Article Text

Download PDFPDF

Observational and cost analysis of the implementation of breast cancer sentinel node intraoperative molecular diagnosis
  1. R I Cutress1,
  2. A McDowell2,
  3. F G Gabriel2,
  4. J Gill1,
  5. M J Jeffrey2,
  6. A Agrawal1,
  7. M Wise1,
  8. J Raftery3,
  9. I A Cree2,3,
  10. C Yiangou1
  1. 1Breast Care Centre, Portsmouth Hospitals NHS Trust, Queen Alexandra Hospital, Portsmouth, UK
  2. 2Department of Pathology, Portsmouth Hospitals NHS Trust, Queen Alexandra Hospital, Portsmouth, UK
  3. 3Wessex Institute, University of Southampton, Chilworth Science Park, Southampton, UK
  1. Correspondence to Ian A Cree, Cancer Laboratory, Portsmouth Hospitals NHS Trust, Pathology Centre, Level F, Queen Alexandra Hospital, Portsmouth PO6 3LY, UK; ian.cree{at}porthosp.nhs.uk

Abstract

Background Accurate intraoperative sentinel lymph node (SLN) assessment enables axillary clearance to be completed immediately in node-positive breast cancer patients. This article reports a study of the introduction of intraoperative molecular SLN analysis in routine clinical practice in the Portsmouth Breast Care Centre.

Design There was prospective analysis of 254 consecutive patients who underwent SLN biopsy in a single centre. Nodes were sectioned at 2 mm intervals and alternate slices were analysed using a CE-marked assay for mammaglobin (MG) and cytokeratin 19 (CK19). Remaining slices of node were sent for histological analysis, which included CK19 immunohistochemistry. While the assay was being carried out, the surgeon performed the breast tumour resection. The cost per patient was estimated retrospectively and the cost effects on the hospital and primary care trust for a typical service were also estimated.

Results A total of 491 SLNs from 254 patients were evaluated. The intraoperative assay showed positivity of SLNs for metastatic cells in 78 patients. There was 100% detection of macrometastases within sentinel nodes analysed by GeneSearch. Overall concordance between histological status, including micrometastases and GeneSearch analysis, was 95% (sensitivity 96%, specificity 95%). The cost per procedure was increased for wide local excision with SLN biopsy and intraoperative assessment compared with other models, but fewer procedures were carried out.

Conclusion Intraoperative assessment of SLNs in breast cancer using a molecular assay is a safe, acceptable and accurate technique that allows a reduction in the frequency of delayed axillary clearance surgery. Take-up of this method may be hampered by perverse incentives operating within healthcare funding.

  • Breast cancer
  • metastasis
  • sentinel node

Statistics from Altmetric.com

Introduction

Sentinel lymph node (SLN) biopsy is associated with reduced arm morbidity and improved quality of life compared with axillary dissection.1 In the UK, following SLN biopsy (SLNB) and pathological examination of SLNs, approximately 70% of patients remain node negative2 and consequently avoid the morbidity of an axillary dissection. There is no significant compromise in oncological safety with this approach, since an axillary recurrence rate of 0.3% following negative sentinel node biopsy has been reported in a meta-analysis of the literature.3 This compares favourably with axillary recurrence rates following sampling or clearance, established as 5.4% and 3%, respectively, in one UK trial of 417 patients performed in the 1980s.4 In approximately 30% of patients, the SLNs contain metastases.2 In 50% of cases with macrometastases in SLNs, there are further positive nodes within the axilla.2 5 Current recommendations are therefore that SLN-positive patients require further axillary treatment.6 Accurate intraoperative staging of the axilla following SLN biopsy would enable immediate definitive axillary surgery to occur during the same operation, avoiding potentially hazardous reoperative surgery and possible delays to adjuvant treatments as a result of the further operation.

Pathological techniques of intraoperative sentinel node analysis include frozen section SLN analysis and touch imprint cytology. These techniques demonstrate high specificity, approaching 100%, but reported sensitivity is variable and generally around 60–70% for the two methods, particularly for micrometastases.7 8 Sensitivity of touch imprint cytology is limited by sampling issues, and frozen section analysis leads to loss of tissue for paraffin block histopathology. Indeed, even the sensitivity of the ‘gold standard’ paraffin block histopathology is less than 100% and increases with serial sections and immunohistochemical stains.9 In contrast to histopathology, molecular techniques sample more tissue as all tissue submitted can be homogenised and analysed. These new methods therefore have the potential to reach greater sensitivity than the traditional ‘gold standard’ of histopathology. There are currently two commercially available molecular tests for the intraoperative analysis of sentinel nodes in breast cancer: the GeneSearch BLN assay (Veridex, LLC, Warren, New Jersey, USA)10–14 and the OSNA assay (Sysmex, Kobe, Japan).10 11 Both of the assays work by amplification and detection of RNA not normally found in lymph nodes except in the presence of metastases. The GeneSearch BLN assay detects cytokeratin 19 (CK19), which is an epithelial marker, and mammaglobin (MG), which is a breast-specific marker, while OSNA detects CK19 only. There are also differences between the two assays in the RNA amplification technique employed, since OSNA utilises reverse-transcription loop-mediated isothermal amplification (RT-LAMP) while GeneSearch BLN utilises one-step quantitative reverse-transcription PCR (qRT-PCR).

The performance of the GeneSearch BLN assay has been assessed in single, multicentre and other studies,12–16 one of which was in the UK healthcare setting.10 The assay has US Food and Drug Administration approval and the EU kitemark for in-vitro diagnostic testing. However, there have been few reports to date on the implementation and use of the technique in a live clinical setting where clinical decisions regarding further axillary surgery following SLNB are influenced by the assay results. In addition, while there have been reports on the cost effectiveness of touch imprint cytology,17 the economic implications of intraoperative molecular analysis have not been previously reported. This observational study reports our experience of the use of the GeneSearch BLN assay in our first 254 patients in a live clinical setting to guide clinical decisions during the patients' operations regarding completion axillary surgery.

Given the importance of costs in modern health services, we used retrospective data to cost the three models of care currently in use: wide local excision (WLE) with axillary dissection, WLE with SLNB, and WLE with SLNB assessed intraoperatively. We also estimated the effects of each model on the income to the hospital and the host primary care trust (PCT) responsible for funding the service.

Methods

Study design and patient selection

This is a prospective observational study of implementation of the GeneSearch BLN assay in the clinical setting of a large district general hospital and cancer centre within the UK. Consecutive patients from November 2007 to December 2008 diagnosed with breast cancer who underwent SLNB in the Portsmouth Breast Care Centre were included. Patients with breast cancer considered suitable for SLNB undergo clinical and ultrasound examination of the axilla, with fine needle aspiration of any suspicious lymph nodes. Patients with axillary disease diagnosed prior to surgery do not have a SLNB and proceed directly to axillary clearance. Once a diagnosis of breast cancer was made, patients were counselled regarding the intraoperative analysis as part of their routine management by clinicians and breast care nurses. Specific patient information leaflets have been produced, and all patients were made aware of the possibility and implications of discordant results between histopathology and the intraoperative analysis. In particular, patients explicitly consented to intraoperative analysis and subsequent axillary clearance, guided by the results of the analysis. Results were reported after 254 patients, as this provided sufficient data to document the learning curve to provide validation of the technique in a clinical setting within our unit and to document the implementation of this method within routine clinical practice.

Surgical management

All surgery took place at Royal Hospital Haslar, Gosport, UK. Sentinel node biopsy was performed according to New Start guidelines.2 Initially 6–40 MBq 99mTc-labelled nanocolloid (Nanocoll; GE Healthcare, Slough, UK) was injected intradermally into the peri-areolar tissue prior to surgery and lymphoscintigraphy was performed. The dose of 99mTc was determined by the scheduling of the isotope injection and surgery. Before surgery, Patent V Blue dye (Laboratoire Guerbert, Paris, France) was injected into the sub-areolar tissue immediately following anaesthesia. Sentinel nodes were harvested, the excess fat was trimmed off, and the nodes were handed to the laboratory scientist performing the intraoperative analysis. While the GeneSearch BLN assay was being carried out, the surgeon performed the breast tumour resection.

Analysis of sentinel nodes by the GeneSearch BLN assay

A dedicated laboratory was set up in the room immediately adjacent to the breast operating theatre. Two qualified biomedical scientists underwent a 3-day training course and evaluation hosted by Veridex before the method was implemented. Sentinel nodes were analysed by the GeneSearch BLN assay according to the manufacturer's instructions and as previously reported.12–16 The nodes were sliced at 2 mm intervals with alternate slices of node undergoing molecular analysis, and remaining slices sent for routine histology. For molecular analysis, the slices were weighed and homogenised, and the RNA was extracted. GeneSearch BLN test kits, the Cepheid SmartCycler (Cepheid, Sunnyvale, California, USA) and associated proprietary software were used for RT-PCR analysis. The SmartCycler completes PCR 40 cycles before giving a result, which takes 24.5 min. Three marker genes were analysed in a single-well multiplex assay: CK19, MG and the housekeeper gene porphobilinogen deaminase, which is constitutively expressed in lymph node tissue and was used as a control. All assays included positive and negative controls to assess any possible contamination and to ensure the validity of the run. The product of the RT-PCR reaction was quantified by fluorescence in the Cepheid SmartCycler. Individual reactions were defined as positive if a predefined threshold (cycle threshold: MG ≤31, CK19 ≤30, porphobilinogen deaminase <36) was exceeded. Overall assay results for each node were classified as positive, negative or, if the controls failed, invalid.

Analysis of sentinel nodes by histopathology

Slices from sentinel nodes were fixed in 10% neutral buffered formalin and embedded in paraffin wax in a vacuum processor according to standard protocols. Sections were cut from each paraffin wax block at three levels at 100 μm intervals and stained with H&E. Slides were then examined by a breast pathologist. In most cases where there was no obvious metastasis and where there was discordance between histopathology and the BLN assay, spare sections were immunostained with CK19 (RCK108; Dako, Glostrup, Denmark). This was performed on an automated immunostainer after epitope retrieval.

Cost analyses

Cost comparisons (before and after implementation of intraoperative SLN diagnosis) were made between three different models of treatment (figure 1). Model 1 included patients who were all treated by primary axillary dissection with no SLNB prior to introduction of this method. Model 2 included patients who were treated by SLNB with routine histopathology of the node. Patients were treated this way prior to the introduction of the intraoperative assessment. In this model, node-negative patients received no further surgical treatment, while those with positive SLNs received delayed completion axillary clearance as a second operation. Model 3 represents the prospective series of 254 patients who were treated by SLNB with intraoperative analysis. In this pathway, node-negative patients received no further surgical treatment, while immediate completion axillary clearance was performed at the same operation if the SLN was positive in the GeneSearch intraoperative assay.

Figure 1

Flowcharts demonstrating the various possible patient pathways considered in the health economic calculations. AxCl, axillary clearance; MDT, multidisciplinary team; SLNB, sentinel lymph node biopsy; SNB, sentinel node biopsy; WLE, wide local excision.

Prior to November 2007, there was insufficient nuclear medicine capacity to perform SLNB on all patients in our institution who were clinically and radiologically node negative, and so patients were treated with either axillary clearance or sentinel node biopsy and delayed completion axillary clearance if the sentinel node was positive on histopathology. Consecutive patients (n=30) for models 1 and 2 were therefore identified retrospectively from the period prior to introduction of the intraoperative analysis in November 2007. The GeneSearch BLN assay was introduced in November 2007, and nuclear medicine capacity for SLNB was increased. This enabled all patients who were clinically and radiologically node negative to undergo SLNB and intraoperative analysis after this time. Only patients undergoing WLE for their breast cancer resection were included in the analysis.

For operations in each group, time in the operating theatre was obtained from the hospital TheatreMan (Trisoft, Chesterfield, UK) computer database. Cost per unit time in theatre, including usage of consumables, was then calculated by the hospital finance department. The number of calendar days spent in hospital for operations in each group was obtained from the hospital patient administration system (iSOFT, Banbury, UK) and cost per day on the ward was then calculated by the hospital finance department. The nuclear medicine costs of the SLNB procedure had been previously calculated and updated as part of a recent business plans for the procedure and these costs were used. Similarly we used updated costs from the business plan for the intraoperative analysis and histopathological analysis. The patients who had a positive SLNB prior to the introduction of the intraoperative analysis required an additional multidisciplinary team (MDT) discussion (3 min MDT time) and an additional MDT outpatient appointment (10 min with consultant surgeon and nurse chaperone/breast care nurse) to discuss a delayed axillary clearance. Additionally patients who underwent an axillary clearance in our unit had a 10 min physiotherapy session on the ward.

The histological node positivity rate within this cohort of 254 patients undergoing intraoperative sentinel node analysis was 28%. This figure for node positivity is within the expected range from the UK New Start programme and the Almanac trial. This figure was therefore used as the assumed node positivity rate across models 1 and 2 for the cost calculations, enabling an equal comparison between groups. Based on an average breast unit throughput of 300 patients per annum, we have also calculated illustrative costs to hospital trusts for each pathway, and the income expected from the PCT.

Data collection and statistical analysis

Information on the Portsmouth Breast Care Centre database is prospectively collected by a dedicated data clerk and contains details on all breast cancers diagnosed and treated within the unit. Information for a second database was also prospectively collected by the biomedical scientists within the cancer laboratory to collate the analytical results from all patients undergoing SLNB with intraoperative analysis; this database contains demographic information, information on operations performed, results of intraoperative analysis and results of histopathological examination. Ethics approval was not required, and all data were handled in accordance with Portsmouth National Health Service (NHS) Trust procedures.

Results

In total there were 259 consecutive SLNB operations in 254 patients, including five patients with bilateral breast cancer in this study. The median age was 61 years (range 31–90 years). Patient characteristics are illustrated in table 1. In total, 213 tumours were treated with breast-conserving surgery, including those who underwent oncoplastic mammaplasties, and 46 tumours were treated with mastectomy with or without reconstruction. One patient was unable to have the intraoperative analysis, as there was no biomedical scientist available at the time of their SLNB, and so this patient was not included in the cohort. Three patients were excluded. One patient who had palpable lymphadenopathy and an axillary clearance with confirmatory intraoperative analysis of the palpable node was excluded as she did not have SLNB. The GeneSearch intraoperative analysis of this node was positive and concordant with histopathology, which demonstrated nodal involvement with tumour (macrometastasis). One patient was excluded because of recording/documentation error such that it was not possible to correlate histopathology to the intraoperative analysis, and one patient because of failed GeneSearch quality controls.

Table 1

Patient demographics and tumour characteristics for the first 254 patients undergoing GeneSearch intraoperative molecular analysis of sentinel nodes

A total of 491 nodes were analysed by GeneSearch in these patients. It was not possible to analyse concordance for 24 of these nodes either because of failed GeneSearch quality controls (n=6) or because no nodal tissue was found (n=18), leaving 467 nodes with GeneSearch and histopathology data. In total, 103 of these nodes were positive by intraoperative analysis in 77 patients (78 breast cancers), including one patient with bilateral breast cancer and positive nodes bilaterally. In total, 72 cancers were histologically node-positive (table 2) in 71 patients. There was a 100% detection of macrometastases within sentinel nodes analysed by GeneSearch, although one cold undyed positive sentinel node was not submitted for GeneSearch analysis and so has been classified as a discordant result. When SLNB with intraoperative GeneSearch analysis was compared with the final histological nodal status, the overall concordance was 95%, with a sensitivity of 96% and specificity of 95%. The negative predictive value was 98%, and the positive predictive value was 88%.

Table 2

A comparison of tumour nodal final nodal status, including axillary clearance where appropriate, by histopathology, and by GeneSearch intraoperative molecular analysis of sentinel nodes

All patients with discordance between GeneSearch analysis and histopathology of individual SLNs are illustrated in table 3. Discordance of individual sentinel nodes did not in all cases lead to discordance between the overall GeneSearch nodal status and histopathological nodal status in individual patients. It was felt that the most clinically meaningful analysis would be on a cancer by cancer basis comparing the GeneSearch nodal status and histopathological nodal status in individual breast cancers, since it is on this basis that clinical decisions regarding immediate completion axillary clearance are made.

Table 3

A summary of all patients with discordant results, including those where individual nodes were discordant although overall patient results were concordant

The GeneSearch assay had a sensitivity of 100% for macrometastases. However, there was one involved cold undyed SLN not submitted for GeneSearch analysis, and this led to overall discordance in the nodal status on that patient, although GeneSearch and histology were concordant on the SLNs analysed. We have classified this as a discordant result. There were two SLNs in two patients that were negative on GeneSearch analysis but contained single subcapsular micrometastases on histology and were therefore discordant. There were nine patients who were positive on GeneSearch analysis and negative on histopathology. In one of these nine patients isolated tumour cells were found on histopathology, but we have still classified this as a discordant result. There were two patients in whom the GeneSearch analysis was positive and an immediate axillary clearance was performed on the basis of this. In these patients while these SLNs were negative on histology, macrometastases were found in further nodes at axillary clearance and these are classified as concordant.

Following validation in the first 125 patients, the assay was used consistently to guide clinical decisions regarding immediate completion axillary clearance (figure 2). While following the first 125 patients there was still a minority of patients who underwent delayed axillary clearance, in the last 60 patients in the cohort, there were 17 patients who were GeneSearch positive, and all 17 underwent immediate axillary clearance. Health economic analysis is based on the assumption that 100% of GeneSearch-positive patients undergo immediate axillary clearance and this appears to be the case now that the technique has been fully implemented.

Figure 2

Axillary clearances, immediate or delayed, expressed as a percentages of histologically node-positive tumours within the validation and implementation phases. There were a total of 72 node-positive tumours: 30 within the initial validation phase from November 2007 to July 2008 (first 125 patients) and 42 in the implementation phase. SNB, sentinel node biopsy.

The cost per patient of each model of care is summarised in table 4. The cost to the hospital of the initial operation to remove the tumour and sample nodes rose with the introduction of SLNB from £2465 to £2891 per patient, a 17% increase. In the absence of intraoperative assessment, the introduction of SLNB increased the number of patient episodes by 28%, as those patients with positive SLN on histology had to return for completion. The addition of PCR reduced this slightly to £2833, but the number of patient episodes remained at 300, as patients did not have to return for completion axillary dissection, though of course theatre time per patient was increased as patients with PCR-positive lymph nodes remained in theatre for immediate completion axillary dissection. The costs to pathology (table 5) for BLN assay are largely due to the high cost of the assay kit. Estimates for frozen section and imprint cytology are included for comparison, though these will vary between centres depending on their practice.

Table 4

Comparative costing by model of care for an individual patient

Table 5

Estimated costs (£) of intraoperative sentinal lymph node assessment by PCR (BLN assay), frozen section and imprint cytology

The effects on the financial position of the hospital and the funding PCT of a notional but typical 300 patients per year are shown in table 6. In each model of care, 300 patients are treated, but in models 1 (primary axillary dissection) and 3 (intraoperative SLN), 300 procedures are involved. Model 2 (delayed assessment of SLN by histology) involves 384 procedures, as 84 patients have to be operated on a second time. The effect of paying the hospital at the rate of £2549 per procedure means that the hospital loses money in models 2 and 3. By contrast the PCT loses in model 2 due to extra surgical procedures being performed, but gains in model 3 compared with model 2 by the number of cases remaining at 300. Thus the current payment system provides disincentives for hospitals to adopt PCR, but incentives for PCTs to favour its introduction.

Table 6

Estimated effects on hospital and the funding primary care trust (PCT) costs assuming a service with 300 cases per annum

Sensitivity analysis shows that the costs are dependent on the number of operations performed rather than the cost of the operation, to which the intraoperative PCR procedure adds around £300. The projections suggest that main savings made by introducing the intraoperative analysis are in terms of hospital bed utilisation. There are also some savings in theatre time due to the reduction in the need for second operations.

Overall these savings offset the cost of GeneSearch intraoperative analysis such that there is a cost saving to the NHS of £58 per patient in comparison with the current standard in most hospitals of SLNB with later histological examination. However due to the nature of the payment by results system, hospital trusts lose money as they are not be able to claim the second tariff for the completion axillary clearance. Prior to introduction of intraoperative analysis, completion axillary clearance is done as a second procedure and so attracts a second tariff for the hospital trusts. Our data suggest that the change in practice from immediate axillary dissection to SLNB costs the NHS £426 per patient, but that by incorporating intraoperative molecular diagnosis the additional cost is reduced to £368. This is currently offset by the additional tariff payable for the second operation in 28% of breast cancer patients. Therefore while the GeneSearch analysis saves money for the NHS overall, it could lead to a loss in revenue to hospital trusts, estimated at £284 per patient, while generating savings within primary care trusts of £714 per patient.

Overall, cost effectiveness is improved by intraoperative testing. Total NHS costs of treating this group of patients are reduced, and clinical/patient outcomes are improved due to 28% of the patients in our series avoiding a second operative procedure.

Discussion

Accurate intraoperative analysis of SLNs enables immediate completion axillary clearance to be performed at the same operation in node-positive patients. Node-negative patients avoid the morbidity and extra hospital stay associated with an axillary clearance and node-positive patients avoid a second operation. There have been various studies examining this technology12–16 and the OSNA assay,10 11 but with one exception18 these are in the context of a research setting. This is believed to be the first report of the assay in a UK healthcare clinical setting where clinical decisions are based on the results of the assay. Our data in an unselected cohort of consecutive patients with breast cancer undergoing sentinel node biopsy demonstrate the feasibility of applying the technique in a ‘real world’ clinical setting. The reproducibility of histopathological assessment of sentinel nodes is not perfect19 and varies with different published interpretations of the tumour, node, metastases system.20 The level of agreement, with an overall concordance of 95% found here between histology and GeneSearch BLN assay, appears to be on a par with the interobserver variability of histopathological assessment of breast biopsies,21 demonstrating that the GeneSearch technique is extremely accurate. Indeed the sensitivity of the ‘gold standard’, histopathology, for diagnosing metastases in lymph nodes is less than 100% and increases with further serial sectioning, immunohistochemistry and training.22 23 Additionally the sensitivity of other intraoperative techniques such as frozen section or touch imprint cytology is significantly less than that obtained here for GeneSearch.7 The lower sensitivity found with these pathological techniques is likely to be because the proportion of the node sampled is much less. We believe sampling differences account for the majority of discordant results. Nodes are sliced at 2 mm intervals and alternate slices sent for GeneSearch and histopathology. It is possible that a micrometastasis may be entirely within one slice and not within any of the others. Consistent with this, the only ‘false-negative’ results of the GeneSearch assay were when micrometastases were found entirely within the slice sent for histopathology. Furthermore, in two patients out of the 11 for whom GeneSearch was positive but histopathology of the sentinel nodes was negative, macrometastases were found within other nodes following axillary clearance. This is similar to the published estimate from a meta-analysis of studies that 20% of patients with micrometastases in sentinel nodes have macrometastases elsewhere in the axilla.24

This series of patients demonstrates that the GeneSearch assay is oncologically acceptable. The GeneSearch analysis did not lead to undertreatment of any patient during the course of the study, since it was possible for all node-positive patients to undergo an axillary clearance, whether immediate following validation of the technique or as a delayed procedure during the validation phase. During the validation phase, a small proportion of patients, following discussion with the patient and at the MDT meeting, declined delayed axillary clearance if, for example, one sentinel node contained a micrometastasis only on histopathology, or they opted for axillary radiotherapy to avoid a second operation. Six patients in this cohort who were histopathologically node negative underwent axillary clearance. In total it is possible that nine patients (3%) could have undergone an unnecessary immediate clearance when arguably it was not required, although these patients probably had micrometastases within the SLN that were not detected by histopathology due to inherent sampling problems. The numbers of patients undergoing axillary clearance are illustrated in figure 2. Following validation of the technique within the first 125 patients within our institution it proved possible to perform an immediate axillary clearance on the majority of GeneSearch-positive patients. In the last 60 patients included, all GeneSearch-positive patients underwent immediate axillary clearance.

Our experience here suggests that there are various logistic hurdles to overcome during introduction of the technique. It is important that the whole team is involved and understands the procedure prior to initiating the validation phase. This includes the whole MDT, the ward, pre-admission clinic, theatres, breast care and clinic nurses, and physiotherapists, to ensure that the patient is appropriately counselled regarding intraoperative analysis at the time that surgery is initially discussed. We have, for example, produced information sheets and patient literature regarding the technique, and patients are explicitly consented to undergo intraoperative analysis. The possibility and implications of discordant results are discussed, as is the possibility of discovering that they are likely to be node positive if awakening from theatre with a drain following immediate completion axillary clearance. To date, no patient has refused to undergo the intraoperative analysis, and anecdotal patient feedback suggests that avoidance of a second operation is popular with patients. We perform clinical examination and a preoperative axillary ultrasound scan with or without fine needle aspiration so that patients who are proven to be node positive on preoperative investigations proceed directly to axillary clearance. In terms of theatre list planning, we perform the sentinel node biopsy first and the breast cancer resection while the node is analysed, providing an opportunity to train junior colleagues in breast cancer resection. It is possible to risk stratify lists according to likelihood of node positivity,25 but flexible theatre working and the understanding that there may be over-runs or under-runs is important. Overall, from our data, performing the intraoperative analysis does not add to the time taken to perform a WLE and SLNB, while in patients found to be node positive the intraoperative analysis and SLNB add a mean of only 14 min to a WLE and axillary clearance. In contrast, a delayed completion axillary clearance takes 100 min and requires a second general anaesthetic.

There are several assumptions that have been made in the health economic calculations. In our group of patients who underwent intraoperative analysis, the node-positive rate was 28% and we have used this figure in our health economic calculation for all three groups to ensure a fair comparison. This figure is similar to a 26% node positivity rate in the SLNB arm of the Alamanac trial1 and the 33% node positivity rate in the UK New Start programme.2 Similarly in this series we harvested a mean number of 1.9 nodes per sentinel node biopsy compared with a median of 2.0 in the SLNB arm of the Alamanac trial1 and a mean of 2.2 in the UK New Start programme,2 suggesting that our results are applicable to practice elsewhere in the UK. We therefore believe that these results from a large district general hospital can be applied to other similar institutions within the UK. We have chosen WLE as the context on which to base the calculations as it is in this context that we had the most data. Additionally it was felt that in this context intraoperative analysis was actually least likely to be cost effective since for mastectomy or breast reconstruction there would be no wait for the assay result—these operations take longer than it takes to obtain results from the assay, and this is not always the case for WLE. We used bed days for our analysis of hospital bed usage since we felt this was the most economically relevant factor. In our analysis an overnight stay therefore counted as two bed days since generally we were unable to admit patients to that bed for same day operation for those 2 days. Additionally we did not have exact and accurate data for times of admission and discharge on each day. It is felt that while this method of analysis leads to hospital stays that appear slightly longer than might be expected, the comparison would apply to all groups equally. Finally the costs of any possible complications such as return to theatre for evacuation of haematoma were not included, as these were assumed to apply to all groups equally.

The calculations of assay costs (table 5) take no account of their different levels of accuracy. Intraoperative molecular assessment of sentinel nodes using a commercial kit is considerably more expensive than the non-commercial alternatives of frozen section or imprint cytology, but the amounts are small in comparison with the cost of the operative procedure. Increased use of commercial kits or the development of non-commercial alternatives for molecular assessment of SLNs are likely to lead to a fall in costs. In the UK, there is a shortage of pathologists and there is therefore no incentive for pathologists to perform frozen sections or imprint cytology in the operating theatre as this takes them away from routine paraffin section reporting.

When savings in theatre time, and in particular hospital bed utilisation, are calculated, they offset the additional costs of the any type of intraoperative analysis such that it is cost-saving to the NHS. While the analysis described is based on the NHS system, intraoperative analysis would be expected to lead to these cost savings irrespective of the health economic or tariff system. There is however a disincentive to hospital trusts to implement this, as within the NHS the second tariff payable for a delayed completion axillary clearance is lost. The current funding model allows NHS trusts to claim for the second operation, so that for a service treating 300 patients, the NHS trust can claim for 384 procedures, whereas the second tariff is lost for those implementing intraoperative testing. This is despite the many advantages for node-positive patients including avoidance of an extra outpatient visit, MDT discussion and (in 28%) a second operation. All patients have the benefit of knowing that definitive axillary surgery will be performed as a ‘one stop’ process. Intraoperative analysis is particularly useful for immediate latissimus dorsi flap reconstructions, as it enables these patients to benefit from SLNB without the risks of a delayed axillary dissection following latissimus dorsi flap reconstruction and concomitant risks to the thoracodorsal pedicle, which supplies the flap. Node-positive patients avoid potentially hazardous re-do axillary surgery by having an immediate completion clearance rather than a delayed axillary clearance as a second operation, and this should minimise delays to adjuvant therapies. In conclusion, intraoperative sentinel node assessment provides many benefits to patients and to the NHS. However, it is an example of an innovation that many hospitals will be unable to take up unless the perverse incentives of the current funding system are addressed.

Take-home messages

  • Accurate intraoperative sentinel lymph node assessment by PCR enables axillary clearance to be completed immediately in node-positive breast cancer patients.

  • There is excellent (95%) concordance between histology and intraoperative PCR—discrepancies are likely to be due to sampling error.

  • Implementation saves patients a second operation, improves bed and operating theatre utilisation, and is cost effective for the National Health Service.

Acknowledgments

We would like to thank Di Eldred for maintaining the Portsmouth Breast Care Centre database and Kirsty Winslade for assistance with the health economic calculations. Portsmouth Breast Care Centre has acted as a mentor centre for the Department of Health National Technology Adoption Centre (NTAC) and we would like to thank NTAC for their support and advice.

References

Footnotes

  • The present address for R I Cutress: Southampton Breast Unit, Southampton University Hospitals Trust, Princess Anne Hospital, Southampton, UK.

  • Funding Other Funders: NHS.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.