Localization techniques for guided surgical excision of non‐palpable breast lesions
Abstract
This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:
To assess the therapeutic outcomes of any new form of guided surgical intervention for non‐palpable breast lesions against wire‐guided localization (WGL), the current gold standard.
Background
Description of the condition
Breast cancer is the most common form of cancer and second leading cause of death in women in Europe (Ferlay 2007). In the developed world, it is estimated that one in 11 women will develop breast cancer, with some figures quoting one in eight for the United Kingdom and one in nine for the United States. Early detection of breast malignancies decreases mortality and morbidity (Philips 2008). This has led to the development of wide‐spread mammographic screening and subsequently a rapid increase in early detected tumours that tend to be small and non‐palpable (Wilson 1993). Amongst the five invasive cancers per 1000 women, detected in screening, 2.7 (54%) were < 15 mm; with others reporting that more than one third of excised breast lesions were clinically occult (NHS Breast Screening Program 2003; Rahusen 2002). In small malignant lesions, current literature suggests breast conserving therapy, including lumpectomy or wide‐local excision together with adjuvant radiotherapy to the tumour bed, as generally suitable and yielding comparable results to mastectomy (Fisher 2002; Schwartz 2006). The challenge remains, however, in accurately and precisely locating small non‐palpable lesions intra‐operatively, for optimal therapeutic outcome.
Description of the intervention
Currently the most widely adopted approach (80% in one survey) in guided breast‐conserving surgery for excising non‐palpable breast lesions is wire‐guided localization (WGL). A wire is inserted to localize the lesion to be excised, commonly under stereotactic or ultrasonographic guidance and alternatively magnetic resonance imaging (MRI) or computerised tomography (CT) (Van Esser 2008).
Key disadvantages of WGL include 1) the presence of a foreign body at pathological assessment; 2) wire transection; 3) wire migration; 4) patient distress and discomfort; 5) injury associated with barbs; 6) pneumothorax; and 7) interference with the surgical approach (De Cicco 2002). This has led to the development of other guidance techniques including 1) radioguided occult lesion localization (ROLL); 2) intra‐operative ultrasound (IOUS) guided resection; 3) cryoprobe‐assisted localization (CAL); 4) carbon marking; 5) methylene blue dye marking; and 6) near‐infrared fluorescence optical imaging (Hirsch 1989; Luini 1998; Rahusen 2002; Rose 2003; Tafra 2006;Tromberg 2008). Amongst these, the ROLL technique, developed in 1998 at the European Institute of Oncology, Milan, has been demonstrated as a feasible technique and is increasingly popular in many countries (Audisio 2005; Rovera 2008). ROLL utilizes a non‐specific radioisotope, commonly technetium‐99m (99mTc), injected intra‐tumorally under stereotactic or ultrasonographic (US) guidance pre‐operatively, with the lesion detected intra‐operatively by a handheld gamma probe. The technique is highly feasible across many settings and variations of the technique have since evolved, including the use of a radioactive iodine (125I) seed (RSL) (Gray 2001; Hughes 2008; Nadeem 2005). The procedures of ROLL, a cryoprobe, carbon and methylene blue dye marking are relatively similar. Their differences are mainly in the agent used for marking the lesion.
In the United Kingdom, all breast lesions are investigated with the triple test of clinical examination, imaging by mammogram or ultrasound, or both, and cytology or histology involving fine‐needle aspiration or core biopsy, or both. Hence both clinicians and patients are aware of the preliminary pathology diagnosis prior to surgery, which may then be confirmed from the excised specimen. However, elsewhere the last step may be missed and an open surgical biopsy performed for suspicious breast lesions (Ernst 2002). For the purpose of this review, we will adopt a pragmatic approach by including studies aimed at completely excising the breast lesion for therapeutic or therapeutic combined with diagnostic purposes but not solely diagnostic purposes.
How the intervention might work
Recently there have been three reviews in the area. Van der Ploeg et al and Rovera et al both reviewed studies investigating ROLL versus WGL, whilst Jakub et al reviewed radioactive seed localization (RSL) versus WGL (Jakub 2010; Rovera 2008; Van der Ploeg 2008). Their findings were similar in that they concluded that ROLL and RSL are reliable and safe alternatives to WGL; and at least equivalent in accuracy, obtaining negative margins, re‐intervention rate and operation time. However, none of them evaluated all the forms of guided surgical techniques together and they included non‐randomised trials. Pleijhuis et al provides the most comprehensive overview of current modalities and future directions in breast‐conserving therapy for early breast cancer (Pleijhuis 2009). They described a number of studies evaluating most techniques of guided surgery. Unfortunately, this was not a structured review.
Why it is important to do this review
With the clinical setting witnessing a shift towards earlier non‐palpable breast lesions being detected through screening, we need to investigate whether the current gold standard in assisting surgical excision of these lesions, WGL, yields the best therapeutic outcome. The lack of evidence to support one specific guidance technique for excising non‐palpable breast lesions calls for a systematic review of the literature and meta‐analysis to synthesize the results of a number of smaller studies investigating the many forms of guided surgery, if appropriate.
Objectives
To assess the therapeutic outcomes of any new form of guided surgical intervention for non‐palpable breast lesions against wire‐guided localization (WGL), the current gold standard.
Methods
Criteria for considering studies for this review
Types of studies
Our main analysis will be targeting randomised controlled trials (RCTs) that compare WGL against another form of guided surgery for non‐palpable breast lesions, for therapeutic and diagnostic purposes.
In the absence of randomised trials, well designed cohort or case control studies will be considered for a narrative description of the evidence available.
Types of participants
Female patients of any age with a diagnosis of a breast lesion. The lesion should be non‐palpable on clinical examination and the patient deemed fit for surgical intervention. Baseline characteristics between each study will be reported.
Types of interventions
Surgical intervention assisted by WGL versus any technique of guidance aimed at the complete excision of the breast lesion for therapeutic or therapeutic combined with diagnostic purposes, but not solely for diagnostic purposes. Guidance techniques to be considered include, but are not limited to:
-
radioguided occult lesion localization (ROLL);
-
radioactive seed localization (RSL);
-
intra‐operative ultrasound (IOUS);
-
cryoprobe‐assisted localization (CAL); and
-
haematoma‐directed ultrasound‐guided (HUG) localization.
Types of outcome measures
Primary outcomes
-
Successful localization of the lesion ‐ using the planned technique to image and localize the lesion pre‐operatively.
-
Successful excision of the lesion ‐ excision of the lesion as planned.
-
Negative excision margins ‐ no cancer in mural edge.
-
Re‐intervention rate ‐ the need for further excision of primary lesion, not including axilla only interventions.
Secondary outcomes
-
Operation time
-
Length of hospital stay
-
Complications
-
Recurrence
-
Cosmesis
-
Patient preference (including as a result of pain)
Other outcomes of interest that will not be featured in our analysis include practicalities and cost.
Search methods for identification of studies
See:Breast Cancer Group methods used in reviews.
Only articles and trials published in English will be included.
Electronic searches
We will search the following databases.
(a) The Cochrane Breast Cancer Group Specialised Register. We will search the Cochrane Breast Cancer Group (CBCG) Specialised Register. Details of the search strategies used by the Group for the identification of studies and the procedure used to code references are outlined in the Group's module (www.mrw.interscience.wiley.com/cochrane/clabout/articles/BREASTCA/frame.html). Trials coded with the key words 'high risk', 'surgery', 'wire guided localisation', 'radioguided occult lesion localization', 'radioactive seed localization', 'intraoperative ultrasound', 'cryoprobe‐assisted localization' and 'haematoma‐directed ultrasound‐guided' will be extracted and considered for inclusion in the review.
(b) MEDLINE (via PubMed) (from July 2008 until search date). See Appendix 1 for the search strategy.
(c) Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library, current issue). See Appendix 2 for the search strategy.
(d) The WHO International Clinical Trials Registry Platform (ICTRP) search portal (http://apps.who.int/trialsearch/Default.aspx), for all prospectively registered and ongoing trials. See Appendix 3 for the search strategy.
Searching other resources
(a) Bibliography searching
We will try to identify further studies from reference lists of identified relevant trials or reviews. A copy of the full article for each reference reporting a potentially eligible trial will be obtained, where possible. Where this is not possible attempts will be made to contact authors to provide additional information.
Data collection and analysis
Selection of studies
Studies identified through the search strategy will be selected independently by two of the authors (BC and RJ) by scanning the title and the abstract; if this is inconclusive the full text will be examined for further assessment. Any discrepancies regarding eligibility will be resolved by discussion involving a third author (RA). Any excluded studies will be recorded in the 'Characteristics of excluded studies' table with their details and the reason for exclusion.
Data extraction and management
All suitable studies will be independently appraised by two of the authors (BC and RJ) using a standardized data extraction sheet, and summarised in the 'Characteristics of included studies' table. Data extracted from included studies will include: study design, study population size, patient baseline characteristics, specimen pathology, therapeutic or combined diagnostic and therapeutic intervention, details of intervention, and outcomes. Where necessary, additional information will be sought from the correspondent author of that study.
Assessment of risk of bias in included studies
Methodological quality will be independently assessed by two authors. RCTs will be assessed using the Cochrane Breast Cancer Group ‘Quality Assessment Form’ (Schulz 1995), available through The Cochrane Breast Cancer Group website. The ‘Risk of Bias Assessment Form’ will be used to assess the specific domain of risk, and the 'Risk of Bias' table will also be utilized for making judgements on the type of risk (low, high or unclear), as per the Cochrane Collaboration's tool for assessing risk of bias (Higgins 2009).
It is equally important to consider sources of bias in non‐randomized studies, however the assessment tools must take into account the varying susceptibilities and designs of these studies (Higgins 2009). Therefore, a Newcastle‐Ottawa scale customised to our review question will be employed as described in the Cochrane Handbook (Higgins 2009).
Potential bias may arise from different diagnostic protocols practiced as mentioned above, giving rise to heterogeneous pathologies, making it difficult to collaborate results on re‐intervention rates, recurrence, and excision margins.
Measures of treatment effect
A pooled estimate for the size of the effect, as specified below, and 95% confidence interval (CI) will be calculated combining all trials. In an effort to determine the likely beneficial therapeutic outcomes between various treatment approaches, dichotomous data will be measured using relative risk (RR) as well as risk difference (RD) for: successful localization, excision of lesion, negative margins, re‐intervention rate, and recurrence.
Mean differences (MD) and standardized mean differences (SMD), when applicable, will be used to measure continuous data: specimen volume and weight, operation time, length of hospital stay. Finally, the outcomes such as grade of complication (grade), cosmesis (grade), and patient compliance (preference and pain) will be categorized as short‐ordinal scales.
Time‐to‐event data will not be measured in this instance.
Unit of analysis issues
For any cluster‐randomised trials, outcomes will be assessed at the ‘group’ level, thus keeping the unit of analysis intact. Cross‐over trials are not anticipated given the topic, however if present they will be analysed using a paired t‐test where each participant acts as his or her own control (The Cochrane Collaboration, 2002). Multiple treatment groups are not anticipated; neither are studies using repeated observations (recurrence will still be captured as a secondary outcome using dichotomous variables (Yes or No), but not in relation to this section).
It will be important to consider effect measures for counts and rates as some participants may have multiple occurrences of the event (that is, excision of a non‐palpable breast lesion). This will be addressed through analysis using the rate ratio (RR), and not dichotomous data, in order to capture each individual event (Higgins 2009).
Dealing with missing data
This will be an available case analysis, where data will be analysed only for the known cases. In order to address the issue of missing data, all data will be obtained whenever possible by contacting the original investigator. In cases where data can be assumed to be missing at random, only the existing data will be analysed. All occurrences of known missing data will be presented and explored within the ‘Discussion’ section, as well.
Assessment of heterogeneity
Clinical heterogeneity will be assessed by examining variability in the participants, interventions and outcomes, as will methodological heterogeneity (that is, variability in study design and risk of bias) (Higgins 2009). A test of statistical heterogeneity will be conducted to detect any differences of effects between studies. This will be carried out 1) by reviewing the confidence intervals (CIs) for each of the individual studies in the systematic review in order to assess overlap. This will be ascertained via a forest plot (Bruce 2008); and 2) by formally conducting a statistical test for heterogeneity using the Cochran’s Q (chi‐squared test for heterogeneity) and I2 statistic (Bruce 2008).This will further assess whether the study estimates of effects are due to chance alone.
Heterogeneity will be addressed by first checking the data for accuracy. Following this, both the fixed‐effect and random‐effects models will be considered: if I2 is less than the defined threshold of 50%, then the fixed‐effect model (FEM) will be performed; however, if the I2 is greater than the defined threshold of 50%, then the random‐effects model (REM) will be performed (Higgins 2009). Additionally, if the results of both the FEM and REM are comparable, the REM will be reported because confidence intervals from REM meta‐analysis gives better coverage. Thus, if heterogeneity exists a meta‐analysis will still be conducted and possible explanations will be explored via the ‘Discussion’ section. Interpretation of these tests will be conducted within the guidelines outlined in the Cochrane Handbook for Systematic Reviews of Interventions 5.0.2 (Higgins 2009).
Assessment of reporting biases
In order to address the issue of reporting bias, risk assessment forms will be utilised in order to determine ‘level’ of risk (high, low, uncertain) and ‘type’ of risk (selection, attrition, performance, other). This information will then be entered into Review Manager software (RevMan 5), where funnel plots will be performed (Egger 1997). Asymmetrical results will be interpreted in accordance with guidelines set out in the Cochrane Handbook for Systematic Reviews of Interventions 5.0.2 (Higgins 2009). Results will be reported using risk of bias graphs and summaries in Review Manager. Risk ratios (RR) will also be plotted on a log scale. Since an asymmetrical funnel plot is not always the result of reporting bias (Sterne 2001; Sterne 2004), alternative reasons will be considered whenever possible. The results of the funnel plot and the possibility of reporting bias will be explored in the Discussion section.
Data synthesis
Continuous data: the inverse‐variance method will employ a fixed‐effect model for continuous data (MD and SMD, when appropriate depending on the scale) or the DerSimonian and Laird (1986) random‐effects model, as appropriate.
Dichotomous data: a random‐effects model employing the DerSimonian and Laird method (1986) will be used for dichotomous outcomes (RR). The Mantel‐Haenszel method will be used for a fixed effect analysis for relevant secondary outcomes.
All analyses will be performed using RevMan version 5 and in accordance with the Cochrane Handbook for Systematic Reviews of Interventions 5.0.2 (Higgins 2009) and Cochrane Systematic Review Workshop (2010). Results will be reported via a graphical assessment using forest plots showing the study estimates, as noted above. Additional subgroup and sensitivity analyses will also be reported.
Quantitative data which are not deemed appropriate for pooled analysis (that is, prospective cohort studies or case control studies) and relevant qualitative data will be presented in table format and described narratively.
Subgroup analysis and investigation of heterogeneity
A separate subgroup analysis was considered, however a search of the literature did not reveal evidence in support of clinical characteristics to be identified a priori.
Since missing data can also impact the subgroup analysis (Higgins 2009), this issue will be addressed as noted in the relevant section above (‘Dealing with missing data’). Interpretation of these findings will be conducted and presented cautiously and in accordance with guidelines set out in the Cochrane Handbook (Higgins 2009) and by Oxman and Guyatt (Oxman 1992).
In order to investigate whether a characteristic may be associated with the treatment effects, and depending upon the number of studies included in the analysis, a meta‐regression will be considered. This is unlikely given the nature of the test and the risk of yielding spurious conclusions.
Sensitivity analysis
A sensitivity analysis will be conducted in order to determine how sensitive (that is robust) the results of the meta‐analysis are to the studies that have been included, given differences in size, quality and other methodological variations of the studies. The results of this analysis will be reported via a summary table (Higgins 2009).
