NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Bruening W, Schoelles K, Treadwell J, et al. Comparative Effectiveness of Core-Needle and Open Surgical Biopsy for the Diagnosis of Breast Lesions [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2009 Dec. (Comparative Effectiveness Reviews, No. 19.)

Cover of Comparative Effectiveness of Core-Needle and Open Surgical Biopsy for the Diagnosis of Breast Lesions

Comparative Effectiveness of Core-Needle and Open Surgical Biopsy for the Diagnosis of Breast Lesions [Internet].

Show details

3Results

Question 1. In women with a palpable or non-palpable breast abnormality what is the accuracy of different types of core-needle breast biopsy compared with open biopsy for diagnosis?

Evidence Base

Our literature searches identified 1,224 potentially relevant articles. After review of the abstracts, the full-length articles of 589 of these studies were obtained and examined in full. Of these, 107 studies met the inclusion criteria for Key Question 1. The excluded studies and primary reason for exclusion are shown in Appendix C. The studies are briefly described in Table 5. Full Full details about the included studies, the enrolled patients, the biopsy methods, and the characteristics of the breast lesions are shown in the evidence tables in Appendix E.

Table 5. Studies addressing Key Questions 1 and 2.

Table 5

Studies addressing Key Questions 1 and 2.

Thirty-five of the 107 studies were prospective in design. Forty-nine were conducted in the United States. Ninety-three were carried out in general hospitals. A total of 57,088 breast lesions were enrolled in the 107 studies. The overall quality of the entire evidence base was rated as low (median score 6.1, range 3.6 to 8.2); see Table 5 for details.

Accuracy of Open Surgical Biopsy

Obtaining information on the accuracy of open surgical biopsy was, not surprisingly, difficult. Practically all authors and experts assume that open surgical biopsy is 100% accurate. We did not identify any clinical studies of open surgical biopsy that met our inclusion criteria (see Methods section).

We identified an article by Antley et al. 1998 that reviewed the accuracy of open surgical biopsy.40 Antley et a l. reviewed the available information (published literature as well as patient charts available in the author’s medical center) on the accuracy of open surgical biopsy and concluded that open surgical biopsy has been reported to miss 1 to 2% of breast cancers (a sensitivity of 98% or greater). This estimate is based upon a re-review of archived open biopsy material by a second pathologist, the charts reviewed by Antley et al., a study of cases of benign results on biopsy after a very suspicious mammogram, and expert opinion.41–43

We did not identify any information on estimates of underestimation rates for open surgical biopsy. However, underestimations are generally thought to be due to failure to sample all important areas of a lesion. For example, a lesion may contain a foci of carcinoma within a cluster of atypical cells. Biopsy samples collected by core-needle may fail to sample any of the carcinoma cells, leading to an underestimation. Because open surgical biopsy samples most or all of the lesion, in theory underestimations should not occur. Therefore, we have assumed that open surgical biopsy has a zero, or close to zero, underestimation rate.

Accuracy of Core-Needle Biopsy

We attempted to fit a bivariate binomial regression model to the data reported by all 107 studies but the data were too heterogeneous to allow a valid model to be fit. Due to obvious differences across studies of biopsy methods and enrolled patient populations, we did not perform further analyses on the full set of data. In the following analyses we have grouped the studies by the type of core-needle biopsy used in the study. The analyses are summarized in Figure 1 A through Figure 4 D in the Executive Summary and in Table 6 and Table 7. Full details of the analyses and reported data are provided in Appendix F.

Table 6. Summary of accuracy by type of biopsy procedure.

Table 6

Summary of accuracy by type of biopsy procedure.

Table 7. Summary of the impact of factors on accuracy.

Table 7

Summary of the impact of factors on accuracy.

Freehand Core-Needle Biopsies

Five studies reported data on the accuracy of non-guided, i.e., freehand, core-needle biopsies performed with automated biopsy gun devices.44–48 We fit a bivariate binomial model. There was very little heterogeneity in the data (I2 = 6.95%). The summary sensitivity was 85.8% (95% CI: 75.8 to 92.1%) and the summary negative likelihood ratio was 0.143 (95% CI: 0.082 to 0.250). This ratio indicates that for a woman with a pre-test probability of malignancy of 30%, her probability of having malignancy after a negative freehand core-needle biopsy would be 5.8%. A pre-test probability of 30% was chosen because the average woman undergoing core-needle biopsy has been categorized as BI-RADS 4 before undergoing the biopsy, and such women have an approximate overall prevalence of malignancy of 30%.15 We have used the 30% pre-test probability in the analyses that follow for the same reason. However, it is important to realize that each individual woman’s pre-test probability may vary from this estimate.

None of the studies reported underestimation rates. Because there were only five studies we did not perform any sub-group or meta-regression analyses.

Cusick et al. noted that smaller lesions(less than 2 cm in diameter) were more likely to be misdiagnosed.48 In contrast, Barretoet al. commented that neither tumor size nor patient age affected the accuracy of the procedure; however, tumors located in the right breast were much more likely to receive false-negative diagnoses, perhaps due to the fact that the persons performing the biopsy procedures were right handed.47 Barreto et al. also noted that operator inexperience was a key factor in misdiagnoses.47 The apparent difference in conclusions about the impact of tumor size on biopsy accuracy is probably due to the fact that the tumors in the study by Barreto et al. were all larger than 2 cm in diameter.

We graded the conclusions from this evidence as Low. The quality of the evidence base was rated as Low (median score 5.7), but quantity, consistency, and robustness were all rated as Sufficient.

Ultrasound Guided Automated Gun Core-Needle Biopsies

Sixteen studies of 7,124 biopsies used ultrasound guidance and an automated biopsy gun.49–64 We could not fit a bivariate binomial model due to heterogeneity. The random-effects model found a summary sensitivity of 97.7% (95% CI: 97.2 to 98.2%) and a summary negative likelihood ratio of 0.030 (95% CI: 0.022 to 0.040). This ratio indicates that for a woman with a pre-test probability of malignancy of 30%, her probability of having malignancy after a negative ultrasound-guided automated gun core-needle biopsy would be 1.3%. Twelve of the sixteen studies reported data on atypia underestimation rates.49,51–53,56–61,63,64; the summary atypia underestimation rate was 29.2% (23.4 to 35.9%). Twelve studies reported data on DCIS underestimation rates.49,51–53,55–61,64 the summary DCIS underestimation rate was 35.5% (27.1 to 45.0%). We graded the conclusions from this evidence as Low. The quality of the evidence base was rated as Low (median score 6.1), but quantity, consistency, and robustness were all rated as Sufficient.

We then proceeded to explore factors that might affect the accuracy of the biopsies by performing meta-regressions. We only performed meta-regressions if all of the studies reported information about the factor being analyzed and at least three studies were different from the rest of the studies for that factor.

Patient and Breast Lesion Factors

The studies reported insufficient information about characteristics of the lesions or the patients to explore the impact of these factors on the accuracy of the biopsies.

Biopsy Procedure Factors

Only seven of the studies reported information about patient position during the procedure, and six of these reported the patients were supine49,51,56,60,62,63 while the seventh reported the patients were seated.58 All but two of the studies reported using a 14G needle ; one of these two studies used an 18G needle, and one used different sizes of needles for different patients.59,62

Three of the fifteen studies verified all core-needle findings with surgery50,58,62 (the rest used a combination of surgery and patient followup), and six of the studies did not follow all patients for at least two years.52,53,55,57,60,63 Meta-regression did not find a statistically significant impact of methods of verification of biopsy on the accuracy of the biopsies.

One study, de Lucen et al., evaluated the impact of number of cores taken on the accuracy of the procedure. The authors of the study reported that taking more than 2 cores did not improve the accuracy of the procedure.50 However, Fishman et al. reported that taking more than 2 cores did improve the accuracy of the biopsy, with 4 cores being the optimal number.55 Fishman et al.’s conclusion was based on one case of DCIS that would have been missed if fewer than 4 cores had been taken; the other 13 tumors identified in the study would have been correctly diagnosed if only 2 cores had been taken. de Lucen et al.’s conclusion was based on the fact that the six tumors (out of a total of 101 tumors identified in the study) that were falsely diagnosed as benign by core-needle biopsy would not have been correctly diagnosed even if up to six cores were taken.

Clinician and Facility Factors

All but one of the studies were performed in general hospitals. The studies were conducted in settings around the world; meta-regression did not find a statistically significant effect of geographic location on the accuracy of the biopsies. Most of the studies did not report data about the training or experience of the persons performing the biopsies.

Stereotactic-Guided Automated Gun Core-Needle Biopsies

Thirty-three studies of 7153 biopsies used stereotactic guidance and an automated biopsy gun.65–97 We were able to fit a bivariate binomial model. The summary sensitivity was 97.8% (95% CI: 95.8 to 98.9%) and the summary negative likelihood ratio was 0.022 (95% CI: 0.012 to 0.043). This ratio indicates that for a woman with a pre-test probability of malignancy of 30%, her probability of having malignancy after a negative stereotactically-guided automated gun core-needle biopsy would be 0.9%. Twenty-six of the 33 studies reported data on atypia underestimation rates.65–69,71,73–90,94,96 and 17 reported data on DCIS underestimation rates. 65–68,73,75,76,78–81,85,86,89–92 The atypia underestimation rate was 43.5% (95% CI: 35.7 to 51.7%) and the DCIS underestimation rate was 24.4% (95% CI: 18.0 to 32.1%). We graded the conclusions from this evidence as Low. The quality of the evidence base was rated as Low (median score 6.1), but quantity, consistency, and robustness were all rated as Sufficient.

We then proceeded to explore factors that might affect the accuracy of the biopsies by performing meta-regression. We only performed meta-regressions if all of the studies reported information about the factor being analyzed and at least three studies were different from the rest of the studies for that factor.

Patient and Breast Lesion Factors

Koskela et al. reported zero false-negatives out of 97 procedures performed on lesions detected as masses on mammography but 4 false-negatives out of 108 procedures performed on lesions with microcalcifications.66 Walker et al. reported that the sensitivity of core-needle biopsy was much lower for microcalcifications than for any other type of lesion.85

The majority of the studies appeared to have enrolled patients with only non-palpable lesions but many of the studies did not report on the palpability of the lesions. The studies reported insufficient information about other characteristics of the lesions or the patients to explore the impact of these factors on the accuracy of the biopsies.

Biopsy Procedure Factors

All but three of the studies used 14G needles,78,79,92 and meta-regression did not find a statistically significant impact of needle size on biopsy accuracy. Twenty-two of the studies reported that the patients were prone,65,67,68,70,73,75,76,80,82–84,86–96 three reported the patients were seated,66,69,72 one reported the patients were in the decubitus position, 77 one reported patients were either prone or seated,81 but six did not report information about patient positioning.71,74,78,79,85,97

Eight of the studies verified all core-needle findings with surgery68,72,73,79,94–97 (the rest used a combination of surgery and patient followup), and 22 of the studies did not follow all patients for at least two years.66,67,69,70,74–78,80–87,89–93 Meta-regression did not find a statistically significant impact of methods of verification of biopsy on the accuracy of the biopsies.

Koskela et al. reported that more than three cores need to be taken from lesions before an accurate diagnosis can be made.66

Clinician and Facility Factors

Twenty-nine of the studies were conducted at a single center (the other four were multi-center studies65,68,70,79). Twenty-six of the studies were conducted in general hospitals,65–69,71–76,78,81–83,85–87,89–92,94–97 four were conducted in free-standing dedicated cancer centers,77,80,88,93 one was conducted in a breast cancer screening clinic,84 and one was conducted in multiple centers of different types.70 Twenty of the studies were conducted within the United States 70,75,76,79–83,86–97 and the rest were scattered worldwide. Meta-regressions did not find that any of these factors had a statistically significant impact on biopsy accuracy.

The majority of studies reported that radiologists performed the biopsies, but many studies did not report information about the training of the operators. Very few of the studies reported the degree of experience of the operators or their caseloads.

Ultrasound-Guided Vacuum-Assisted Core-Needle Biopsies

Seven studies of 507 biopsies used ultrasound guidance and a vacuum-assisted device to perform breast biopsies.56,98–103 There was no significant heterogeneity in the data (I 2= 0.0%). We fit a bivariate binomial model to the data. The summary sensitivity was 96.5% (95% CI: 81.2 to 99.4%) and the summary negative likelihood ratio was 0.036 (95% CI: 0.006 to 0.212). This ratio indicates that for a woman with a pre-test probability of malignancy of 30%, her probability of having malignancy after a negative vacuum-assisted ultrasound-guided core-needle biopsy would be 1.5%. The studies reported no cases of atypia underestimation and only a single case of DCIS underestimation.56 We graded the conclusions from this evidence as Low. The quality of the evidence base was rated as Low (median score 5.9), but quantity, consistency, and robustness were all rated as Sufficient.

Due to the lack of heterogeneity in the data, we did not perform any meta-regressions to explore the impact of factors on accuracy. The following differences between studies do not appear to affect accuracy.

Patient and Breast Lesion Factors

The studies reported very little information about the patients or lesions.

Biopsy Procedure Factors

All of the studies verified core-biopsy results by a combination of open surgery and patient followup. Only one of the seven studies followed all patients for at least two years.98

Five of the studies used the Mammotome device with an 11G needle,56,99,101–103 one study used a VACORA device with a 10G needle,98 and one study did not report information about the device or needle gauge.100 Four of the studies reported the patients were supine 56,100,102,103 and the others did not report details of patient positioning.

Clinician and Facility Factors

Two of the studies were conducted in free-standing cancer centers101,103 and the others were performed in general hospitals. The studies were conducted in many different countries worldwide. The studies generally did not report information on operator training or experience.

Stereotactic-Guided Vacuum-Assisted Core-Needle Biopsies

Twenty-two studies of 7,153 biopsies used stereotactic guidance and a vacuum-assisted device to perform core-needle biopsies.76,80,104–123 We were able to fit a bivariate binomial model. The summary sensitivity was 99.2% (95% CI: 98.1 to 99.6%) and the summary negative likelihood ratio was 0.009 (95% CI: 0.004 to 0.021). This ratio indicates that for a woman with a pre-test probability of malignancy of 30%, her probability of having malignancy after a negative vacuum-assisted stereotactically-guided core-needle biopsy would be 0.4%. All of the studies reported information about atypia and DCIS underestimation rates. The summary atypia underestimation rate was 21.7% (95% CI: 17.7 to 26.4%) and the summary DCIS underestimation rate was 12.9% (95% CI: 11.1 to 15.1%). The low DCIS underestimation rate may affect treatment planning. The surgeon performing the followup open surgical procedure can be reasonably confident that a malignant tumor is not present, and therefore may plan to remove the lesion using a breast-conserving approach, and may decide to not sample the axillary lymph nodes. Some women and physicians may decide that the ADH underestimation rate is low enough to safely substitute surveillance for an open biopsy procedure after diagnosis of ADH on core-needle biopsy. We graded the conclusions from this evidence as Low. The quality of the evidence base was rated as Low (median score 6.1), but quantity, consistency, and robustness were all rated as Sufficient.

We then proceeded to explore factors that might affect the accuracy of the biopsies by performing meta-regressions. We only performed meta-regressions if all of the studies reported information about the factor being analyzed and at least three studies were different from the rest of the studies for that factor.

Patient and Breast Lesion Factors

Two studies reported that stereotactic-guided vacuum-assisted core-needle biopsy was equally accurate for lesions with microcalcifications and lesions detected as masses on mammography.110,117

Nine of the 21 studies reported that all of the lesions were non-palpable76,104,108,110–112,114–116 but the other studies reported no information on palpability of enrolled lesions. The studies reported insufficient information about characteristics of the lesions or the patients to explore the impact of these factors on the accuracy of the biopsies.

Biopsy Procedure Factors

All 21 studies used the Mammotome device either exclusively or in part. Seventeen of the studies used an 11G needle,80,104–114,117–120,122 two used a 14G needle,76,116 one used either a 14G or an 11G needle,121 and one did not report the size of the needle. 115 All but one of the studies used a combination of open surgery and patient followup to verify the results of the biopsies, and it used open surgery on all patients.117 Only three studies followed all patients for at least two years.104,110,116 Meta-regression found that method of biopsy verification did not affect the accuracy of the biopsies.

The majority of the studies reported that patients were prone,76,80,104,105,107–112,114,116–122 two reported that patients were seated,113,115 and one did not report information about patient positioning.106

Lomoschitz et al. reported that 12 cores were necessary for accurate diagnosis and taking more than 12 cores did not improve accuracy.110

Clinician and Facility Factors

Only two of the 21 studies were multi-center studies. Three of the studies were conducted in free-standing dedicated cancer centers,105,106,116 one was conducted in an ambulatory surgical center,80 and the rest were conducted in general hospitals. Six of the studies were conducted in the USA76,80,113,114,116,118 and 12 were conducted in Europe.104,106–112,115,117,120,121 Meta-regression did not find that the type or location of facility affected the accuracy of the biopsies.

Very few of the studies reported any information about the training or experience of the persons performing the biopsies. Pfarl et al. noted that for six of the seven false-negatives that occurred in the study, the biopsy procedure had been performed by an operator who had previously performed fewer than 15 stereotactic-guided biopsies.117

MRI-Guided Core-Needle Biopsies

Only one study reported data on the accuracy of MRI-guided biopsies performed with automated biopsy guns.124

Perforated Compression Grid Guided Core-Needle Biopsies

Only one study reported data on the accuracy of biopsies performed with automated biopsy guns guided by a perforated compression grid.125

Multiple Core-Needle Methods

There were an additional 24 studies that used multiple core-needle biopsy methods in their studies and did not report the data for different biopsy methods separately.126–149 Some of these studies reported information relevant to this topic as discussed below.

Patient and Breast Lesion Factors

Abdasaleh et al. reported that technical failures were more likely to occur with women with very dense breast tissue.130

The authors of Ciatto et al., who used multiple methods of performing core-needle biopsy, reported the percentage of procedures that gave false-negative results by lesion type: 2.7% palpable lesions, 2.2% nonpalpable lesions, 2.3% masses on mammography, 1.4% distortions on mammography, and 2.5% of microcalcifications.126 Cipolla et al. reported that correspondence between core-needle biopsy and surgical biopsy results was 100% for palpable lesions but only 88% for nonpalpable lesions.127 Fajardo reported that the sensitivity of core -needle biopsies for nonpalpable lesions and lesions with microcalcifications was 90.7%, much lower than the 97.4% sensitivity of core-needle biopsy for masses detected on mammography.129

Biopsy Procedure Factors

Abdasaleh et al. reported that taking two cores instead of one increased the accuracy of the procedure.130

Helbich et al. randomly assigned patients to be biopsied in different positions-seated upright, supine, or prone. The accuracy data were not reported separately for each group, but the authors did comment that patient position did not affect the biopsy procedure.144

Clinician and Facility Factors

Ciatto et al. reported that sensitivity of core-needle biopsies improved as the operators (radiologists) gained experience, from 88% in the first year of the study to 96% in the last year (eight years overall) of the study.126.

Question 2. In women with a palpable or non-palpable breast abnormality, what are the harms associated with core-needle breast biopsy compared to the open biopsy technique in the diagnosis of breast cancer?

The evidence for Key Question 1, 107 studies of overall low quality, was used to address Key Question 2. Fifty of the 107 included studies did not report any harms (see Appendix F); whether this was because no harms occurred is unclear. Five studies only reported that no severe complications or harms occurred. Tonegutti and Giradi reported that (unspecified) complications only occurred during the first year of performing stereotactically-guided vacuum-assisted biopsies.104

Very few of the included studies reported information about complications occurring in association with open surgical biopsy procedures. We consulted a narrative review published in 2007 to obtain further information about complications of open surgical biopsy procedures. In this review, Vitug and Newman report that 2 to 10% of breast surgeries are complicated by hematoma formation, and that 3.8% are complicated by infections.152 Rissanen et al. reviewed a series of 425 wire-localized open biopsy procedures and reported that 10.2% were complicated by vasovagal reactions.153

Use of Pain Medications

Four studies reported information on the use of pain medications.80,101,102,124 These studies reported that 100% of patients were sent home with narcotics after an open biopsy procedure, and only one patient (0.17%) required narcotics after a core-needle procedure.102 Twenty (3.5%) patients were reported to have required acetaminophen after a core-needle procedure.101 Note that being sent home with a medication may not necessarily mean the patients required or used the medication.

Bruising, Bleeding, and Hematomas

Twenty-four studies of 17,585 core-needle biopsy procedures reported that only 0.085% were complicated by hematomas that required treatment.56,57,59,66\,81,90,94,96,100–102,104,106,109,112–115,120,138,139,143,146,149 These studies reported that 3.85% of vacuum-assisted procedures were complicated by hematoma formation, and only 0.14% of vacuum-assisted procedures were complicated by hematomas that required treatment. In comparison, only 0.24% of non-vacuum-assisted procedures were reported to be complicated by hematoma formation, and only 0.035% of non-vacuum-assisted procedures were complicated by hematomas that required treatment. Due to inconsistency in reporting, these percentages should be used with caution; however, vacuum-assisted procedures do appear to have a higher rate of hematoma formation than other core-needle biopsy methods, although overall, hematomas rarely complicate core-needle procedures.

Twenty-four studies of 8,474 core-needle biopsy procedures reported that 1.4% were complicated by bleeding, but only 0.3% were complicated by bleeding that required treatment. 44,46,53,56,69,80,85,97,98,100,102–104,109,112–115,121,130–132,135,138 Of the vacuum-assisted procedures, 0.94% were reported to be complicated by bleeding, but only 0.34% of vacuum-assisted procedures were complicated by bleeding that required treatment or termination of the procedure. In comparison, 0.55% of non-vacuum-assisted procedures were reported to be complicated by bleeding, and only 0.20% of non-vacuum-assisted procedures were reported to be complicated by bleeding that required treatment. Due to inconsistency in reporting these percentages should be viewed with caution; however, vacuum-assisted procedures do appear to be complicated by bleeding more often than non-vacuum-assisted procedures, although bleeding is a rare complication of core-needle procedures.

Nine studies reported that bruising occurred after core-needle biopsy procedures. 46,57,59,85,90,99,101,108,141 Three of the nine reported that bruising was a common event, 46,85,141 two reported that approximately 50% of patients had bruising, 90,101 and four studies reported that 45 out of 976 patients (4.6%) had severe bruising.57,59,99,108 These nine studies used a variety of core-needle procedures.

Infections

March et al. reported that 2.1% of open biopsy procedures were complicated by the development of an abscess, but zero abscesses complicated 234 ultrasound-guided vacuum-assisted core-needle procedures.101 Tonegutti and Girardi reported that one abscess that required surgical treatment occurred in a series of 268 stereotactically-guided vacuum-assisted procedures.104 None of the other studies reported the occurrence of abscesses.

Twenty studies of 16,407 core-needle procedures reported that only 0.15% of the procedures were complicated by infections.44,53,57,59,66,77,81,85,93,94,97,98,102,106,108,109,133,135,139,149 Zannis and Aliano reported that 6.3% of open surgical biopsies were complicated by infections.80

Pain

Three vacuum-assisted biopsy procedures (out of over 6000 performed) were reported to have been terminated after patients complained of severe pain.76,108,114 No other types of biopsy procedures were reported to have been terminated due to patient complaints of pain. Seventeen studies of a wide variety of biopsy methods reported information about patient pain during the procedure, and overall only 1.7% of patients were reported to have experienced severe pain.44,46,76,84–86,93,94,96,100,101,108,112,114,121,146

Frazee et al. reported the mean pain score (10-point VAS scale) was 2.5 for open biopsy procedures and 2.8 for stereotactically-guided automated gun core-needle biopsies (the difference was not statistically significant).86

Wong and Hisham reported no difference in the amount of pain experienced by patients undergoing a 14G core-needle procedure vs. a 16G core-needle procedure.44 McMahon et al. reported that patients undergoing 18G core-needle procedures had significantly less pain than patients undergoing 14G core-needle procedures, but there was no significant difference in pain between 14G and 16G procedures.46

Vasovagal Reactions

Twenty-two studies of 7,526 core-needle procedures reported that 1% were complicated by vasovagal reactions (fainting or near -fainting).58,66,69,72,77–79,85,94,97,98,104,109,113–115,125,131,138,139,143,144 More than 40% of the vasovagal reactions occurred in patients who were reported to have been positioned sitting upright for the biopsy procedure (many of the studies did not report patient position so the other 60% of vasovagal reactions could have occurred in patients positioned in a variety of positions, or could have occurred primarily in seated patients).

Kirshenbaum et al. commented that the majority of vasovagal reactions occurred when inexperienced operators performed the biopsy procedures.131

Time to Recovery

One study, Frazee et al., reported information about time to recovery, measured by asking patients how long it had taken for them to return to their normal activities after the biopsy procedure. This study reported that the average time of recovery was 3.5 days for open biopsy procedures and 1.5 days for stereotactically-guided automated gun core-needle biopsy procedures.86

Impact of Biopsy Procedure on Usual Activities

One study, March et al., reported that ultrasound-guided vacuum-assisted procedures did not impact the usual activities of 47% of the women at all.101

Impact of Biopsy Procedure on Subsequent Mammographic Procedures

Three studies reported information about the impact of core-needle biopsies on subsequent mammographic examinations.109,120,121 All three studies performed stereotactic-guided vacuum-assisted core-needle procedures. These three studies enrolled 3,748 patients of whom 3,345 (89.2%) were reported to have no mammographically visible scarring after the biopsy procedures. Only seven of the patients (0.19%) were reported to have scars that were potentially diagnostically confusing on subsequent mammographic procedures.

Miscellaneous Reported Harms

Four studies of 2,600 patients reported that four cases of pneumothorax, none of which required treatment, had occurred.104,109,113,120 None of these four studies used the same method of performing the core-needle biopsies.

Two studies reported that one patient per study (out of 3,487 patients) had suffered a seizure during a stereotactic-guided vacuum-assisted procedure.109,120

One study of 268 patients undergoing stereotactic-guided vacuum-assisted biopsies reported that three patients developed acute inflammation at the biopsy site after the procedure.104

One study of 185 stereotactic-guided vacuum-assisted procedures reported that one patient vomited during the procedure.113

Dissemination of Cancerous Cells During the Biopsy Procedure

To address this possible harm of a breast biopsy we did not use formal inclusion criteria; any clinical study that addressed the topic was included for discussion. Full details of the studies are shown in Appendix E. The results of the studies are summarized in Tables 8 through 10.

Table 8. Dissemination of cancerous cells during biopsy procedures.

Table 8

Dissemination of cancerous cells during biopsy procedures.

Table 9. Summary of harms complicating core-needle biopsies.

Table 9

Summary of harms complicating core-needle biopsies.

Table 10. Summary of the impact of factors on harms.

Table 10

Summary of the impact of factors on harms.

We identified ten studies that used histopathology to demonstrate dissemination of cancerous cells by core-needle biopsy procedures.154–163 The percentage of needle tracks reported to contain displaced cancerous cells ranged from 0% to 65%. Diaz et al. demonstrated that the time elapsed between core-needle biopsy and examination of the needle track strongly influenced the findings, with fewer and fewer displaced cancerous cells observed the longer the interval, suggesting that the majority of displaced cancerous cells die off over time.154 However, we also identified six case reports of patients developing tumor recurrences at the site of prior core-needle biopsies, indicating that not all displaced cancerous cells are non-viable.155,161,164 Three of these six women were reported not to have received radiation therapy for the primary tumor; for the other three women it was not reported whether or not they had received radiation therapy.

The risk of tumor recurrence following biopsy was explored by four retrospective studies of 1,879 women. 165–168 Three of these four studies reported that women who did not have a pre-operative needle biopsy had a higher rate of tumor recurrence than women who did receive a pre-operative needle biopsy;165–167 the fourth study reported the opposite. 168 The majority of the women in these four studies were treated with breast-conserving surgery and radiation therapy.

The risk of seeding the lymph nodes with cancerous cells by biopsy procedures was examined in three retrospective studies of 3,103 patients. 169–171 Two of the three studies reported that the method of biopsy did not affect the rate of positive sentinel lymph nodes;169,171 the third study reported that the rate of metastases to the sentinel lymph node was higher in women who underwent some form of pre-operative biopsy.170

In 2006 Bleiweiss et al. reported 25 cases of false-positive sentinel lymph nodes.172 All 25 cases appeared to be caused by displacement of benign epithelial cells during a prior biopsy procedure. Twelve of the false-positive cases had undergone core-needle biopsy prior to the sentinel lymph procedure, 12 had undergone wire-localization open biopsy procedures, and one had undergone a fine-needle aspiration procedure. Although these cases are not, strictly speaking, cases of seeding lymph nodes with cancerous cells, this study is of clinical importance. False-positive sentinel lymph node procedures are likely to lead to over-treatment of patients, thus causing harm. These false-positive cases had stained positively for the presence of cytokeratins due to the presence of benign breast epithelial cells in the lymph nodes. Fifteen of the false-positives occurred in women with pure DCIS, and the remainder had DCIS plus invasive carcinoma. Twenty-two of the 25 cases had intraductal papilloma, (a not uncommon breast lesion) at the biopsy site and showed signs of displacement of benign cells at the biopsy site. The authors of this series of case reports suggest using caution when interpreting sentinel lymph node histopathology in cases where intraductal papilloma was noted during the initial biopsy procedure.

Question 3. How do open biopsy and various core-needle techniques differ in terms of patient preference, availability, costs, availability of qualified pathologist interpretations, and other factors that may influence choice of a particular technique?

We did not use formal inclusion criteria to select literature that addressed Key Question 3 due to the nature of the question. Data addressing this question were collected and are shown in Appendix E. The data are summarized in Table 13, Table 14, and Table 15, and are discussed outcome-by-outcome below. Economic factors that may influence the choice of a particular technique are discussed first, followed by factors highly important to patients, followed by other factors such as availability of equipment. Because of the nature of the quest ion and the sources of information used to address it, we did not draw many formal evidence-based conclusions for this question, nor, in most cases, did we attempt to rate the quality of the studies or grade the strength of the evidence.

Table 13. Summary of economic aspects of core-needle biopsy.

Table 13

Summary of economic aspects of core-needle biopsy.

Table 14. Summary of patient perspectives on choice of biopsy method.

Table 14

Summary of patient perspectives on choice of biopsy method.

Table 15. Summary of clinician and facility factors related to core-needle biopsy.

Table 15

Summary of clinician and facility factors related to core-needle biopsy.

Relative Costs

Articles identified by our searches that analyzed the costs of open and various core-needle biopsy techniques in the U.S. health care system within the last five years (published in or after 2004) are summarized in Table 11. The relative costs of open surgical biopsy and various core-needle biopsy techniques have been evaluated by six studies. Some of the studies developed models, while others prospectively followed a patient population. When evaluating the costs of these techniques and procedures, the studies have reviewed factors such as the initial purchase price of the devices used, the costs of staffing, the costs of processing and analyzing the biopsy samples, the patient volume where the device will be utilized, if the device is used as a complementary procedure, and what mammography results determine the use of a core-needle biopsy technique.

Table 11. Economic considerations.

Table 11

Economic considerations.

According to the literature reviewed, the costs of open surgical biopsy are substantially higher than core-needle techniques. A study by Hatmaker et al. in 2007 found that the average total cost of an open surgical biopsy performed in the operating room was $4,368 (presumably 2003–2005 U.S. dollars) with a median cost of $3,479 and the average total cost of image-guided core-needle biopsy was $1,267 with a median cost of $1,239.173

The results of a mammogram help surgeons and radiologists decide which core-needle technique, if any, would be beneficial and ultimately cost-effective for the patient and facility. Soo et al. used a decision analysis model to compare the costs of a 14-gauge core-needle biopsy to a 14-gauge and 11-gauge vacuum-assisted biopsy for noncalcified lesions. They found that the 14-gauge CNB is less costly for noncalcified lesions, which is not surprising since vacuum-assisted equipment is more expensive.174 Golub et al. prepared a cost-minimization model and found that image-guided core-needle biopsy was favored (cost the least) over open biopsy for low suspicion lesions, calcifications, and masses, primarily due to savings from reducing the overall number of surgeries performed.175

The cost to purchase a core-needle biopsy system is another factor of interest to facilities. In an article published in 2003, Kirshenbaum et al. reported that t he average list price for a breast imaging center to make an existing mammography unit biopsy ready (i.e add-on unit) was $90,000 and the average list price for a dedicated prone biopsy table was $226,000.131 Current quoted prices (not list prices) are about $170,000 (2008 U.S. dollars) for a dedicated table (which also requires a large dedicated room) and about $100,000 for an add-on unit.176 Unlike a dedicated prone biopsy table, a mammography unit with an add-on device can be used for general mammography purposes when not being used for a biopsy procedure. However, add -on units have limitations, including limited access angles, limited ability to restrict patient movement, and less patient comfort than dedicated units.176

Ultrasound-guided core-needle biopsies do not require special equipment and can be performed with a standard multi-purpose US device. Vacuum-assisted core-needle devices currently cost around $37,000 (2008 U.S. dollars) to purchase a console, and require $270 single-use probes.176 MRI-guidance is the most expensive method of performing core-needle biopsies, requiring expensive specialized equipment as well as access to an MRI facility.176

Spared Surgical Procedures

We identified 31 studies that reported information on how the use of core-needle biopsy spares women additional surgical procedures (see Table 12; also see Appendix E for further details). Women who undergo open biopsy with positive findings often undergo additional surgical procedures to ensure the entire lesion has been removed and to sample the lymph nodes. Women who undergo a core-needle biopsy procedure with positive findings may be able to undergo a single surgical procedure that simultaneously confirms the diagnosis and removes the entire lesion, and samples the lymph nodes if necessary, thus being spared additional surgical procedures. Women who undergo a core-needle biopsy with negative findings may be able to avoid surgical procedures altogether. Liberman et al. reported that, before the introduction of core-needle biopsy, 29% of women diagnosed with cancer had only one surgical procedure, but after the introduction of core-needle biopsy that number rose to 84%.177 The studies consistently reported that approximately 75% of women who underwent a core-needle biopsy procedure were spared further procedures, with a mean of approximately 1.2 procedures per woman compared to 1.5 to 2.0 procedures per woman who was initially evaluated with open surgical biopsy.

Table 12. Key Question 3: other outcomes.

Table 12

Key Question 3: other outcomes.

Seven of the studies reported information about the percentage of women who, after being diagnosed with breast cancer by either core-needle or open biopsy, were able to be treated for their cancer with a single surgical procedure. We combined the data reported by these studies in a meta-analysis. The data were consistent (I2= 2.2%). The summary odds ratio is 13.7 (95% CI: 5.6 to 34.6), an extremely large magnitude of effect. We felt that the strength of association between the type of biopsy and being able to treat the breast cancer with only one surgical procedure was strong enough to support an evidence-based conclusion. Although the internal validity (study quality) was low, the evidence was robust, consistent, and had an extremely large magnitude of effect. We therefore graded the strength of evidence supporting the conclusion as Moderate.

Procedure Preference

We identified 20 studies that reported data on patient preferences (see Table 12; also see Appendix E for further details). Ten of the 20 studied vacuum-assisted methods. The majority of the studies did not directly compare different biopsy procedures and instead reported information such as that the patients tolerated the procedure well or would recommend it to others in the future. One study reported that patients preferred the decubitus position to the prone position.138 Two studies reported that vacuum-assisted procedures were more comfortable than other types of core-needle biopsies.178,179 Two authors reported that patients lost less time to core-needle procedures than to open procedures.100,180 The majority of the studies concluded that core-needle biopsies were preferable to open biopsies, but one study reported that a survey of patients found that 90% were satisfied with their open surgical biopsy compared to only 80% satisfied with a vacuum-assisted core-needle biopsy.181

Cosmetic Results

We identified ten studies that reported information on cosmetic results (see Table 12; also see Appendix E for further details). The studies all used vacuum-assisted core-needle biopsy methods. The authors of the studies reported information on how patients felt about the cosmetic results post-procedure. Overall, patients were reported to have been satisfied with the cosmetic results. Only one of the ten studies, Chun et al., compared a group of patients undergoing core-needle biopsy to a group of patients undergoing open biopsy.181 Chun et al. compared cosmetic results of patients undergoing wire-localized open biopsy to patients undergoing vacuum-assisted 11-gauge core-needle biopsy two years post-procedure. Ninety-five percent of the core-needle biopsy group and only 25% of the open biopsy group were very satisfied with the appearance of their breast. None of the core-needle biopsy group said the cosmetic results were unacceptable compared to 20% of the open biopsy group who found the results unacceptable.

Although all of the studies reporting on cosmetic results used vacuum-assisted methods, it is likely the results apply to most forms of core-needle biopsy. Regardless of the needle gauge or method used, the actual incision cut in the skin for core-needle procedures is always approximately ¼” long.176

Physician Experience

We identified ten studies that reported information concerning physician experience (see Table 12; for further details see Appendix E). Authors of some of the studies commented that certain devices were easier for inexperienced physicians to use. In general, however, the authors of the studies concluded that greater experience with particular devices improved the accuracy of the biopsy procedures, shortened procedure duration times, and led to a decrease in the number of open biopsies that were performed.

Availability of a Qualified Pathologist

We identified two studies that discussed pathologist qualifications and availability (see Table 12; for further details see Appendix E). One reported that whether a specimen was read by a local or central pathologist made little difference because concordance between readings was 96.1% (κ= 0.90) for core needle biopsy and 92.6% (κ = 0.93) for open surgical biopsy. However, there was greater disagreement with respect to ADH and ALH for both biopsy types, with underestimation of the lesion by local pathologists in comparison to the central pathology laboratory (for CNB, ADH agreement 63% and ALH agreement 53%; for open, ADH agreement 45% and ALH agreement 73%).182 The authors of the other study speculated that lack of an experienced pathologist was the cause of the low accuracy of the core-needle biopsies performed during the course of their study.183

Availability of Equipment

We identified three studies that talked about the impact of equipment availability (see Table 12; for further details see Appendix E). One reported that vacuum-assisted devices were more commonly available in the U.S. than in Europe.184 One reported that wait times for access to core-needle procedures were significantly shorter than wait times for access to open surgical procedures.185 The authors of the third study reported that wait times for access to a dedicated prone biopsy table were longer than wait times for other types of core-needle biopsy.186

Resource Usage

We identified two studies that talked about resource usage (see Table 12; for further details see Appendix E). The authors of one study reported that vacuum-assisted procedures required more physician and room time than free-hand ultrasound-guided procedures.187 The other study reported that dedicated prone tables use four times as much space as non-prone units.188

Procedure Duration Time

We identified 40 studies that reported information about the duration of different biopsy procedures (see Table 12 ;for further details see Appendix E). The studies reported a wide range of times, from 10 minutes to 128 minutes. The wide range of times may be in part due to different definitions of when exactly the procedure was defined as starting and ending: for example, does the procedure start when the patient enters the room? When the incision is made? Does it end when the sample is collected or when the patient is released to go home? In general, study authors did not define what exactly they meant by procedure duration time.

The reported mean or median time to perform core-needle biopsies under ultrasound guidance ranged from 10 to 60 minutes; the mean or median time to perform core-needle biopsies under stereotactic guidance ranged from 19 to 70 minutes; and the mean or median time to perform core-needle biopsies under MRI guidance ranged from 31 to 70 minutes. Vacuum-assisted core-needle biopsies were reported to have a mean or median duration of 10 to 70 minutes. Open surgical biopsies were generally reported to have longer duration times than core-needle procedures, but only two studies reported estimated duration times of open biopsy—40 to 45 minutes.51,189

Wait Time for Test Results

We identified two studies that reported mean or median times to get a diagnosis following a breast biopsy (see Table 12; for further details see Appendix E). The authors reported that wait times after a core-needle procedure were 7 to 10 days shorter than after an open excisional biopsy.183,185

Previously Published Systematic Reviews

Our searches identified two previously published systematic reviews. Verkooijen et al. reviewed the literature published prior to 1999 on core-needle biopsy of non-palpable lesions.234 Fahrbachet al. reviewed the literature published from 1996 to 2004 on core -needle biopsy of patients referred for biopsy after screening mammography.235

We assessed the quality of each systematic review using the ‘assessment of multiple systematic reviews’ (AMSTAR) measurement tool.236 The AMSTAR consists of 11 items, which have been tested for face and content validity. The items assess whether or not a systematic review includes important elements, such as a comprehensive literature search, assessment of study quality, appropriate methods to combine study findings, and assessment of publication bias. Responses to each item are checked as ‘Yes’ if the review includes that item, ‘No’ if it does not, ‘Can’t tell’ if the item cannot be answered by the information provided in the review, or ‘Not applicable’ if the item is not applicable. The AMSTAR does not provide a method for rating the quality of a review. To rate the quality of the reviews, we applied the following criteria: a rating of ‘High’ if the review received mostly ‘yes’ responses (at least 8), a rating of ‘Low’ if the review received mostly ‘no’ responses (at least 8), and a rating of ‘Moderate’ if the review received mixed responses. Both systematic reviews were rated as Moderate quality. The reviews were not rated as High quality because neither systematic review stated conflicts of interest or incorporated ratings of the quality of the literature into their conclusions. See Appendix E for details about the quality rating.

Verkooijen et al. included only five cohort studies in their review. Their inclusion criteria were studies of non-palpable lesions, either surgical biopsy or at least two years of followup to verify the true diagnosis, and a minimum of five cores taken per lesion. All included studies happened to have used stereotactic guidance. The authors assumed core-needle biopsy had no false-positives (i.e., malignant diagnoses on core needle that were not found on open surgery were assumed to have been completely removed by the core-needle procedure). Their analyses found that the DCIS underestimation rate was 15% (95% CI: 8.0 to 26.0%), the ADH underestimation rate was 40% (95% CI: 26.0 to 56.0%), and the overall sensitivity of core-needle biopsy for non-palpable lesions was 97.0% (95% CI: 95.0% to 99.0%). Only two complications were reported, one hematoma and one case of infection.234

Fahrbach et al. included 12 studies of stereotactically-guided vacuum-assisted core-needle biopsy and compared them to 25 studies of stereotactically-guided automated gun core-needle biopsy. One of their inclusion criterion was that the study must have been conducted in a western-style health care system (North America, Europe, Australia, or New Zealand). Their analyses found the false-negative rate of vacuum-assisted biopsy was 1.2%, the DCIS underestimation rate was 13.7%, and the ADH underestimation rate was 29.2%. Automated gun core-needle biopsy had a false-negative rate of 2%, a DCIS underestimation rate of 27.1%, and an ADH underestimation rate of 47.4%. Further, the authors performed analyses of possible factors that may have affected the results. Study location was a significant predictor of the false-negative rate, but type of reference standard and patient position had no significant impact on the results.

The authors of both systematic reviews concluded that core-needle biopsy rarely misdiagnosed malignant lesions as benign. Fahrbach et al. concluded that vacuum-assisted biopsy may provide lower miss and underestimation rates than automated gun core-needle biopsy.235