| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Tristán Associates Harrisburg, PA 17111
Regarding the amount of tissue obtained during a stereotactic breast biopsy, the question "How much is enough?" has been discussed since the dawn of the minimally invasive breast biopsy era [1]. In the September 2002 issue of the American Journal of Roentgenology, Liberman et al. [2] conclude that attempting excision of all mammographic evidence of a lesion during vacuum-assisted stereotactic breast biopsy, as opposed to sampling the lesion, has little merit. Unfortunately, problems with selection bias preclude such a conclusion.
Liberman et al. [2] retrospectively analyzed 788 consecutive biopsies of lesions ranging in size up to 10 cm, during which no attempt was made either to excise or to avoid excision of the targeted lesion. The authors simply recorded whether they felt they had excised each lesion and reported no difference between the sampled and excised groups in rebiopsy rate or sparing surgery.
The potential for bias under these conditions should be obvious. Faced with a highly suspicious spiculated 2-cm mass, for example, few radiologists would find it necessary or even worthwhile to attempt complete excision during a stereotactic biopsy—even the first sample is highly likely to contain the expected result of infiltrating ductal carcinoma [3]. Stereotactic biopsy of such a lesion will almost invariably spare the patient one operation, because wide surgical excision and lymph node sampling can be accomplished during one surgical procedure. It is likely that these lesions were preferentially sampled, rather than excised, in the current study [2]. Therefore, the sampled group in all likelihood contained a disproportionate number of such lesions and consequently appeared to have a more favorable rate of sparing surgery than the excised group.
The data suggest that such a bias was in fact present. Table 2 in Liberman et al. [2] shows that cancer was 63% more common in patients in the sampled group, compared with those in the excised group. Because the diagnosis of cancer ordinarily increases with a greater volume of acquired tissue, the observed difference can only be explained by selection bias resulting from steering higher-suspicion lesions preferentially into the sampled group.
A better study design would be to select prospectively only those patients in whom excision is even feasible in the first place—why analyze merits or drawbacks of excision in instances when it is not possible to perform it?— and to randomly assign the patients to excision or sampling. Then a bona fide attempt at excision could be made in those patients assigned to the excision group, and outcomes in the two groups could be compared.
Paradoxically, excision improved the ability to diagnose atypical ductal hyperplasia (ADH) but detracted from its apparent performance, because a diagnosis of ADH leads to a recommendation for surgery. ADH was diagnosed 46% more frequently in the excised than in the sampled group [2]. Given that ductal carcinoma in situ, ADH, and usually hyperplasia constitute a spectrum, and given that obtaining a greater volume of tissue has been shown to lead to less-frequent mislabeling of ductal carcinoma in situ as ADH [4], it stands to reason that the greater volume of tissue acquired during excision ought to yield less-frequent mislabeling of ADH as usual hyperplasia, which could account for the difference in frequency of diagnosis of ADH.
The authors did not address ease of subsequent mammographic interpretation after the targeted lesion has been removed. I have seen several instances in which calcifications were reported to have been sampled at an outside institution and no marking clip was deposited because many calcifications were left in place. Management in these cases can be difficult if no objective evidence that appropriate sampling has been accomplished can be seen 6 months after the biopsy; deciding whether to recommend rebiopsy in these patients or to observe is problematic.
Finally, the wording of their conclusion is revealing: "Complete excision rather than sampling of the mammographic target was associated with a significantly lower frequency of discordance and a trend toward lower frequency of ductal carcinoma in situ underestimation but had no other significant advantage or disadvantage" [2]. These factors are two of the most important measures of stereotactic biopsy performance. Why minimize them? It is a bit like saying, "Cracking the Enigma code helped shorten World War II but had no other advantage or disadvantage."
In summary, selection bias in this study by Liberman et al. [2] nullifies the conclusion that attempting excision during stereotactic biopsy is of little benefit. Personal experience has shown otherwise, and a prospective study may be a more appropriate test of the hypothesis.
References
Memorial Sloan-Kettering Cancer Center New York, NY 10021
We thank Dr. Guenin for his interest in our article [1]. Unfortunately, he misstates our conclusions. In the first paragraph of his letter, he states that "Liberman et al. conclude that attempting excision of all mammographic evidence of a lesion during vacuum-assisted stereotactic breast biopsy, as opposed to sampling the lesion, has little merit." In the last paragraph, he speaks of our having made the "conclusion that attempting excision during stereotactic biopsy is of little benefit." The words "little merit" and "little benefit" are Guenin's, not ours.
Our study was a retrospective analysis of data from 800 lesions in 797 patients who underwent stereotactic 11-gauge vacuum-assisted biopsy. As we stated in the materials and methods section, no attempt was made specifically to achieve or to avoid complete excision of the imaging target. We compared lesions in which the imaging target was excised with lesions in which the imaging target was sampled with respect to a variety of parameters that can be used to evaluate the outcome of stereotactic biopsy: discordance, complete histologic removal of cancer, histologic underestimation, immediate and delayed rebiopsy, sparing surgery, and complications. The group in which the imaging target was excised, rather than sampled, had a significantly lower frequency of discordance, a significantly higher frequency of complete histologic removal of cancer, and a trend toward fewer ductal carcinoma in situ underestimates. Statistical analysis showed no other significant differences between the groups [1].
Different readers may interpret our results differently. Some may interpret the data to indicate that because the potential benefits of complete excision would impact on few women (imaging–histologic discordance and ductal carcinoma in situ underestimation each occurred in approximately 2% of women who had stereotactic biopsy during the study period), it is not worthwhile to attempt it. Others may interpret the data to indicate that complete excision is preferable because it offers a few potential clinical advantages (lower discordance and trend toward fewer ductal carcinoma in situ underestimates) without any demonstrable disadvantage. We did not endorse either interpretation in our article, but presented the facts [1]. Guenin has interpreted the data to indicate that "attempting excision during stereotactic biopsy is of little benefit," and then rejected that conclusion because his "personal experience has shown otherwise."
With the advent of percutaneous biopsy technology capable of removing larger volumes of tissue, some investigators have suggested that complete removal of the imaging target is preferable [2], but no data prior to ours have specifically addressed the potential advantages or disadvantages of this approach. Retrospective analyses have limitations, including potential selection bias [3]. However, retrospective analyses still have clear scientific advantages over anecdotal personal experience because they use existing data to assess the value of clinical interventions [3]. Medicine cannot be evidence-based unless someone gathers the evidence.
The "Goldilocks" question—how much tissue is too much, how much is not enough, and how much is just right?—has been asked since the advent of percutaneous imaging-guided breast biopsy. As indicated in the concluding paragraph of our article, further study, including long-term follow-up of lesions excised versus sampled at stereotactic biopsy, is needed to resolve this question [1]. A prospective investigation may add valuable information. Neither our statistical analysis of data from 800 lesions nor Guenin's personal experience should constitute the last word.
References
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
