Long-term mortality in women treated for cervical intraepithelial neoplasia

Authors


Dr M. Jakobsson, Department of Obstetrics and Gynaecology, University Hospital, Helsinki, Finland. Email maija.jakobsson@fimnet.fi

Abstract

Objective  The objective of this study was to study whether women surgically treated for cervical intraepithelial neoplasia (CIN) have increased mortality later in life. We also wanted to study whether pregnancy beyond 22 weeks post-treatment affects the risk.

Design  Register-based retrospective cohort study from Finland.

Setting  National data of the Hospital Discharge Register and the Cause-of-Death Register during 1986–2003.

Population  A total of 25 827 women who had surgical treatment for CIN during 1986–2003.

Methods  We calculated standardised mortality ratios (SMRs) by dividing the numbers of observed deaths (until 31 December 2006) by the numbers of expected deaths.

Main outcome measures  SMRs for different causes-of-death groups.

Results  The overall mortality increased by 17% after treatment for CIN, including increased risk of dying from all diseases and medical conditions (SMR 1.13, 95% CI 1.01–1.26), cancers (SMR 1.09, 95% CI 0.91–1.27) and injury deaths (SMR 1.31, 95% CI 1.03–1.58). As expected, the mortality from cervical cancer was high (SMR 7.69, 95% CI 4.23–11.15). Women who had delivered post-treatment tended to have decreased overall mortality (SMR 0.78, 95% CI 0.52–1.04) and decreased disease mortality (SMR 0.63, 95% CI 0.37–0.90). However, the mortality rate was significantly increased for women who had subsequent preterm delivery (SMR 2.51, 95% CI 1.24–3.78). In this subgroup, there was a tendency of increased mortality from diseases of the circulatory system, alcohol-related causes and injury deaths.

Conclusions  Mortality rate was increased after surgical treatment for CIN. However, women who had delivered post-treatment had decreased overall disease mortality rate. Subsequent preterm delivery may be a risk marker for increased long-term mortality.

Introduction

We have previously studied a cohort of women treated for cervical intraepithelial neoplasia (CIN) and demonstrated that any cervical surgical treatment for CIN is associated with preterm delivery.1,2 There is strong evidence that women treated for CIN3 are at increased risk of developing invasive cervical, vaginal or anal cancer.3,4 Increased risk of smoking-related cancers has also been demonstrated.5 Women with CIN are generally young, and CIN lesions are caused by sexually transmitted high-risk human papillomavirus (HPV) types. Risk-taking behaviour also increases the risk for other sexually transmitted infections. Thus, our hypothesis was that these factors may predispose to increased morbidity and mortality.

The link between history of CIN, delivery and mortality for other causes has not been systematically studied. In general, women with recent delivery have lower mortality rates.6,7 However, women with preterm delivery are at increased risk for cardiovascular diseases.8–11

We had an opportunity to use Finnish register data to find out whether women treated for CIN have increased mortality later in life. We also studied mortality among women with subsequent term or preterm delivery.

Materials and methods

The study population consisted of 25 827 women of reproductive age (15–49 years) who had been surgically treated for CIN during 1986–2003. These women were identified from the Hospital Discharge Register (HDR), maintained by STAKES (National Research and Development Centre for Welfare and Health). The HDR collects information on all inpatient episodes in public and private healthcare facilities and since 1998 also all outpatient episodes in public hospitals. Day-surgical outpatient procedures have been recorded since 1994. The information on treated women was linked using unique personal identification numbers to the 1986–2006 Cause-of-Death statistics gathered by the Statistics Finland to identify all deaths. Deaths had been classified according to the ICD (ICD-8 1969–86, ICD-9 1987–95 and ICD-10 1996–2006). Information on subsequent deliveries was received from the Medical Birth Register. We had information on all live births and stillbirths (gestational age ≥ 22 weeks) since 1987. The Finnish health registers cover the total population, and the quality of data is excellent.12 The data linkages between registries were performed after the register-keeping organisations of STAKES, and Statistics Finland had given the authorisation required by legislation.

Women who had surgical treatment for CIN were divided into three categories using surgical procedure codes based on the national classification of the Finnish Hospital League (1986–96) or the Finnish version of Nordic Classification on Surgical Procedures (since 1996). The conisation group (n = 15 865) included cold knife conisation, large loop excision of the transformation zone (LEEP, LLETZ or LOOP) and laser conisation. The ablation group (n = 8 612) included cryotherapy, electrocoagulation and laser vaporisation. The other excisional treatments group (n = 2 450) included treatments such as cervical amputation, other cervical excisions and excision of the lesion of cervix uteri. All diagnoses in the last group were carefully reviewed, and women with irrelevant diagnoses were excluded. Treatment modalities have been previously described in detail.1 We also had information on repeat treatments. Women were included in the analysis according to the first treatment. Different treatment groups were analysed separately.

We used general mortality rates for the female population in 5-year age groups to calculate the expected numbers of deaths. We calculated standardised mortality ratios (SMRs) for total number of deaths between the time of treatment for CIN and 31 December 2006 and for different causes-of-death groups using the Finnish 54-item short list for causes of death13 with 95% CI by dividing the number of deaths by the number of expected deaths. The causes of deaths were further collapsed to ten groups for all causes (Tables 1 and 2) and to six groups for cancers (Table 3).

Table 1.  Mortality among women treated for CIN during 1986–2003
Short list of causes-of-death number*Mortality classICD-10 codesTotalConisation (n= 15 865 women) SMR (95% CI)Ablation (n= 8 612 women) SMR (95% CI)Other excisional treatments (n= 2450 women) SMR (95% CI)
Observed/expectedSMR (95% CI) (n= 25 827 women)
01–53All deathsA00-Y89395/338.91.17 (1.05–1.28)1.15 (1.00–1.30)1.09 (0.86–1.32)1.30 (1.03–1.56)
01–41All deaths from diseases and accidental poisoning by alcoholA00-R99, X45307/270.81.13 (1.01–1.26)1.06 (0.90–1.22)1.09 (0.83–1.35)1.38 (1.08–1.68)
04–22All cancersC00-D48145/133.01.09 (0.91–1.27)1.11 (0.87–1.34)0.72 (0.42–1.02)1.46 (1.01–1.90)
27–30Diseases of the circulatory systemI00-I425, I427-I9963/58.01.09 (0.82–1.35)0.93 (0.60–1.26)1.20 (0.61–1.78)1.39 (0.73–2.04)
41Alcohol-related diseases and alcohol poisoningF10, G312, G4051, G621, G721, I426, K292, K70, K860, K86, O354, P043, X4537/26.51.40 (0.95–1.85)1.56 (0.92–2.19)1.80 (0.74–2.87)0.54 (0–1.14)
01–03, 23–26, 31–40Other diseases 62/53.41.16 (0.87–1.45)0.84 (0.51–1.17)1.54 (0.85–2.24)1.59 (0.86–2.33)
42–53Injuries from accidents, suicides and assaultsV01-X44, X46-Y8988/67.31.31 (1.03–1.58)1.52 (1.12–1.91)1.09 (0.57–1.62)0.98 (0.47–1.50)
42–49Injuries from accidentsV01-X44, X46-X59, Y10-Y15, Y85-Y8627/29.70.91 (0.57– 1.25)0.84 (0.40–1.29)1.46 (0.55–2.36)0.48 (0–1.02)
50Suicides and sequelae of intentional self-harmX60-X84, Y87.051/30.51.67 (1.21–2.13)2.12 (1.42–2.81)0.85 (0.17–1.54)1.39 (0.48–2.31)
51Assaults and sequelae of assaultsX85-Y09, Y87.14/4.70.85 (0.76–0.93)1.14 (0–2.43)0.92 (0–2.72)
52–53Injuries from unclear intention **Y16-Y86, Y872, Y88-Y896/2.42.51 (0.50–4.52)3.00 (0.06–5.94)3.95 (0–9.43)
Table 2.  Mortality among women treated for CIN by delivery post-treatment
Short list of causes-of-death number*Mortality classNo deliveryAny deliveryPreterm delivery
Observed/expected (n= 19 667 women)SMR (95% CI)Observed/expected (n= 6160 women)SMR (95% CI)Observed/expected (n= 690 women)SMR (95% CI)
01–53All deaths361/295.31.22 (1.10–1.35)34/43.60.78 (0.52–1.04)15/6.02.51 (1.24–3.78)
01–41All deaths from diseases and accidental poisoning by alcohol285/235.91.21 (1.07–1.35)22/34.80.63 (0.37–0.90)11/4.82.31 (0.94–3.67)
04–22All cancers136/115.91.17 (0.98–1.37)9/17.10.53 (0.18–0.87)3/2.31.28 (0–2.73)
27–30Diseases of the circulatory system59/50.51.17 (0.87–1.47)4/7.50.54 (0.01–1.06)4/1.03.92 (0.08–7.75)
41Alcohol-related diseases and alcohol poisoning35/23.11.52 (1.01–2.02)2/3.40.59 (0–1.40)2/0.54.29 (0–10.23)
01–03, 23–26, 31–40Other diseases55/46.51.18 (0.87–1.50)7/6.91.02 (0.26–1.78)2/0.92.13(0–5.08)
42–53Injuries from accidents, suicides and assaults76/58.71.30 (1.00–1.59)12/8.71.39 (0.60–2.17)4/1.23.37 (0–6.67)
42–49Injuries from accidents25/25.90.97 (0.64–1.31)2/3.80.52 (0–1.25)1/0.51.91 (0–5.65)
50Suicides and sequelae of intentional self-harm42/26.61.58 (1.10–2.06)9/3.92.29 (0.80–3.79)2/0.53.72 (0–8.88)
51Assaults and sequelae of assaults4/4.10.97 (0.02–1.92)0/0.60/0.1
52–53Injuries from unclear intention **2.3/2.12.40 (0.30–4.50)1/0.33.25 (0–9.62)1/0.023.73 (0–790.23)
Table 3.  Mortality of certain cancer types among women treated for CIN
Mortality classICD-10 codesTotalNo deliveryDelivery
Observed/expected (n= 25 827 women)SMR (95% CI)Observed/expected (n= 19 667 women)SMR (95% CI)Observed/expected (n= 6160 women)SMR (95% CI)
All cancersC00-D48145/1331.09 (0.91–1.27)136/115.91.17 (0.98–1.37)9/17.10.53 (0.18–0.87)
Cancers of the larynx, trachea and lungC32-C3415/11.81.27 (0.63–1.91)15/10.31.46 (0.72–2.19)0/1.5
Breast cancerC5035/37.80.93 (0.62–1.34)32/32.90.97 (0.64–1.31)3/4.90.62 (0–1.32)
Cervix cancerC5319/2.57.69 (4.23–11.15)19/2.28.83 (4.86–2.80)0/0.3
Uterine cancerC54-C553/2.61.17 (0–2.50)3/2.21.35 (0–2.87)0/0.3
Ovarian cancerC567/10.20.69 (0.18–1.20)7/8.90.79 (0.20–1.38)0/1.3

Results

We observed 395 deaths among 25 827 women treated for CIN, which is more than expected (SMR 1.17, 95% CI 1.05–1.28) (Table 1). The majority of deaths from diseases and medical conditions were caused by cancers and diseases of the circulatory system. The risk of death from accidental and intentional injuries was increased (SMR 1.31, 95% CI 1.03–1.58), mainly due to increased suicide rate (SMR 1.67, 95% CI 1.21–2.13). Also the mortality from alcohol-related diseases and alcohol poisoning was increased, but only with borderline significance (SMR 1.40, 95% CI 0.95–1.85).

An increased overall mortality rate was observed in all treatment groups, except in the ablation group. The mortality was increased both in the conisation group (SMR 1.15, 95% CI 1.00–1.30; 217 deaths) and in the other excisional treatments group (SMR 1.30, 95% CI 1.03–1.56; 93 deaths) (Table 1). Women in the conisation group had increased mortality rate for all injury deaths (SMR 1.52, 95% CI 1.12–1.91) and suicides (SMR 2.12, 95% CI 1.42–2.81). Women in the other excisional treatments group had increased risk of all deaths from diseases and medical conditions (SMR 1.38, 95% CI 1.08–1.68) and cancers (SMR 1.46, 95% CI 1.01–1.90). Women in the ablation group had a trend towards increased risk for alcohol-related deaths and injury deaths from accidents, but the numbers were small.

Women with no delivery post-treatment had increased risk of death (SMR 1.22, 95% CI 1.10–1.35, Table 2). The risk followed general mortality pattern for all women with treatment for CIN, but with somewhat higher SMRs.

Women who had delivered post-treatment tended to have decreased overall mortality (SMR 0.78, 95% CI 0.52–1.04) and decreased disease mortality (SMR 0.63, 95% CI 0.37–0.90), especially from cancers (SMR 0.53, 95% CI 0.18–0.87). In addition, these women tended to have decreased mortality of diseases of the circulatory system but increased risk of injury deaths and suicides (Table 2).

Women with at least one preterm delivery after treatment for CIN had increased mortality (SMR 2.51, 95% CI 1.24–3.78, n = 690 women, 15 deaths) due to deaths from diseases of the circulatory system, alcohol-related deaths, injuries and suicides (Table 2). Majority of these women (n = 11) had delivered between 33 and 36 gestational weeks, one at 31st gestational week and three before 28th gestational week.

Cancer deaths were studied in more detail (Table 3). There was a tendency towards increased cancer mortality after treatment for CIN (SMR 1.09, 95% CI 0.91–1.27). However, women who had delivered post-treatment had decreased cancer mortality (SMR 0.53, 95% CI 0.18–0.87). Among all women, the risk for cervical cancer was increased (SMR 7.69, 95% CI 4.23–11.15). This risk was even higher among women who had not delivered post-treatment. There was a trend towards increased risk from cancer of the larynx, trachea and lung, but for a decreased risk of breast and ovarian cancers.

Discussion

We demonstrated that women treated for CIN have increased mortality rate, especially disease mortality and injuries from accidents and suicides. Mortality of cervical cancer was high, as expected. Women who had delivered post-treatment had decreased overall disease mortality rate. However, women who had preterm delivery had increased mortality rate.

We used Finnish population-based registers that include reliable data on hospitalisations, deliveries and deaths. Women treated for CIN are relatively young with low expected mortality rates. Our observation period was long, up to 20 years, and the study population was large. Although the number of deceased women who had delivered preterm was relatively small, we were able to demonstrate an increased overall mortality. LOOP conisation was introduced in Finland in early 1990s, and the procedures have been registered since 1994.

The total number of key variables available in routine administrative registers is limited. We lacked information of other potential confounders, such as smoking, alcohol consumption and socio-economic status. We had information on smoking habits only for women who had delivered post-treatment. These women smoked during pregnancy almost twice (27%) as often as pregnant women in general (15%).1 Smoking is known to be a strong risk factor for cervical neoplasia.5 It is likely that smoking rates are high among all women with CIN. This may also explain increased tendency towards other smoking-related cancers. In some studies, smoking has also been linked to preterm delivery. Smoking may predispose to infections that can lead to preterm birth.14 CIN is caused by the HPV infections that are mainly sexually transmitted. Risk-taking behaviour among these women may also increase the risk for other sexually transmitted infections. Certain HPV types, most notably HPV 16/18, are associated with anogenital, laryngeal and oropharyngeal carcinomas,15 which may explain the increased mortality from these cancers as well. Low socio-economic position is also associated with increased risk for CIN and cervical cancer.16 In Finland, both smoking in general17 and during pregnancy18 is strongly related to low socio-economic status.

Alcohol-related, accidental and intentional injury deaths were more common than expected, which led to the assumption that alcohol consumption among these women was high. In Finland, the proportion of alcohol-related deaths is high, constituting about 2% of all deaths among women.19 This proportion is considerably higher for injury deaths than deaths from diseases and medical conditions and in young populations,20,21 such as the population of women in our study. The results suggest that our findings can be partly explained by risk-taking behaviour and poor health habits. It must, however, be recognised that these findings demonstrate associations rather than causality. Beside high mortality rate from alcohol-related causes, Finland has also a high suicide rate among women, that is 9/100 000 in 2006.22 Therefore, generalisation of our results on other countries may be difficult.

Our study confirms that the risk for cervical cancer remains high after treatment for CIN. The risk for invasive cervical cancer seems to persist at least 20 years after the treatment.3,4 In some studies, pregnancy has been shown to protect against ovarian cancer. Preterm delivery seems to increase the risk for ovarian cancer compared with women with term deliveries.23 In our study, the number of ovarian cancers was too low for statistical significance, although there was a tendency towards decreased mortality.

Previous studies on maternal deaths and pregnancy-related deaths have shown a so-called healthy pregnant woman effect, that is parturients are healthy and their mortality from diseases and medical conditions is lower than that among nonpregnant women or among women with adverse pregnancy outcomes.6,7 Their behaviour protects babies and themselves, which further explains low injury mortality. The protective effect can be seen in long-term mortality even among grand multiparous women.24 In our data, overall disease mortality including cancer deaths was decreased among women who delivered post-treatment. Also, alcohol-related deaths and injury deaths from accidents were rare. We found a tendency towards high mortality from all injuries and suicides, which may be explained by background characteristics and health behaviour.

Women who deliver preterm are at increased risk for cardiovascular disease later in life. A strong association between preterm delivery and cardiovascular mortality has been demonstrated.8–11 Mechanisms linking preterm delivery with excess maternal cardiovascular mortality are not well understood. Infection and inflammation is present in most early preterm deliveries.25 This can also contribute to the development of coronary artery disease later in life.26 These findings are in accordance with our observations. It is noteworthy that our material included three deaths among women with extremely preterm birth. Their proportion (20%) was much higher than expected since these deliveries constitute less than 1% of all deliveries in Finland.27

In conclusion, we found an increased mortality among women treated for CIN. This can be partly explained by risk-taking behaviour among these women. However, those women who had subsequent term delivery had decreased mortality rates. Preterm delivery was an indicator of increased mortality, especially from diseases of the circulatory system. Although this is biologically plausible, the pathways should be studied in more detail.

Disclosure of interest

All authors declare that they have nothing to declare.

Contribution to authorship

M.J. conceived the study in collaboration with A.-M.T., M.G. and J.P. M.G. retrieved register data and performed the statistical analyses. All authors contributed to the data interpretation. M.J. wrote the first draft of the manuscript, and all authors contributed to the revision and accepted the final version.

Details of ethics approval

The analyses were performed after the register-keeping organisation, STAKES, had given the authorisation required by the national data protection legislation.

Funding

This study was supported by the Clinical Graduate School in Pediatrics and Obstetrics/Gynecology, University of Helsinki, Finland, and a grant from the Paulo Foundation. STAKES covered the expenses for data linkages and analyses.

Acknowledgement

Mark Philips, language reviser at STAKES.

Ancillary