Fax: (011) 39-045-8027136
Is there still a role for fine-needle aspiration cytology in breast cancer screening?†
Experience of the Verona Mammographic Breast Cancer Screening Program with real-time integrated radiopathologic activity (1999–2004)
Article first published online: 27 FEB 2008
Copyright © 2008 American Cancer Society
Volume 114, Issue 2, pages 74–82, 25 April 2008
How to Cite
Manfrin, E., Mariotto, R., Remo, A., Reghellin, D., Dalfior, D., Falsirollo, F. and Bonetti, F. (2008), Is there still a role for fine-needle aspiration cytology in breast cancer screening?. Cancer, 114: 74–82. doi: 10.1002/cncr.23412
See editorial on pages 65–6, this issue.
- Issue published online: 11 APR 2008
- Article first published online: 27 FEB 2008
- Manuscript Accepted: 10 JAN 2008
- Manuscript Revised: 8 JAN 2008
- Manuscript Received: 17 OCT 2007
- Foundation Savings and Loan Company of Verona Vicenza Belluno and Ancona. “Carcinoma of the mammella: fenotipici markers and molecular pointers of prognosis and therapeutic answer
- Ban of scientific search 2004, biomedical address
- breast cancer;
- fine-needle aspiration cytology;
- onsite cytology;
- quality assurance
In mammography screening, the majority of abnormalities are the result of nonmalignant proliferative breast lesions.1, 2 Preoperative diagnostic procedures are adopted in the investigation of suspicious or equivocal screen-detected breast lesions to avoid unnecessary open biopsy. The high-quality performance of a nonoperative procedure provides the best benign-to-malignant ratio in open biopsy. The quality assurance system is an integrated part of breast screening programs3 and is necessary to evaluate screening performance.
Fine-needle aspiration cytology (FNAC) has been used extensively over the years in the diagnosis of breast lesions. However, despite the many benefits (eg, fast turnaround time, patient compliance, and its reputation as a simple and cost-effective method), it has been increasingly criticized of having a high inadequate rate (IR) and suboptimal accuracy.4–6 The preoperative confirmation of breast cancer by FNAC has gradually been reduced in many screening programs in which the results were not satisfactory and replaced by microhistology, consisting of ultrasound-guided core needle biopsy (USCNB) or stereotactic core needle biopsy (SCNB).4, 7–9 The IR for FNAC is inversely proportional to the operator's level of experience and the presence of specialized cytologists.10–13 Radiologic imaging5, 6 and the type of guidance system used are additional factors that could affect the performance of FNAC.5 Diagnostic accuracy can be achieved through a multidisciplinary consultation, combining FNAC results with clinical and radiologic data (the triple test).14 The diagnostic value of FNAC improves with the immediate onsite evaluation of specimens.10, 15 Immediate cytologic diagnosis in real time is cost-effective and allows those patients with benign diseases to be given immediate reassurance whereas the management of patients with malignant or suspicious lesions can be quickly planned.16, 17
The current study describes the quality performance results of FNAC over a 5-year period of activity at the Breast Cancer Screening Program in Verona (BCSPV), in which the integrated radiopathologic, real-time assessment of suspicious lesions and immediate cytologic reports are performed. The objective of the current study was to highlight FNAC accuracy when it is part of an integrated radiopathologic service in which the cytologic sample results are immediately matched with clinical and imaging findings.
MATERIALS AND METHODS
A breast cancer screening program for women ages 50 to 69 years was initiated in Verona, Italy in 1999.
Integrated Radiopathologic Screening Activity Organization
Compared with “classic” screening, the BCSPV is organized differently, similar to that of a breast clinic, and operates in a building exclusively devoted to breast screening. The first-level mammograms and all the second-level examinations (ie, ultrasound [US], galactography, FNAC, CNB, and vacuum-assisted biopsy [VAB]), are performed in the same structure by a specialized team comprised of 2 radiologists, 2 consultant pathologists, 4 technicians, and 2 nurses. The radiologists and pathologists are fully specialized in breast disease, the former with years of experience in breast cytology and histopathology. All women undergo mammography using 2 projections; the radiologists then interpret the mammograms and the patients receive the results within 10 minutes. Patients with negative mammograms are dismissed after reassurance whereas patients with suspicious lesions are immediately submitted to second-level examinations. In the majority of cases, FNAC is the first technique of preoperative sampling to be used. The FNAC samples are obtained by the radiologist either freehand or by US or computer-assisted stereotactic mammography (STX) guidance (using a 22-gauge needle). A pathologist evaluates the slides for adequacy in real time. The conclusive diagnostic report is given to the radiologist within 4 to 7 minutes. Cases with inadequate or unsatisfactory FNAC samples or a nonconcordant radiocytologic degree of suspicion are immediately discussed by the medical team and new cytologic samples are obtained or core needle biopsy performed to determine the nature of the lesion. Patients with benign results from second-level examinations are sent for regular screening follow-up, whereas a surgical examination is scheduled for those patients with lesions determined to be cytohistologically malignant or suspicious.
All the data from those patients participating in the BCSPV were collected in the screening computer system. A special file was adopted for all patients who underwent cytohistologic analysis. Cytologic reports were expressed in diagnostic categories according to the European guidelines for quality assurance in breast cancer screening and diagnosis.3 The diagnoses of were C1 (inadequate), C2 (benign), C3 (atypia probably benign), C4 (suspicious of malignancy), and C5 (malignant). FNAC diagnoses were compared with the final outcome.18 Only 1 lesion per patient was eligible. For patients with bilateral disease, the tumor at the worst stage was taken into consideration. Multiple cancerous foci in 1 breast were counted as 1 cancer. FNAC performance indicator values were calculated according to the standards of quality assurance as stated in the European quality assurance guidelines (Table 1).3 The patients with a benign diagnosis at FNAC (C2 category) were followed for the next 3 years. A case was considered to be false-negative (FN) if cancer developed within the same area previously submitted to FNAC within 3 years after the C2 diagnosis. FNAC performance was evaluated over the entire series and a separate calculation also was made in accordance with the radiologist's degree of risk, radiologic imaging (calcification cluster, mass lesion, and mass lesion with calcifications), and the guidance system adopted to perform the cytologic sampling (freehand, US, and STX). The results were correlated with the thresholds for FNAC performance suggested by the European guidelines for quality assurance.
|Definition of quality assurance standards in the European guidelines|
|Absolute sensitivity||No. of cancers diagnosed as C5 expressed as a % of the total no. of cancers sampled|
|Complete sensitivity||No. of cancers that were not definitely negative or inadequate on FNAC as a % of the total no. of cancers|
|Specificity (full)*||No. of correctly identified benign lesions (C2) expressed as a % of the total no. of benign lesions aspirated|
|Positive predictive value for C5||No. of correctly identified cancers (no. of C5 results minus the no. of false-positive results) expressed as a % of the total no. of positive results (C5)|
|Positive predictive value for C4||No. of cancers identified as suspicious (no. of C4 results minus the no. of false suspicious results) expressed as a % of the total no. of suspicious results (C4). This excludes those not confirmed by histology.|
|Positive predictive value for C3||No. of cancers identified as atypical (C3 results from cancers) expressed as a % of the total no. of C3|
|Negative predictive value||No. of benign lesions (C2) expressed as a % of the total of benign findings on histology|
|False-negative case||A case that subsequently is found (during the next 3 y) to be a carcinoma, having had a negative cytology result|
|False-negative rate†||No. of false-negative results expressed as a % of the total no. of carcinomas sampled|
|False-positive rate||No. of false-positive results (C2) expressed as a % of the total no. of carcinomas sampled|
|Inadequate rate||No. of inadequate specimens (C1) expressed as a % of the total no. of cases aspirated|
|Inadequate rate from cancers||No. of inadequate specimens (C1) expressed as a % of the total no. of cancers|
|Suspicious rate||No. of C3 + C4 cases expressed as a % of the total no. of cases aspirated|
FNAC Accuracy According to Radiologic Imaging
Calcification cluster, mass lesion, and mass lesion with calcifications were the mammographic lesion categories studied to verify the impact of the radiologic features of the lesions on cytologic accuracy. FNAC performance indicators were separately calculated for each category of lesions sampled and the differences were statistically evaluated.
FNAC Accuracy According to the Guidance System
Cytologic samples were performed either freehand, with US, or with STX using 22-gauge needles. To evaluate the effect of the sampling method on the accuracy of FNAC, the performance indicators were calculated separately for each of the 3 guidance systems and the results were then compared. The differences were statistically evaluated.
Positive Predictive Value of FNAC in Combination with Clinical and Imaging Findings
To evaluate the effect of the interdisciplinary consultation on FNAC accuracy, the positive predictive value (PPV) of FNAC was calculated separately combining the cytologic diagnostic categories of C3, C4, and C5 with the radiologist's degree of clinical and imaging suspicion. The clinical and imaging suspicion of a cancerous lesion was graded into the categories of low-risk, medium-risk, and high-risk/positive for cancer.
The differences between the performance indicator values in the lesion type group and in the sampling method group were evaluated with the z-test and Fisher exact test. The tests were considered statistically significant if the P value was <.05.
Between July 1, 1999 and June 30, 2004, 54,472 women were screened. The BCSPV results and performance indicators have been reported previously.19 Of the 1286 screen-detected abnormalities submitted to preoperative diagnostic procedures (FNAC, CNB, and VAB), cytologic examination was performed in 1263 cases, accounting for 98.2%. Open biopsy was performed in 510 of the total number of women screened. On histology, 427 cases (83.7%) were found to be malignant and 83 (16.3%) were nonmalignant. The benign-versus-malignant biopsy ratio was 0.2.
Cytologic Performance of the Entire Series
Cytology was chosen in 98.2% of cases (1263 of 1286 cases) as the primary preoperative tissue assessment method. FNAC was performed freehand in 72 cases (5.7%), by US guidance in 786 cases (62.2%), and by STX guidance in 381 cases (30.1%). In 24 of 1263 cases (2%), an apposition slide was performed on nipple discharge. The complete cytologic findings in relation to the final outcome are summarized in Table 2.
|Histology report||No. of cases|
|FNAC report||Benign on follow-up||Benign on histology||Carcinoma in situ||Invasive cancer||Total|
|Total||764 (60.5%)||80 (6.3%)||94 (7.5%)||325 (25.7%)||1263|
In the BCSPV, FNAC quality performance indicators have met or exceeded the majority of the measures recommended by the European quality assurance guidelines (Table 3).3 The complete sensitivity was 95.4%, the specificity was 74.8%, and the PPV of a diagnosis of category C5 was 99.3%. The FN rate (FNR) was 2.4%; 10 cases were reported as C2 at FNAC, but were determined to be cancer on final histologic examination (Table 2). The total IR was 9.8% (125 of 1263 cases). The IR from cancer (IRc) was 2.4% (10 cases). The false-positive rate (FPR) was 0.5%.
|FNAC Performance Indicators||BCSPV %||European guidelines|
|% Minimum||% Preferred|
|Positive predictive value for C5||99.3||>98||>99|
|Positive predictive value for C4||71.4||Not defined||Not defined|
|Positive predictive value for C3||18.9||Not defined||Not defined|
|Negative predictive value||98.6||Not defined||Not defined|
|Inadequate rate from cancers||2.4||<10||<5|
FNAC Accuracy According to Radiologic Imaging
Of the 1263 FNAC samples, 1026 were stratified according to the lesion type at mammography. The performance of FNAC was calculated separately for calcification cluster, mass lesion, and mass lesion with calcifications categories. The results are summarized in Table 4. A total of 310 cases (30.2%) were calcification clusters, 520 cases (50.7%) were mass lesions, and 196 cases (19.1%) were mass lesions with calcifications.
|FNAC performance indicators||BCSPV||European guidelines|
|Calcification clusters n=310, %||Mass lesion n=520, %||Mass lesion + calcifications n=196, %||Minimum %||Preferred %|
|Positive predictive value for C5||96.9||100||100||>98||>99|
|Positive predictive value for C4||70.8||76.4||69||Not defined||Not defined|
|Positive predictive value for C3||30.0||16.7||30||Not defined||Not defined|
|Negative predictive value||100||98||96.5||Not defined||Not defined|
|Inadequate rate from cancers||4.5||1.9||0.9||<10||<5|
The statistical analysis indicated a few significant differences in the performance indicators between the 3 categories (Table 5). FNAC performance in mass lesions versus calcification clusters was statistically better with regard to full specificity (P < .001), the PPV for category C5 (P = .029), the FPR (P = .030), and the IR (P < 0.001). FNAC performance in calcification clusters versus mass lesions with calcifications was better with regard to the NPV (P = .026), but was worse with regard to the IR (P < .001). FNAC specificity (full) was significantly better in mass lesions versus mass lesions with calcifications (P = .010).
|FNAC performance indicators||Comparison among different FNAC performances in relation to the type of lesion detected at mammogram|
|Mass vs calcification cluster % (P < .05)||Calcification cluster vs mass+calcifications % (P < .05)||Mass vs mass+calcifications % (P < .05)|
|Specificity (full)*||80.5 vs 63.8 (.001)||NS||80.5 vs 67.1 (.010)|
|Positive predictive value C5||100 vs 96.9 (.029)||NS||NS|
|Positive predictive value C4||NS||NS||NS|
|Positive predictive value C3||NS||NS||NS|
|Negative predictive value||NS||100 vs 96.5 (.026)||NS|
|False negative rate†||NS||NS||NS|
|False positive rate||0 vs 2.2 (.030)||NS||NS|
|Inadequate rate||5.0 vs 19.4 (.001)||19.4 vs 6.1 (.001)||NS|
|Inadequate rate from cancers||NS||NS||NS|
FNAC Accuracy According to the Guidance System
FNAC performance was calculated separately for the 3 different guidance systems and the results are summarized in Table 6. FNAC was performed freehand in 72 cases (5.7%), by US guidance in 786 cases (62.2%), and by STX guidance in 381 cases (30.1%). Twenty-four cases of nipple discharge evaluation were excluded from the statistical analysis. The FNAC performance using the STX method was lower than the US and freehand methods for the majority of indicators, but the majority of the differences were not found to be statistically significant (Table 7). The absolute IR and IRc were significantly higher in the STX sampling method when compared with the freehand method (P = .001 and P = .005, respectively). Comparing FNAC performance between the group of lesions sampled with US guidance and those sampled with STX guidance, the specificity (full), PPV for category C5, and IR were found to be significantly better in the US guidance group (P < .001, P = .014, and P < .001, respectively).
|FNAC performance indicators||BCSPV||European guidelines|
|Freehand guidance n=72, %||US guidance n=786, %||STX Guidance n=381, %||% Minimum||% Preferred|
|Positive predictive value for C5||100||100||97.2||>98||>99|
|Positive predictive value for C4||62.5||75||66.7||Not defined||Not defined|
|Positive predictive value for C3||33.3||14.6||26.1||Not defined||Not defined|
|Negative predictive value||96.8||98||99.4||Not defined||Not defined|
|Inadequate rate from cancers||6.1||1.8||2.9||<10||<5|
|FNAC performance indicators||Comparison among different FNAC performances in relation to the type of sampling guidance|
|Freehand guidance vs US guidance % (P < .05)||US guidance vs STX guidance % (P < .05)||STX guidance vs freehand guidance % (P < .05)|
|Specificity (full)*||NS||79.3 vs 65.1 (.001)||NS|
|Positive predictive value for C5||NS||100 vs 97.2 (.014)||NS|
|Positive predictive value for C4||NS||NS||NS|
|Positive predictive value for C3||NS||NS||NS|
|Negative predictive value||NS||NS||NS|
|Inadequate rate||NS||6.4 vs 17.8 (.001)||17.8 vs 6.9 (.001)|
|Inadequate rate from cancers||NS||NS||2.9 vs 6.1 (.005)|
PPV of FNAC in Relation to Clinical and Imaging Findings
Table 8 shows the PPV of FNAC in the C3, C4, and C5 diagnostic categories when combined with the radiologist's degree of suspicion resulting from summarizing clinical and imaging findings. The PPV of FNAC was increased by an integrated analysis of risk in all the cytologic categories.
|Degree of risk at clinical and imaging findings||FNAC No. of cases||PPVC3 No. (%)||PPVC4 No. (%)||PPVC5 No. (%)|
|Low-grade suspicious lesion||671||44 (14)||26 (38)||19 (95)|
|Medium-grade suspicious lesion||220||19 (21)||28 (50)||34 (97)|
|High-grade suspicious lesion/positive for cancer||372||12 (33)||70 (80)||254 (100)|
False-negative, False-positive, and Inadequate Cases from Cancers
Ten FNAC cases (2.4%) were FN (Table 2). Five of these 10 cases were carcinoma in situ, 3 were lobular carcinoma in situ (LCIS), and 2 were ductal carcinoma in situ (DCIS). Five of the 10 cases were invasive carcinoma, 2 of which were tubular type and 3 of which were invasive ductal carcinoma (IDC). All the cancers were well differentiated (grade 1 according to the Combined Nottingham Grading System), except for 1 IDC that had a moderate grade of differentiation (grade 2). All invasive cancers were classified as pT1 on pathologic staging (according to the American Joint Committee on Cancer Pathological Grading System) (Table 9).
|Case||Histologic diagnosis||pT classification||Grade||pN classification|
Of the total 419 cases that underwent FNAC, 10 cases (2.4%) were inadequate. The missed cancers were 3 cases of DCIS, 1 case of ductal microinvasive carcinoma, 1 case of IDC, and 5 cases of invasive lobular carcinoma (ILC). Four of the ductal carcinomas were graded as grade 3 and 1 was graded as grade 2. The ILC cases were graded as classic (grade 1) in 4 cases and pleomorphic (grade 3) in 1 case. On pathologic staging, 2 of the 5 IDC cases were classified as pT1b, and 3 were classified as pT2 (Table 10).
|Case||Histologic diagnosis||pT Classification||Grade||pN Classification|
Two cases (0.5%) were FP (Table 2). On histology, 1 case was determined to be florid adenosis and the other was classified as ductal atypical hyperplasia.
In the BCSPV, FNAC performance indicator values were found to have met or exceeded the majority of the measures suggested by the European quality assurance guidelines (Table 3).3 The IR of FNAC is low (9.8%), particularly the IRc (2.4%) (Table 3) and the assessment efficiency and accuracy is explained by the optimal benign-to-malignant ratio noted in open biopsy (0.2).19 These data are among the favorable mean values of local and national Italian breast cancer screening programs.20 In the BCSPV, the PPV for cancer was 99.3% (Table 3), which is better than both the Italian local screening mean value (93.7%)21 and the preferred value stated in the European guidelines for quality assurance (>99%).3 In our integrated radiopathologic screening system, the FNR was low (2.4%) (Table 3) and the PPV of the cytologic C5, C4, and C3 FNAC diagnostic categories was positively influenced by the radiologist's degree of suspicion (Table 8). In the BCSPV, the positive effect of the integration of radiology and pathology, as demonstrated by the observed improvement in PPV, highlights the possibility for patients to undergo planned management of their disease in real time.14 It has been proved that including clinical examination, additional mammographic views, US, FNAC, and CNB in the standard diagnostic assessment reduces the possibility of delayed diagnoses in FN assessments.22 These results can be easily achieved in centers using a multidisciplinary approach to screening. In the BCSPV, specialized pathologists are present onsite to audit technical FNAC procedures and to perform diagnostic consults with radiologists. In our screening organization, all patients are immediately sent to a second-level examination when a suspicious lesion is detected. Those cases with nonconcordant radiocytologic diagnoses are sent immediately for further assessments and new FNAC samples taken. This favors low FP and FN results and a low IR (Table 3).
The results of the current study confirm that the PPV of FNAC increases when combined with the radiologic grade of suspicion (Table 8)14, 23 and better accuracy is achieved with a specialized cytopathologic and multidisciplinary approach.10–13, 24 The experience in breast cytology and in sampling techniques11, 12, 25 associated with the onsite evaluation of FNAC specimens have been reported to provide an accurate assessment and more efficient clinical care of patients.10, 13, 15 In this context, the cost-effectiveness of FNAC10, 13, 15 makes it one of the better methods to apply during the course of a breast cancer screening program.
Data from the current study indicate that FNAC is a reliable and cost-effective method for the diagnosis of breast lesions in the correct setting. Multi-institutional trial studies performed in North America5–7 and the U.K.4 have reported unsatisfactory FNAC results in the majority of screening centers in which no special FNAC or specimen preparation training methods were adopted and in cases in which a cytopathologist was not available on site. The high IR of FNAC samples in this setting has made its use impractical and has encouraged the use of CNB in breast screening programs.4, 7–9
FNAC performance in the BCSPV is only marginally affected by the radiologic imaging of the lesions (Table 4). The absolute IR is higher in lesions with calcifications, particularly calcification clusters (IR of 19.4%), than in mass lesions (Table 4), with a statistically significant difference noted (P < .0001) (Table 5). However, the majority of the inadequate cases are found to be benign on histology or follow-up (Table 2). The FNAC IRc is higher in calcification clusters (Table 4), but comparison of the IRc among the 3 types of radiologic lesions in the current study demonstrated no statistically significant differences (Table 5). The current study data confirm that FNAC performance for the majority of indicators is lower in calcification clusters compared with mass lesions.6 In the BCSPV, the integrated radiopathologic activity most likely guarantees the global accuracy of FNAC that is not affected by the low performance in calcification clusters (Table 3).
In the current series, FNAC performance was found to be better when the lesion sampling was performed with US guidance and/or freehand rather than with STX guidance (Table 6). The IR in the STX guidance group of lesions was found to be significantly higher than with the other sampling methods (P < .0001); when compared with US, the STX sampling method was found to have a significantly lower performance in the diagnosis of true benign and malignant lesions (PPV of category C5) (Table 7). This low performance is possibly explained by patient selection, because it should be considered that STX guidance is adopted more frequently in cases of lesions with calcifications and, for this reason, the FNAC performance reflects the same limitations observed for this type of lesion.
FNAC must be included in an integrated radiopathologic breast cancer screening program to guarantee the best performance. Its cost-effectiveness makes it one of the better methods to apply in a breast cancer screening program.
The multidisciplinary approach is necessary to increase the quality of FNAC and to reduce its diagnostic limititations. When this model of activity is not available, the role of FNAC is less effective and adding CNB to FNAC should be considered.
- 3European Guidelines for Quality Assurance in Breast Cancer Screening and Diagnosis. 4th ed. Luxembourg: Office for Official Publications of the European Communities; 2006., , , , , .
- 5Rate of insufficient samples for fine-needle aspiration for nonpalpable breast lesions in a multicenter clinical trial: the Radiologic Diagnostic Oncology Group 5 Study. The RDOG5 investigators. Cancer. 1998; 82: 679–688., , , et al.
- 20Trend temporali di alcuni indicatori dei programmi di screening mammografico in Italia: 1996–2004. Quinto Rapporto. Osservatorio Nazionale Screening. Milano: Ed. Inferenze scarl, 2005: 32–46., , , et al.
- 21I programmi di screening oncologici del Veneto. Rapporto 2005. Istituto Oncologico Veneto, Registro dei tumori del Veneto. Ed. CLEUP, 2007: 45., , , .