Sentinel lymph node status in vulval cancer: systematic reviews of test accuracy and decision-analytic model-based economic evaluation

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 09/112/03. The contractual start date was in November 2010. The draft report began editorial review in April 2012 and was accepted for publication in May 2013. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors' report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Copyright statement

© Queen's Printer and Controller of HMSO 2013. This work was produced by Meads et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Description of underlying health problem

Vulval cancer is a relatively rare gynaecological malignancy diagnosed mainly among elderly patients. In 90% of cases it develops as a squamous cell carcinoma (SCC) and the remaining 10% are melanomas, Paget's disease, Bartholin's gland tumours, adenocarcinomas and basal cell carcinomas. 1 Lesions mainly occur on the inner edges of the labia majora (around 50% of the cases), less often in the labia minora and very rarely on the clitoris or in the Bartholin glands. The symptoms of vulval cancer include a lump on the vulva, vulval bleeding, itching, pain or ulceration, and approximately 90% of women present with a visible tumour. 1 Diagnosis is by biopsy with histological examination of the sample. Histological examination can include immunohistochemical analysis which may enable more a precise interpretation of the degree of dysplasia compared with conventional haematoxylin and eosin (H&E) staining. 2

Vulval cancer can be locally invasive and can spread via the lymphatic system to the inguinal or femoral nodes. If the primary tumour is laterally located in the vulva, spread may be only to that side; if the tumour is central, spread may be to either side. Lymphatic spread is strongly related to lesion size: metastasis is present in 20–30% of tumours < 2 cm in diameter and in 44% of tumours > 2 cm. 3 Correlation of lymph node status with depth of invasion is also important. Tumours with a < 1-mm depth of invasion have a < 1% risk of nodal spread. 4,5 Lymph node status is regarded as an important predictor of survival. 6 Malignancy in the lymph nodes can result in invasion of the blood vessels of the groin, including rupture of the femoral blood vessels. Distal spread via the bloodstream is relatively rare. 2

Once vulval cancer is diagnosed, staging is important in order to plan treatment and estimate prognosis. Two staging systems are frequently used, one developed by the International Federation of Gynecology and Obstetrics (FIGO)7,8 and one by the tumour node metastasis (TNM) classification of malignant tumours7 (Tables 1 and 2). Both staging systems have gradually evolved over time. Table 3 gives a comparison of FIGO and TNM staging. 2

The grade of malignancy refers to the extent of differentiation seen on microscopic examination and gives an indication of how fast the malignancy is likely to develop. In vulval cancer three grades are defined:

• Grade 1: cells are low grade or well differentiated and histologically look very much like normal vulval cells.

• Grade 2: cells are medium grade or moderately differentiated and look more abnormal than grade 1 cells, but not so much as grade 3 cells.

• Grade 3: cells are high grade or poorly differentiated or can be undifferentiated, they are very unlike normal vulval cells.

Protocol development and overview of review methods

A protocol was developed for undertaking systematic reviews of test accuracy, diagnostic and therapeutic impact and effectiveness of treatment for vulval cancer (see Appendix 1). Scoping searches for relevant systematic reviews were conducted in MEDLINE, EMBASE and The Cochrane Library [systematic reviews, HTA, Database of Abstracts of Reviews of Effects (DARE)] (see Appendix 2).

Systematic reviews were carried out using established methods in line with the recommendations of the NHS Centre for Reviews and Dissemination and the Cochrane Collaboration,42,43 including those of the Cochrane Methods Working Group on Screening and Diagnostic Tests. 44 Presentation of results is according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines. 45

Inclusion of studies, data extraction and quality assessment were carried out in duplicate, with differences resolved by consensus and/or arbitration involving a third reviewer. The selection process was piloted by applying the inclusion criteria to a sample of papers first. A two-stage process was then followed. First titles and abstracts were screened, then, for all references categorised as ‘include’ or ‘uncertain’ by both reviewers, the full text was retrieved wherever possible and final inclusion decisions were made based on the full paper. Reference Manager version 12.0 (Thomson Reuters, New York, NY, USA) software was used to construct a database of citations for all systematic reviews.

Clinical, methodological and statistical data extraction was conducted into data extraction sheets by at least two reviewers and discrepancies were resolved through discussion. If consensus could not be reached, disagreements were resolved by arbitration by a third reviewer. For diagnostic studies, information regarding study design and methods, characteristics of participants, SLN biopsy and comparison tests, and outcomes of interest were extracted using data extraction forms (see Appendix 3). For the effectiveness review, separate data extraction forms were used for different study designs: comparative experimental study (part A), comparative observational study (part B) and non-comparative study (part C). The data extraction sheets used are shown in Appendix 4. The quality assessment questions for randomised controlled trials (RCTs) were included in the data extraction sheet, but a separate form was used for case series (see Appendix 4). Data extraction was managed with Microsoft Word 2003 (Microsoft Corporation, Redmond, WA, USA) and Microsoft Excel 2003 (Microsoft Corporation, Redmond, WA, USA). Quality was also assessed independently by two reviewers. Any disagreements were resolved through discussion or by arbitration by the third reviewer. For each review a comprehensive master database of articles was constructed using Reference Manager 12.0 software.

Methods for test accuracy, diagnostic and therapeutic impact review

Methods for effectiveness reviews

Study selection

From the searches, 2942 citations were identified, of which 82 full papers were obtained. Included were 26 relevant studies (38 publications) (Figure 3). Excluded full-text articles are listed in Appendix 9 with reasons for exclusion, which were mostly because of small sample size or type of publication (reviews, abstracts). Some studies were excluded because they gave results only per groin rather than per patient.

FIGURE 3.

A PRISMA diagram for diagnostic review.

Quality of included studies

Quality assessment is reported in Table 10. Of the 26 studies included, four53,66,67,71 provided no information about the histological staining method used. Brunner et al. ,52 Camara et al. ,53 Hauspy et al. 58 and Rob et al. 69 used frozen section as the reference standard, rather than more routine histopathological techniques. In 16 studies, on receipt of negative results by H&E procedures, immunohistochemical tests using specific protein antibodies such as AE1, AE3, S-100, human melanoma black monoclonal antibody (HMB)-45, monoclonal antibody, cytokine myocyte nuclear factor (CKMNF), cytokine (CK)-88 and epithelial membrane antigen were conducted. In others, ultrastaging was used if samples were negative by H&E staining and standard sectioning. The thickness of slices varied from one study to another so that some studies were more likely to find small metastatic deposits than others because of the thinner sections taken.

There were four studies50,65,70,73 in which, if the SLN was found to be negative, no IFL was performed but patients were followed up clinically instead. In a study by Van den Eynden et al. ,72 10 out of 32 patients had a SLN biopsy plus full IFL. In the remaining 22 patients, an IFL was performed only if the SLN was positive or not found. In a study by Johann et al. 59 and another by Vidal-Sicart et al. ,74 some patients had SLN biopsy and full IFL regardless of node statistics and some only had IFL if the SLN was positive, but the results for the two groups were reported separately. Only the results for SLN biopsy and full IFL are reported here. In Crosbie et al. ,54 Klat et al. ,60 Martine-Palonez et al. ,63 Vakselj and Bebar71 and Vidal-Sicart et al. ,74 clinical follow-up was reported, and, for all except Vidal-Sicart et al. ,74 this was reported according to whether patients had been SLN positive or negative.

Because the main aim of the included studies was the analysis of diagnostic procedures, most did not report information about the number of patients who had undergone specific types of surgery or other treatment procedures (see Appendix 10, Table 53, for treatment descriptions). Usually, patients underwent radical vulvectomy, wide local excision or hemivulvectomy. RT was performed in only six studies (as adjuvant therapy in Hauspy et al. ,58 Levenback et al. ,61 Moore et al. ,65 Vakselj and Bebar71 and Van der Zee et al. 73 or as palliative treatment in Terada et al. 70). Additionally, a study by Levenback et al. 61 mentioned that, in one patient in whom SLN was grossly positive after SLN biopsy, the surgeon aborted IFL in favour of RT. Adverse events (AEs) were reported in five studies54,65,70,72,73 (see Appendix 10, Table 55).

With regard to blinding of index and reference test results, it would have been possible for the SLN and the full IFL nodes to be examined by different pathologists blind to each other's reports, but only de Hullu et al. 56 achieved this, and only De Cicco et al. 55 mentioned that they had not blinded pathologists. The remaining studies did not mention any blinding.

Test accuracy results

Results of test accuracy studies for which all patients had IFL as the reference standard are given in Table 11 and for which IFL was the reference standard in test positives and clinical follow-up in test negatives is given in Table 12. Most of the studies reported their results per groin rather than per patient; therefore, teasing out the results per patient was difficult in several of the papers (noted as numbers unclear in the comments column of Tables 11 and 12).

For calculation of sensitivity and specificity, studies have been categorised by the reference standards used, the index test used and the histopathological techniques used as follows:

• IFL for all

  • ^99mTc with blue dye

    • H&E only or insufficient details to determine whether immunohistochemistry or ultrastaging were used (Table 13)

    • immunohistochemistry (Table 14)

    • frozen section only (Table 15)

    • immunohistochemistry with ultrastaging (Tables 16 and 17)

  • ^99mTc only

    • H&E only or insufficient details to determine whether immunohistochemistry or ultrastaging were used (Table 18)

    • immunohistochemistry (Table 19)

  • blue dye only

    • immunohistochemistry (Table 20).

• IFL for SLN positive and clinical follow-up for SLN negative

  • ^99mTc and blue dye

    • immunohistochemistry (Table 21)

    • ultrastaging (Tables 22 and 23).

The sensitivities and specificities in the results tables are based on the number of patients with one (or more) SLNs found, rather than the total number of patients. They could have had SLN biopsy in one or both groins; therefore, the results are giving the sensitivity and specificity of malignancy or none when a SLN (or more than one) has been found. It would not be possible for a SLN biopsy to be false-positive for malignancy, so all the point estimates of specificity are 100%. For this reason, only one receiver operating characteristic (ROC) plane is given, for the category of IFL for all, ^99mTc with blue dye: ultrastaging and immunohistochemistry, in which there were 11 included studies (see Tables 16 and 17, Figure 4). All of the points are clustered along the top left-hand-side and the plot is not particularly informative. Unfortunately, there were insufficient studies of similar clinical characteristics to be able to conduct metaregression comparing ^99mTc and blue dye results.

FIGURE 4.

Receiver operating characteristic plane for IFL for all, ^99mTc with blue dye: ultrastaging and immunohistochemistry.

The results of sensitivities and specificities show that, although the point estimates are either 100% or close to 100%, the CIs are wide, reflecting the small samples available. However, the pooled sensitivity of the studies with clinical follow-up was less than the ones with IFL for all.

In a test for cancer, it is important not to miss malignancy. If a FN diagnosis is made, the patient is at risk of developing malignancy in the groin or disseminated malignancy. If a FP diagnosis is made, the patient would undergo unnecessary IFL with resulting morbidity. So FNs may be considered to be relatively more important than FPs. In SLN biopsy, there would be no false-positive diagnoses unless an error was made in the histological examination, which would be rare, so we have to evaluate only the FN diagnoses. The results suggest that, if we evaluate SLN biopsy with clinical follow-up for node-negative patients, many more FN diagnoses will be made because of the longer time of follow-up, enabling more development of observable metastases. However, one would assume that the accuracy of a test would be a function of the test itself, rather than the method of evaluation. It is known that differential verification usually leads to increased estimates of sensitivity and specificity. The opposite is seen in this example and may be because there is more time to develop a recurrence with a long clinical follow-up. In addition, total recurrence rates were used rather than groin recurrence rates only because some studies only gave total recurrence rates. It is reasonable to assume that IFL should not influence recurrence in the vulval area. Therefore, the recurrence rates for clinical follow-up studies were recalculated, using only groin and distant recurrence, where given (only available in the study by Van der Zee et al. 73). The results are shown in Table 23 and show a higher pooled sensitivity, comparable to the estimates for which all patients received IFL as the gold standard.

The probability of curing vulval cancer is greatest when it is diagnosed at an early stage. The studies included patients with disease at a variety of FIGO stages; some included only patients with early-stage disease51,63,67,71 and others patients with disease of all stages50,54,65 (when reported). A study by Nyberg et al. 66 had a mixture of stages but reported results for each stage separately so results for stages I and II are reported here. There was a trend for the studies with early FIGO stage patients to have higher sensitivities than studies with mixed-FIGO-stage patients, but this was not consistent across all studies.

The vast majority of vulval cancer (> 90%) is SCC. Vulval melanoma is the second most common vulval malignancy but, unlike SCC, melanoma has a high risk of metastasis and the overall prognosis is poor. However, studies that included malignancies other than SCCs did not appear to have noticeably different results to those with SCCs only, but this may be a result of small sample sizes.

Sentinel lymph node detection rate

The accuracy of a diagnostic test such as SLN biopsy depends on the ability of the surgeon to identify the SLN; therefore, SLN detection rates for each of the analysed techniques (blue dye, ^99mTc, blue dye/^99mTc) are presented (Table 24). The detection rate calculated per patient was available in all included studies. Some of the studies also gave detection rates per groin for ^99mTc and blue dye separately, but these are not reported here. The detection rates do not obviously vary by whether or not patient groups included non-SCCs, such as melanomas, or whether both early and late stage were included rather than early only. However, it is clear that blue dye detects fewer SLNs than ^99mTc and that both used together is the most successful strategy.

Training and experience

There is a learning curve for surgeons performing SLN biopsy and IFL. Several of the included studies mention this (Table 26), but for most of the studies the learning curve was taken to mean that, after the first few cases, only SLN biopsy without full IFL was to be performed. Only Levenback et al. 61 calculated that the rate of SLN detection was worse in the first 2 years of the study but then continues with full IFL for all patients regardless of node status.

Recurrence rates

Two groups of studies gave recurrences at follow-up. The first group are those that used full IFL regardless of node status at initial operation to establish diagnostic accuracy, but also followed up patients afterwards. The second are those that used clinical follow-up for SLN-negative patients to establish diagnostic accuracy.

In group 1 are Crosbie et al. ,54 Klat et al. ,60 Martinez-Palonez et al. ,63 Vakselj and Bebar71 and Vidal-Sicart et al. 74 Overall recurrence rates are given in Table 11. Vidal-Sicart et al. 74 did not give recurrences by SLN status. In Crosbie et al. ,54 6 out of 31 patients were SLN positive at biopsy and five SLN-positive patients developed recurrences. No SLN-negative patients developed recurrences. In Klat et al. ,60 14 out of 23 patients were SLN positive at biopsy (and one with a malignancy not in SLN) and three SLN-positive patients developed recurrences. One SLN-negative patient developed recurrence. In Martinez-Palonez et al. ,63 6 out of 27 patients were SLN positive at biopsy (and one with malignancy not in SLN) and four SLN-positive patients developed recurrences. Recurrences also developed in possibly three SLN-negative patients and in one patient for whom the SLN was not found (the numbers are unclear from the journal article). In Vakselj and Bebar,71 10 out of 35 patients were SLN-positive at biopsy and six SLN-positive patients developed recurrences. Three SLN-negative patients also developed recurrences.

In group 2 are Achimas-Cadariu et al. ,50 Moore et al. ,65 Terada et al. ,70 Van den Eynden et al. 72 and Van der Zee et al. 73 The overall recurrence rates are given in Tables 12 and 22. Achimas-Cadariu et al. 50 did not give recurrences by SLN status, so test accuracy could not be calculated. Van der Zee et al. 73 gave separate results for groin and local recurrences. In addition, Van der Zee et al. 73 gave recurrences for node-negative and node-positive patients in separate papers (node negative, Van der Zee et al. ;73 node positive, Oonk et al. 83). It was curious that there were 34 local recurrences in each category.

Given the recurrence results, it is reasonable to assume that the number of clinically apparent recurrences is likely to be smaller than the number of SLN-positive patients at biopsy. This may be because the subsequent IFL is removing malignancy that might otherwise develop into a recurrence. In addition, in a very small proportion of patients who do not undergo IFL, some recurrences will be distant rather than groin because the groin metastases may stay very small and not be noticeable.

In addition, it shows that some SLN-negative patients will develop recurrences so will be FNs. In Martinez-Palonez et al. 63 and Vakselj and Bebar,71 there was a higher rate of recurrences in SLN-negative patients than known FN SLN biopsies. In Crosbie et al. 54 and Klat et al. ,60 the rates were the same.

Survival rates

Nine studies gave information about survival. 50,54,60,63,65,70,71,73,74

In the study by Achimas-Cadariu et al. ,50 12 out of 46 patients died during follow-up; median survival was 61.2 months for the whole cohort and 16.2 months for the eight patients with relapse.

Crosbie et al. 54 reported that 2 out of 32 patients died from disease during a median follow-up period of 62 months (range 33–84 months).

Klat et al. 60 reported that 1 out of 23 patients died from disease, with a follow-up of 8–46 months.

Martinez-Palonez et al. 63 did not mention any deaths in a group of 28 patients followed up for a median of 22.5 months (range 0–64 months).

Moore et al. 65 reported that 1 out of 35 patients died from intercurrent disease, with a median follow-up of 29 months (range 8–51 months).

Terada et al. 70 gave overall survival and disease-free survival curves for all patients and node-negative patients. Median follow-up of 21 patients was 4.6 years (range 2–8 years) and largely reflects losses to follow-up, as none of the node-negative patients died of cancer. Two out of three node-positive patients died of cancer and three patients died of other illnesses.

Vakselj and Bebar71 gave the status of each of the node-positive and -negative patients and length of follow-up. Out of the 10 node-positive patients, six died of disease at follow-up times of 42, 10, unstated, 3, 18 and 10 months (total range of follow-up was 3–55 months). Of the 25 node-negative patients, one died of disease at 49 months and one died of another cause (no time given) (total range of follow-up 2–52 months).

Van der Zee et al. 73 gave a disease-specific survival curve for node-negative patients with a median follow-up of 35 months (range 2–87 months) for 202 of the 276 patients with at least 24 months of follow-up. Four patients were lost to follow-up, 10 patients died of vulval cancer and 16 died of intercurrent disease. Five out of 34 patients with local recurrents eventually died of distant metastases at 15, 18, 22, 41 and 41 months after primary treatment. The 3-year disease-specific survival rate for patients with unifocal vulval disease and negative SLNs was 97.0%. In the Oonk et al. paper,83 in the group of patients with positive SLNs, the 5-year disease-specific survival was 77.3%. Survival was varied depending on the pathology, and was 64.9% when the SLNs were identified by routine pathology and 92.1% when identified by ultrastaging. During the follow-up of median 31 months (range 0–109 months) 15 died of other causes and 28 died of vulval cancer.

In the study by Vidal-Sicart et al. ,74 1 out of 50 patients died of disease, with a mean follow-up of 20 months.

In summary, vulval cancer is largely a disease of older women, so a relatively large proportion die of other causes during follow-up. Survival rates are better in women with a node-negative status than in those with a node-positive status.

Quality of life

The Van der Zee et al. 73 study had a substudy investigating QoL that was published separately (Oonk et al. 82). The authors used the European Organisation for Research and Treatment of Cancer QoL Questionnaire-C30 (EORTC QLQ-C30), the Functional Assessment of Cancer Therapy – Vulvar (FACT-V) and a patient’s opinion questionnaire. The EORTC QLQ-C30 consists of five functional scales, global health status and nine symptom scales. The FACT-V consists of five functional scales and four symptom scales. In both of these questionnaires, a higher score indicates better functioning on the functional scales and worse functioning on the symptom scales. The patient opinion questionnaire asked what they would recommend to a friend or family member with vulval cancer: IFL or SLN biopsy (with the chance of missing metastases at a rate of 1 in 10, 1 in 100 or 1 in 1000). Patients included in the Van der Zee et al. 73 study between March 2000 and December 2005 were eligible. Questionnaires were sent to 37 patients with positive SLN and 37 age-matched patients with a negative SLN. Sixty-two of the 74 patients (84%) returned the completed questionnaires.

The results of the EORTC QLQ-C30 are in given Table 27, the results of the FACT-V in Table 28 and the results of the patient’s opinion questionnaire in Table 29. The most useful result was the global health status/QoL, which had a mean of 80 and a standard deviation (SD) of 18 for the SLN group and a mean of 80 and a SD of 23 for the IFL group. The authors state that ‘our present study does not support our original idea that a decrease in especially long-term morbidity also translates into an improved overall quality of life’. 82 It may be that there were too few participants to detect a small difference in QoL.

The EORTC QLQ-C30 showed very few differences between the two groups; only on the financial difficulties scale was the score statistically significantly worse in the IFL group. For the FACT-V questionnaire, results were significantly worse for the contentment functional scale, oedema, complaints and stockings symptom scales.

For the patient’s opinion questionnaire, the authors analysed the maximum FN rate of the SLN procedure that would be acceptable to patients. There were significant differences between the SLN biopsy patients and the IFL patients for all three FN rates, so that the SLN biopsy patients were more likely than the IFL patients to accept the FN rates.

Adverse events

Information about AEs was generally poorly reported. Eight studies provided data. 50,52,54,59,65,70,72,73 In Brunner et al. ,52 8.7% of patients' groins had postoperative inguinal morbidity (inguinal seromas, abscess, wound breakdown), but this information was not given by group.

Adverse events according to surgical procedures (SLN biopsy, SLN biopsy plus IFL) and time interval (short and long term) are as follows:

Short-term AEs:

• For SLN biopsy only: transient lymph oedema (13%),59 wound breakdown (11.7%) and wound cellulitis (4.5%). 73

• For SLN biopsy + IFL: transient lymph oedema (39%),59 postoperative groin lymphocele (5.5%) and cellulitis arising in the labia majora (2.8%),65 wound cellulitis (9.5%) and seroma (4.3%),70 wound breakdown (34%) and wound cellulitis (21.3%),73 cellulitis (5.9%) and lymphocele (11.8%). 72

Longer term AEs:

• For SLN biopsy only: lymphoedema (1.9%) and recurrent erysipelas (0.4%). 73

• For SLN biopsy + IFL: wound infection (31%), wound dehiscence (5%), lymphocyst (22%) and chronic lymphoedema (16%),54 lymphoedema (25.2%) and recurrent erysipelas (16.2%). 73

In Achimas-Cadariu et al. ,50 AEs were presented according to surgical procedures and by vulva and groin locations:

• For SLN biopsy only: wound breakdown – 3% (vulva) and 3.6% (groin); haematoma – 3.6% (vulva) (no AEs from blue dye only).

• For SLN biopsy + IFL: wound breakdown – 5% (vulva) and 6% (groin); haematoma – 1% (vulva) and 3.7% (groin); chronic lymphoedema – 3.7% (vulva) and 8% (groin).

• For IFL only: wound breakdown – 2% (vulva) and 7% (groin); haematoma – 6.9% (vulva) and 6.9% (groin); chronic lymphoedema – 1.7% (vulva) and 4% (groin).

Study selection

The final search retrieved 14,038 potentially relevant citations, which were screened for relevance to the inclusion criteria. Relevant full-text articles for 313 citations were retrieved and 295 articles were subsequently excluded. The most common reasons for study exclusion were lack of the full-text version, wrong study design and wrong population. The list of excluded studies and reasons for their exclusion can be found in Appendix 11. Eighteen publications (corresponding to 17 studies) fulfilled the inclusion criteria (Figure 5). There was one RCT, three case–control studies and 13 case series included.

FIGURE 5.

A PRISMA diagram of the selection process for the clinical effectiveness systematic review.

Randomised controlled trials

Case–control studies and case series

Sixteen studies were included in this section (see Table 31). There was one matched case–control study (Helm et al. 37), two unmatched case–control studies (Manavi et al. 88 and Stehman et al. 89) and the remainder were retrospective case series. One study (Katz et al. 90) included more than 25% FIGO stage III and IV patients but the results for recurrence for stage I and II patients only by treatment received were clear, so this subgroup was reported here. The characteristics of the included studies are shown in Table 31. Almost all studies were conducted in the USA or Europe and the follow-up times varied between 1 and 30 years.

The numbers of patients given different types of treatments are shown in Table 32. The table indicates that the different studies evaluated several interventions and many of the groups were small. Many of the case series compared radical vulvectomy with hemivulvectomy and wide excision and it was frequently difficult to determine how many patients had received unilateral or bilateral IFL. The largest study was that by Kumar et al. ,99 which reported results from the US Surveillance, Epidemiology and End Results (SEER) database between 1988 and 2005. With this paper, it was impossible to determine the number of women who had had IFL so the results have been placed in the surgery to primary site-only category.

The case–control study by Helm et al. 37 matched 32 women who had a triple incision for IFL with 32 women of similar FIGO stage, lymph node status, greatest diameter of lesion and site of lesion with a single incision for IFL. The case–control study by Manavi et al. 199788 compared women who had had inguinofemoral RT with those who had no RT. IFL was not performed on any of the patients but all had had simple vulvectomy. The case–control study by Stehman et al. 89 compared women who had modified radical hemivulvectomy and ipsilateral superficial inguinal lymphadenectomy with a historical comparator group of women who had had radical vulvectomy and bilateral IFL. Cases and controls had comparable stage I disease, 5 mm or less invasion, no vascular space involvement and negative inguinal and femoral nodes.

The baseline characteristics of patients in the included studies are shown in Table 33. Their median or mean ages were mostly of 60 years, which reflects the fact that vulval cancer is mostly a disease of older women. However, some younger women were included down to age 20 years and Kumar et al. 99 compared characteristics of younger women (aged < 50 years) with older women (aged ≥ 50 years). Where location of primary tumour was specified in studies, the majority were on the labium majus and were medial rather than lateral. The vast majority of malignancies were SCCs but also included were a few adenocarcinomas, MMs and others in some of the studies. Some of the studies included just FIGO stage I, whereas others had a spread of stages from I to IV. Some of the studies gave TNM classification as well as FIGO stage but others just gave one or the other. Node status was given in eight studies only. 37,89,91,92,95,96,100,101

Objective

Patients with FIGO stage I or stage II vulval cancer may have an IFL as part of the treatment in order reduce the probability of groin and distant recurrence. As has been shown previously in this report, the morbidity of this procedure is high. Moreover, it is recognised that, although there is a possibility of groin lymph node metastases before IFL is conducted, for many patients the IFL is subsequently found to have been unnecessary since no metastases were found. Therefore, an accurate method of identifying which patients need IFL would help to reduce unnecessary morbidity in many patients.

The SLN biopsy may use one or both of blue dye and ^99mTc and is intended to identify one or more SLNs. The SLN is then examined using histopathological techniques to identify patients with metastases that need an IFL. The potential advantage of SLN biopsy is that there is much less morbidity because of a smaller incision if patients have no metastases in the SLN, although there is a risk that patients may test false-negative for metastasis and then go onto have a groin recurrence with an associated higher risk of mortality. The important potential advantage with SLN biopsy is that only one lymph node is removed, thereby reducing the incision, the extent of surgical dissection required and the potential for subsequent lymphocyst and lymphoedema formation as a result of fewer lymph nodes being removed.

The objective of this economic evaluation is to compare the relative cost-effectiveness of undertaking a range of SLN biopsy options that examine for the presence of metastases compared with implementing a strategy of ILN for all without first identifying whether or not patients need such radical treatment.

Developing the model structure

A decision tree was developed in TreeAge Pro 2001 software (TreeAge Software Inc., Williamstown, MA, USA). It was felt that a decision tree approach would be most appropriate for this economic evaluation for two reasons:

• The short time horizon.

• There are no examples of multiple recurrences (the same event happening to the same patient many times) seen in the model structure.

Women enter the model having been identified with a previous biopsy with presumed FIGO stage T1 or T2 unifocal vulval cancer (but not T1a). Only patients with SCC are considered because the vast majority of vulval cancer is SCC. Patients with basal cell carcinoma and MM are excluded from this analysis because of the different characteristics of these types of tumours.

The patients are assumed to follow one of seven different pathways that describe alternative approaches to the SLN biopsy and the treatment of vulval cancer. The first pathway is the comparison arm, which is the implementation of an IFL without a SLN biopsy and is used to show how this more morbid treatment compares to different SLN biopsy options. In the case of pathways 2–7, a SLN biopsy is performed using either blue dye, ^99mTc, or both, in order to identify the SLN. This is followed by histopathology, which is some combination of H&E staining and ultrastaging in order to test for the presence of metastasis, in which ultrastaging can be considered to be representative of more sensitive techniques such as immunohistochemistry. For patients with metastasis, an IFL is performed, with RT also given when necessary (for example, Van der Zee et al. ,73 in which RT was given when more than one intranodal metastasis and/or extranodal growth was detected). Following an IFL, patients are considered to be monitored every 3 months for 2 years for evidence of recurrence. If patients are only given a SLN biopsy and do not go on to IFL because no metastases were found, then patients are considered to be monitored every 2 months for 2 years. The seven patient treatment pathways are defined as follows:

1. IFL: this is performed on all patients, with no SLN biopsy being given.

2. Blue dye test and H&E: a blue dye test is administered intraoperatively to identify the SLN. This is followed by histopathology consisting of H&E staining of the SLN in order to identify the presence of metastasis.

3. Blue dye test and ultrastaging: a blue dye test administered intraoperatively to identify SLN, followed by histopathology consisting of H&E staining of the SLN. If no metastasis is detected then ultrastaging/staining is performed to confirm the absence or detect presence of metastases.

4. ^99mTc and H&E: a ^99mTc test is administered, patient imaging preoperatively and then a radioactive probe used detect signal at surgery to identify the SLN. This is followed by histopathology consisting of H&E staining of the SLN to identify presence of metastasis.

5. ^99mTC and ultrastaging: a ^99mTc test administered, patient imaging preoperatively and then a radioactive probe used detect signal at surgery to identify the SLN. This is followed by histopathology consisting of H&E staining of the SLN. If no metastasis is seen then ultrastaging/staining is administered to confirm absence or detect presence of metastases.

6. Blue dye and ^99mTc and H&E: both blue dye and ^99mTc test are administered to identify the SLN. Followed by H&E staining to identify the presence of metastasis.

7. Blue dye and ^99mTc and ultrastaging: both blue dye and ^99mTc test are administered to identify the SLN, followed by H&E staining. If no metastasis is seen, then ultrastaging/staining is administered to confirm absence or detect presence of metastases.

Morbidity can occur in the short and long term as a result of complications due to a SLN biopsy or an IFL. Local recurrence can occur at any time following either of these procedures, the probability of which is informed by data (see Table 42), whereas groin recurrence may occur depending on the outcome of biopsy result and the treatment response. RT may be implemented alongside an IFL, or in response to a recurrence among patients who have not previously received it. Chemotherapy is assumed to be administered to all patients who have a recurrence but have already received an IFL and RT. These points are illustrated in Figures 6–11:

FIGURE 6.

Summary of the decision pathway used in the economic model.

FIGURE 7.

Model Structure showing each of the seven treatment pathways and the treatment pathway for blue dye + H&E. This pathway is repeated for each of the pathways that include either blue dye and/or ^99mTc.

FIGURE 8.

Treatment pathway for IFL. Note: patients who have a local recurrence are given a primary excision. Patients who have a groin recurrence following IFL + RT are given chemotherapy. Patients who have a groin recurrence following an IFL only are given RT.

FIGURE 9.

Treatment pathway for strategies 2–7 following a TP result for metastasis. Note: patients who have an IFL with or without RT may still go on to have either a local or groin recurrence. In the case of a local recurrence, all patients are given a primary excision. Patients who have a groin recurrence having previously received an IFL + RT are given chemotherapy, whereas for those patients who have only previously received a IFL RT is given.

FIGURE 10.

Treatment pathway for strategies 2–6 following a FN result for metastasis. Note: patients who have a FN biopsy test result, subsequently go on to have a groin recurrence. In this case, all the patients receive both an IFL + RT.

FIGURE 11.

Treatment pathway for strategies 2–6 following a TN result for metastasis. Note: patients who have a TN biopsy test do not subsequently go on to have a groin recurrence. However, a local recurrence is still possible and under these circumstances the patients are given a primary excision.

Model assumptions

A number of assumptions are required in order to develop a workable model structure and enable the analysis to be carried out. The assumptions made in this study are described below and grouped into those that refer to the general pathway, recurrence and the wider model:

Data requirements

The data requirements for the economic evaluation are fulfilled by using the findings of the systematic review. However, when the results of many studies could not be generalisable to obtain an overall parameter value owing to differences in the study protocols, etc., the findings of larger and more recent studies have been preferred.

Outcomes

The main focus of this economic evaluation is how the different treatment scenarios impact on the mortality and morbidity of the patients. Therefore, the following outcomes have been examined in this analysis:

• case of death avoided within 2 years

• case of morbidity-free survival within 2 years

• case of long-term morbidity-free survival within 2 years.

In each case, these are compared again the comparison scenario of implementing an IFL to all patients.

As has been highlighted in the systematic reviews, there are no studies that have measured a generic QoL such as Euroqol EQ-5D in vulval cancer. Furthermore, the only study to measure global health status (QoL)82 showed no difference for patients who received the different procedures that are considered in this analysis. Therefore, the option of using the quality-adjusted life-year (QALY) as an outcome measure was not available in this economic evaluation.

The outcome measures in this study are considered to be reasonable given that the main focus of this study is whether or not a negative result from a SLN biopsy and no further treatment, with the possibility of recurrence, is preferable to the highly morbid IFL which has a lower risk of recurrence after the procedure.

Analysis

The model used in the economic evaluation begins with a hypothetical cohort of women who are presumed to have FIGO stage I or stage II vulval cancer. The model estimates the mean costs associated with each of the treatment strategies and the base case assumes that all women entering the model are aged 65 years. The time horizon of the model is 2 years, which was chosen as it was felt that any groin recurrences that might appear as a result of a FN SLN biopsy would be detected within this time frame. Owing to this short time horizon, and with the majority of costs occurring in the first year, no discounting was applied. This economic evaluation takes the form of a cost-effectiveness analysis and is carried out from the UK NHS perspective. Therefore, only direct costs and resources associated with the intervention and outcomes are incorporated in the analysis. The results of the analysis are presented in terms of incremental cost-effective ratios (ICERs) for each of the three outcome measures considered.