen-detected cancers as ‘cancers which were detected after a diagnostic process that began with a cytology test taken up to three months before the test due date or up to six months after the test was due’.6 However, apart from women who chose to have tests early or later, we feel that these criteria should not exclude ‘interval cancers’ or ‘lapsed attenders’, many of which, in our experience, are screen-detected. Especially with the advent of multidisciplinary meetings, it is usually relatively easy to obtain information and segregate screen-detected from symptomatic cancers. However, it might be better to separate ‘screen-detected’ from ‘not screen-detected’ cases, as the latter could include doubtful cases; that was our intention at the time of the audit, but our study shows that it can occasionally be difficult to decide the route to diagnosis.

There are very few reports in the existing literature14–16 that have compared screen-detected cervical cancers as such with symptomatic cancers. However, Sigurdsson et al.9 reported that women who had attended the cervical screening programme fared significantly better than those who had never attended. Young women had a significantly better prognosis than older women. The prognostic effect of screening was mainly attributed to the more favourable distribution of early stages and younger age at diagnosis among the screened women. After all the analyzed parameters had been adjusted, the non-attenders still had poorer prognosis. Pretorius et al.10 found that women who had abnormal cervical smears had better disease-free survival than those who presented with symptoms of abnormal vaginal bleeding or pain. Our study shows that there was a significant trend towards screen-detected cancers across the period of the study, even in the youngest women aged 20–34 years. In the last period of the study, as many as 75% of cancers in that age band were screen-detected. Benedet et al.17 have said that age, per se, represents a valuable prognostic factor but is an independent prognostic factor only for women with stage I cancers. We suggest that the strong association between screen-detected cancers and younger age groups accounts for the better prognosis reported in young women with cervical cancer. We also believe that detection of microinvasive cancer is a positive outcome of screening that should be taken into account, especially in the current climate of necessarily conservative management of young women. That these cancers are increasingly found in young women and are seldom found on the first cytology test may be an argument against delaying the first screening invitation to age 25 years.18

Our study showed that cancers were most frequent in women aged 30–49 years, which is consistent with an increased risk in women born since 1940.2,3 Women in that age band with cancers diagnosed between 1985 and 1996 would have been born between 1966 and 1936. That age band held 16 of 24 (75%) screen-detected microinvasive cancers in the third period of the study when there was a peak in the incidence of that category. We found an interesting association between screen-detected cancers and their likelihood of being interval cancers. Interval cancers were progressively more frequent with screen-detected microinvasive cancers compared with screen-detected fully invasive cancers and with the latter group compared with symptomatic cancers. Thus, the peak in screen-detected microinvasive cancers in the third period of our study would have involved a high proportion of interval cancers. This is considered in more detail in a separate study,12 which examines the reasons why cancers develop in previously screened women.

We have demonstrated the clinical importance of cervical cancers being screen-detected rather than being symptomatic, especially once organised screening is established. These cancers are more likely to be diagnosed at an earlier stage in younger women and in women who have previously been screened.

Disclosure of interest

None of the authors have interests that conflict with the study.

Contribution to authorship

A.H. conceived and carried out the initial 12-year audit while employed as Lead Cytopathologist at Southampton University Hospitals Trust (SUHT). Anshu while a Commonwealth Fellow at GSTFT helped write the paper and prepared the tables and graphs from the spreadsheet. S.S.G. carried out the statistical analysis. M.G. led the Southampton cytology laboratory throughout the time of the study and collaborated in collecting the information about screening at that time. N.S. is now the Lead Cytopathologist at SUHT and helped with this study and was co-author of the previous study of the same data set when it was transferred to an anonymous spreadsheet.

Details of ethics approval

Approval for this audit was gained at the time the project was started in January 1992 when a 3-year grant was successfully applied for and Catherine Breen appointed as a research assistant.

Funding

Wessex Cancer Trust and Cansearch, which was the campaigning arm of the British Society for Clinical Cytology, jointly provided the grant that enabled the original project to be set up and carried out. Guy’s & St Thomas’ Charity provided a grant for A.H. to carry out the final analysis of the data set as a baseline for a GSTT cervical cancer audit.

Acknowledgements

We thank Catherine Breen, the research assistant who set up the original audit database at Southampton General Hospital and Jennifer Smith, Director at that time of Wessex Cancer Intelligence Unit who provided population and coverage data for a previous study of the full data set referred to in the text.

Ancillary