Dr K. C. K. Lim, Department of Obstetrics and Gynaecology, University Hospital of Wales, Heath Park, Cardiff, UK.
Objective To assess the effectiveness of clinical follow up after primary surgery for early stage cervical cancer.
Design Retrospective analysis of clinical follow up after radical hysterectomy and node dissection for early stage cervical cancer.
Setting Gynaecological Oncology Cancer Centre.
Sample Two hundred and ninety-one patients who underwent surgery for cervical cancer.
Methods Follow up data were collected retrospectively from hand-searched patients notes, as well as a computerised database (Information System for Clinical Organisation [ISCO]). The data were analysed using the SPSS for windows (SPSS, Chicago, Illinois) statistics package, using χ2, Kaplan–Meier life tables and Cox Linear regression analysis.
Main outcome measures To determine whether routine follow up was useful for detecting early recurrent disease.
Results Two hundred and ninety-one patients treated by radical hysterectomy and node dissections were followed up. The cumulative five-year survival for all cases in our series was 80% and 53/291 patients (18.2%) were found to have recurrent disease. The median period from surgery to recurrence was 17.6 months (3.0–60.0). Seven patients with recurrence were detected at a routine follow up examination, and two out of seven of the patients were asymptomatic. Detection of the recurrence on routine follow up was not an independent prognostic factor for survival when compared with age, stage and whether the patient received post-operative adjuvant therapy.
Conclusions Routine follow up in patients following radical hysterectomy and node dissection for early stage cervical cancer is not a sensitive way of detecting recurrent disease, as a high proportion of patients were symptomatic at the time of detection. As there are other reasons for follow up, we propose alternative methods of structuring the programme.
In the UK cervical cancer still claims 1222 lives a year.1 Despite the introduction of the National Cervical Screening Programme, women still present with advanced cervical cancer. Radical hysterectomy and node dissection is recognised as a treatment for stage IA2 to IB2 cervical cancer. However, following treatment, the follow up of these women is less evidence based. To date, only six studies have assessed the benefits of follow up in cervical cancer patients.2–7 The data from those studies conclude that very few patients were asymptomatic and diagnosed with recurrent disease at the time of routine follow up examination. In one UK series, of 674 women treated with radical hysterectomy for stage IB disease, only 13% of the patients were asymptomatic when recurrent disease was diagnosed.2 Studies of surgical and radiotherapy follow up of patients from Sweden4 and the United States,3 showed the incidence of asymptomatic recurrences to be 16% and 14.3%, respectively. However, the latter study concluded that the asymptomatic recurrence was worthwhile detecting as their survival was significantly increased. In all these analyses, most of these recurrences were diagnosed within the first two years after treatment. The majority of asymptomatic recurrences were discovered by pelvic examination, and not on vault cytology, which was rarely helpful. Conversely, some older series concluded that a larger proportion (25–37%) of the recurrences were found in asymptomatic patients.5,6
The reasons why patients are followed up after surgery for gynaecological oncology varies, but a UK survey on follow up protocols by Consultant Gynaecologists suggest that they follow up patients to detect early recurrences, for reassurance to the patient, to collect data on outcome and also for medico-legal purposes.8
It is clear from studies on the follow up of patients with endometrial cancer that there is little evidence that follow up is associated with better survival.9–12 Furthermore, there have been other studies showing no long term difference in survival rates after recurrence between patients whose recurrences were detected routinely and those who initiated their visits when they were symptomatic.12 The guidance from the NHS Executive with regards to cervical cancer follow up again emphasises that very few asymptomatic recurrences were detected routinely and that in the long term no survival difference was detected between symptomatic women and those whose recurrences were detected at routine follow up.13
In Wales, there has been a trend for all the cervical cancer cases to be centralised to a cancer centre, as proposed by the NHS executive guidelines.13 Over the last 15 years, data have been collected in a prospective and blind manner on patients with early stage cervical cancer who were treated surgically. We have now retrospectively analysed these data through the use of a computerised data bank as well as hand searches through patient notes.
Patients were referred throughout South and Mid Wales to the South Wales Regional Oncology Unit for the period 1985–1999. During this period, standard treatment for stage 1B and 2A disease was a consideration for radical hysterectomy and bilateral node dissection if it was deemed suitable. If the histology demonstrated more than one positive lymph node, lymphovascular invasion, close or incomplete resection margins or an unfavourable tumour type, then further adjuvant therapy was advised and the patient was referred on to the regional oncology unit.
For those whose operative findings and histology confirmed stage 1B cervical cancer with no added risk factors, follow up was with clinical surveillance. This consisted of an outpatient visit every three months for the first year, every six months for the second year and yearly thereafter until five completed years. If there was no recurrence, the patient would be discharged back to their GP.
Clinical surveillance consisted of an abdominal, pelvic bimanual and speculum examination. Follow up in the Gynaecological Oncology clinic was entirely clinical, routine cytological and radiological investigations were not employed on the basis that they were not specific enough to aid in the detection of recurrences.3,4,7
The findings at each visit and details of recurrences and further treatment/management/outcomes were recorded through hand searching of the case records as well as via computerised database records of the clinical episodes at the oncology centre (Information System for Clinical Organisation [ISCO]). All patients were checked electronically using the Welsh NHS network, the University Hospital of Wales Patient Management System (PMS) and ISCO. The result meant that the disease-free intervals as well as death events were not only recorded but also double-checked.
Statistical analysis was undertaken using SPSS v11.5 for windows (SPSS, Chicago, Illinois) statistics package. Analysis of non-parametric categorical data was performed using the χ2 test. A Cox's proportional hazard regression analysis was used in the analysis of the effects of demographics as well as clinical variables on survival. The variables were either continuous (i.e. age) or categorical and converted to binary data where appropriate. Each variable had a reference category (i.e. 0 = not detected at follow up, 1 = detection at follow up; and 0 = no post-operative adjuvant therapy and 1 = received post-operative adjuvant therapy). Stage was categorised as 1 = 1A2, 2 = 1B, and 3 = 2A. Kaplan–Meier curves were generated for graphical representation of survival. The censor date was the last date seen at follow up.
Between November 1985 and November 1999, a total of 318 patients referred to the University Hospital of Wales underwent radical hysterectomy and node dissection either on its own or in conjunction with a lower segment caesarean section in three cases. Twenty-seven patients were lost to follow up, leaving 291 cases for the study. As all the operations were performed or overseen by the same gynaecological oncologist (A.S. Evans), therefore a consistent operative method was employed. The mean and median age of the patient was 42.7 and 40.5 years, respectively; this was slightly skewed over towards the younger population. The median survival time was 47.4 months (range 5.5–58.0) and the median time to recurrence from date of radical hysterectomy was 21.8 months (range 3.0–60.0). The median time of follow up was 47.1 months (range 0.5–189.1).
Of the 291 cases, 274 (94.2%) were stage 1B disease and the remainder consisted of stage 1A2 and 2B disease. One hundred and ninety-eight (68.0%) were squamous cell carcinomas; the other 93 (32.0%) were a mixture of adenocarcinoma, adenosquamous carcinomas, clear cell carcinomas, mullerian adenosarcoma, planocell-ceratoides and small cell carcinoma.
The cumulative five-year survival for all cases in our series using the Kaplan–Meier survival tables was 80% with a median survival of 48.9 months. Fifty-three patients out of the 291 patients (18.2%) developed recurrent disease, and in a further 18 patients, it was unclear from the notes whether they had developed a recurrence. The majority of the recurrences (63%) occurred before 24 months and 77% before 36 months as shown diagrammatically in Fig. 1. For the patients who developed recurrence, the five-year survival was 30% and the median survival was 25.3 months. In six of the patients with a recurrence, it was not possible to be certain how this was found. Seven patients out of the remaining 47 (14.9%) had their recurrence detected during a planned follow up examination; the remaining 40 patients (85.1%) necessitated further investigations after self-presentation or through a referral from another medical practitioner (Table 1).
Table 1. Point of referral of patients who recurred.
Point of referral
Accident and Emergency
Furthermore, only two out of the seven 7 patients who were picked up on routine follow up were completely asymptomatic at time of detection, both had received adjuvant therapy and had their recurrences picked up on check CT scans. This makes the asymptomatic recurrence rate 2/291 (0.7%).
For the five patients with recurrent disease detected at a planned follow up clinic who were symptomatic, the median survival was 37.5 months as opposed to only 8.3 months for the two patients with recurrent disease who were asymptomatic at the time of their clinic appointment. When compared with the median survival for patients with a recurrence who self-presented (26.0 months), the differences were not statistically significant.
Fifty-six patients (19.3%) were referred to an oncology centre (Velindre Hospital) for further adjuvant therapy, which consisted of 41 post-operative radiotherapy patients (73.2%), 1 post-operative chemotherapy patient (0.02%) and 14 post-operative chemoradiotherapy patients (25.0%). There was a significant difference in the recurrence rate between the patients who did not receive (30/234; 12.8%) and those who received (23/57; 40.4%) post-operative adjuvant therapy (χ2 test; P= 0.000003) as diagrammatically shown in Fig. 2. This is reflected in the difference in the five-year survival for patients who had post-operative adjuvant therapy (49%) and without therapy (88%).
When the data were analysed using the Cox Regression analysis, detection of the recurrence on routine follow up was not an independent prognostic factor for survival when compared with age, stage and whether the patient received post-operative adjuvant therapy.
Despite the fact that the value of follow up in post-radical hysterectomy patients has been questioned in the past, traditional methods and practices continue to dictate current practice in the UK. There have been previous reports suggesting that routine post-therapy clinical surveillance may not be effective2,3,7 and furthermore, symptomatic patients who had not been adequately ‘educated’ to present early may defer presenting till their next appointment.14 This would represent a delay in detection and subsequent treatment. Our study, with 291 patients, supports these previous publications (patient numbers ranging from 96 to 674) and is the first to be described from South Wales. The period of study, which spanned 15 years, afforded a substantial period of follow up for the majority of the patients. The five-year overall survival rate of 80% and a recurrence rate of 18.2% are also consistent with other national studies.2–4 The patient population and type are stable and represent a fairly captive population with little movement throughout the country, and hence, we were able to obtain good follow up data.
The evidence in our study demonstrates that the patients who did recur were mainly symptomatic and will self-refer earlier to the hospital, with only a handful of cases (7/47) detected on routine examination in the outpatient clinic. Strikingly though, only two out of the seven were completely asymptomatic, making the asymptomatic recurrence rate (percentage of patients who were found to be symptom-free when recurrence was detected) 0.7%, which is lower than the findings in the UK and US studies.2–4 Two important points can be inferred here, firstly, there were five patients who preferred to wait till the next clinic appointment rather than present early, a well documented phenomenon14 which should be adequately addressed through good communication and patient information. Secondly, the survival data of these patients, who were detected through routine follow up clinics, did not appear to be better than patients who presented with their symptoms (as the median survival appeared to be below that of patients who recurred and self-presented: 11.9 vs 26.0 months). Although the numbers were too small for statistical significance, there are no published data on survival benefits of early detection with routine follow up. Studies of follow up in endometrial cancer have also shown the ineffectiveness of follow up as well as lack of good evidence that detection of recurrences at a routine follow up clinic changes the prognosis of the disease.9–12
Points of referral included other gynaecology consultants, physicians, urologist, orthopaedic surgeons and only a small proportion were referred through their primary care physician (Table 1). This may be because of the follow up umbrella of a Gynaecological Oncology clinic, which had open access to the patients and they were told to present to the clinic if they were in any way worried. Our study demonstrates the fact that the majority of patients with recurrent disease will present earlier with symptoms and therefore, if the aim of regular follow up is to detect early recurrent disease,8 this is not being achieved.
The majority of the recurrences (63%) occurred before 24 months and 77% before 36 months, although there were recurrences that occurred as late as 60 months. These data are entirely consistent with that of other similar cancer centres in the UK2 and the US3,5,7 and implies that any surveillance schedule should concentrate on the first two post-operative years.
Patients referred for further adjuvant therapy will, by definition, have poor prognostic markers, which reflect the much higher recurrence rate of 40.4% compared with 12.8% in those who did not receive adjuvant treatment. Interestingly, this group of patients was followed up by the clinical oncologists and despite a lower threshold in this group for further imaging for recurrences, only two cases of recurrent disease from this group were discovered at routine follow up (as opposed to five cases from the group who did not require adjuvant therapy). It could be inferred from these data that if resources were limited, it would be better to concentrate it on this group or possibly modify the follow up schedules to be more intense. The extent of the follow up examination and investigation is very variable and tends to be more conservative in the UK and comprises mainly of a thorough examination looking for lymphadenopathy and recurrence within the pelvis through a visual and digital examination. The current accepted UK follow up regimen does not involve the use of cytology or radiological imaging for screening out recurrences although there are other centres in the world where this might be seen as a useful adjunct to clinical follow up (e.g. a chest routine vault smears and chest X-rays also form part of the follow up practice in the US15). This obviously had cost and time implications, although some advocate that it is better for detecting asymptomatic early recurrences.7,16 Despite our streamline approach, on average, patients who are followed up for five years will have at least nine clinic appointments after their surgery and this has cost and clinical time implications in service provision. Cost benefit analysis on endometrial cancer routine follow up demonstrated that there was an unfavourable cost to follow up for the benefits that it returns.17 Another study of follow up on endometrial cancer reported that one asymptomatic recurrence was diagnosed for every 653 consultations.11 Although this has not been done for cervical cancer follow up, it is likely that the outcomes would be roughly similar.
Although our data support the idea that routine follow up has no benefit in detecting recurrences earlier, there are other claimed reasons for follow up.8 Psychological benefits to the patient, monitoring of late complications (e.g. bladder dysfunction), as well as the collection of other parameters such as survival and recurrence statistics, could equally be as important as the detection of recurrence. Further evaluation and rehabilitation of sexual function after extensive surgery, especially if post-operative irradiation has been applied, can also be performed as part of the follow up.15
Based on our findings in this study, we propose a more individualised approach to surveillance. By patient choice, they could enter into a contract with the gynaecological oncologist, to allow follow up to be performed in the primary care setting, or even by telephone, as successfully demonstrated in a breast cancer model.18 This would be on the provision that the patient presents as soon as possible if symptomatic. The alternative for the patient would be regular follow up in the oncology clinic as currently practised, and the choice of which could be left with the patient.
Because this was a retrospective study, we felt that despite our best efforts, the strength and depth of data collected was not as comprehensive as a prospective study. There were patients in the study whom it was not possible to obtain all the follow up data. Furthermore, there was no possibility of evaluating the patient's views on follow up. Another criticism of the study was that the referral patterns for pre- and post-operative chemo/radiotherapy have now changed since 1985 and that the more recent patients having surgery as a primary treatment may have smaller tumours and there may be a lower threshold for offering post-operative adjuvant therapy.
To conclude, follow up after radical cancer surgery of the cervix in stage 1B disease can be an expensive process with poor cost benefit ratio, as the detection of asymptomatic recurrence is poor. Although there are other benefits to post-therapy surveillance, it may be that the follow up programme could be more individualised to suit the patients' needs and expectations, and we have initiated a prospective study looking into how our post-therapy surveillance can be improved.
The authors would like to thank Ms Wendy Jones (Department of IT, Velindre Hospital) and Mrs Anne O'Brien (Clinic Coordinator, University Hospital of Wales) for their help in compiling the data.