nueva página del texto (beta)
Inglés (pdf)
Artículo en XML
Referencias del artículo
Traducción automática
Enviar artículo por email
Citado por SciELO 
Similares en
SciELO 
Permalink
Introduction
Breast cancer is the leading cause of cancer mortality in Mexican women since 2006, accounting for 14% of cancer-related deaths1, and yet, scarce information is available on tumor sampling and subcellular determinations2. Estrogen receptors (ER) and progesterone receptors (PR) along with c-erbB2 human epidermal growth factor receptor 2 (Her2) and Ki-67 expression are currently used to classify invasive epithelial breast tumors into four clinically relevant categories3. Immunohistochemistry (IHC) is the current method to accomplish such measurements, quantifying its nuclear or membrane signals2,3. Concern about IHC results compelled its automatization and nowadays to "next generation IHC"4. However, for economic reasons in developing countries most laboratories are still doing them manually. How much confidence should be given to measurements in these settings?
The trend for less radical surgeries as a treatment for breast cancer brings the opportunity to use core needle biopsies (CNB) and excisional biopsies (EB) in the same patient and, it offers the chance to bring about concordance studies regarding IHC analyses. Few studies addressing concordance of ER, PR, and Her2 have been published5-17. Such studies have shown a good proficiency of CNB for morphologic diagnosis and subcellular studies. Most of the published series are based on automatized protocols of IHC excluding patients submitted to neoadjuvant therapy and showed a concordance rate for ER, PR, and Her2 in core needle biopsies and excisional resections of 62-98%, 69-89%, and 60-98%, respectively5-17.
Our aim was to calculate concordances among core needle biopsies and EB regarding histological type, ER, PR, and Her2 in consecutive patients, including those receiving neoadjuvant treatments without a complete pathologic response.
Materials and methods
Study population
A search of medical records at the Instituto Nacional de Ciencias Medicas y Nutricion Salvador Zubiran was performed and patients with suspicious breast lesions submitted to CNB with subsequent EB, being lump excision, radical mastectomy, or cuadrantectomy due to in situ or invasive breast carcinoma were retrieved since January 2008 through December 2012. Medical charts of each patient were reviewed, and clinical information on gender, presentation, and age at diagnosis was recorded. The pathology reports from the diagnostic CNB and EB were reviewed. The data recorded from the CNB were core number, histological type of lesion, and IHC results. Data recorded from the excision were type of excision, tumor type according to the WHO classification, size, and ER, PR, and Her2 results. In both samples, the reported ER, PR, and Her2 score was compiled. The study followed the guidelines of the Declaration of Helsinki and was approved by the institutional review board. Patient identity and its pathological specimens remained anonymous in the context of this study. Cases were grouped as tumors as in situ carcinoma (either ductal or lobular), invasive carcinoma (either ductal or lobular), mixed, or other types of invasive carcinoma.
IHC analyses
In our institution, IHC is done manually and is interpreted according to the college of the American pathologists guidelines. Briefly, deparaffinized 1.5 μm tissue sections on charged slides were submitted to antigen retrieval for 20´ in an electric pressure cooker and mounted in cover plates in a plastic rack (Thermo Scientific, Cashire UK). Protein and endogen peroxidase was blocked for 5´ and the primary antibody incubated for 45´. After washing with Tris-buffered saline (TBS), two drops of mouse/rabbit immunodetector biotin link was applied for 10´ and rinsed with TBS. Finally, two drops of mouse/rabbit immunodetector HRP were used and slides developed with DAB under microscopic (Olympus CH2) observation. Extension of the stain was evaluated in neoplastic cells and classified as 0% or negative, 1-25%, 26-50%, 51-75%, and > 75% as positive and, nuclear staining intensity in a scale from 0 to 3+. Her2 score was obtained according to the American society of clinical oncology guidelines, considering 0/1+ as negative, 3+ as positive, and 2+ equivocal. ER detection was performed with 6F11, RBT11, or 1D5 clones, for PR either 16, RBT22, or PgR636 clones were used. All Her2 assessments were made using HercepTest, Dako (Glostrup, Denmark), and the manufacturer protocol. Equivocal Her2 results were considered in the analysis only if gene amplification was confirmed or discarded by either chromogenic (CISH) or fluorescent (FISH) in situ hybridization.
Disagreement among CNB and EB was considered when a change from noninvasive to invasive carcinoma or vice versa was observed in hematoxylin and eosin sampling or when a change from positive/negative in incisional biopsy turned on negative/positive in excisional resection for hormonal receptors (HR) and Her2 determinations.
Statistical analyses
Descriptive statistics were employed to summarize the characteristics of the study population. Discordant reports led us to review all samples from that patient to ascertain pre analytical, analytical, or post-analytical reasons. Concordance analysis by means of a kappa coefficient test was performed using SPSS 15.0 for Windows. Sensibility, specificity, and predictive values were calculated considering excisional resections as the gold standard. Significance test was two sided.
Results
Clinicopathologic characteristics of the study population
We recorded 424 surgical specimens from 321 patients, and 70 women with both CNB and surgical excision turn out to be the body of this study. When a patient had more than two breast biopsies, the one obtained closer to CNB sampling was preferred, if complete pathologic response was observed in the final excision, the case was discarded. Median age was 58 (36-87) years. Initial diagnosis of the patients in CNB was invasive ductal 56 (80%), lobular 10 (14%), and mixed 4 (6%) carcinomas; four were in situ lesions. Half of the patients had stage I/II disease, and neoadjuvant chemotherapy was received by 38 (54%).
The concordance rate of primary diagnosis of in situ or invasive cancer between the core biopsy and the excision specimens was 97%. Nine discordances in morphologic diagnosis were explained by mixed phenotypes of the invasive carcinoma in four EB (two ductal in CNB become lobular in EB; two lobular in CNB become ductal in EB), and because only two in situ lesions were observed in CNB, whereas infiltrative carcinoma admixed with the in situ lesion was identified in EB or because in 2 CNB of invasive ductal carcinoma, only in situ carcinoma was observed in EB. The one left was a poorly sampled CNB (one tissue fragment) with crushing artifact diagnosed as ductal invasive, which turned out to be invasive lobular carcinoma in the final excision.
IHC findings
Six pairs of samples were excluded from the IHC analysis because the surgical pathology report of CNB (2) and EB (4) did not include ER, PR, or Her2 determinations. In addition, four Her2 misleading samples lacking CISH/FISH amplification procedures were excluded. Thus, concordance analysis was based on 64 patients for HR and in 60 pairs of samples for Her2. A concordance of 92% (kappa 0.75), 75% (kappa 0.26), and 67% (kappa 0.47) was observed for ER, PR, and its two determinations between CNB and EB, and the positive predictive values were 0.96, 0.89, and 0.44, respectively.
ER discordances between CNB and EB in six patients were due to tumor heterogeneity with weak nuclear staining (2 cases), associated with neoadjuvant treatment (1 case), as well as with technical variations in IHC, leading to slides without adequate external controls (2 cases) and misinterpretation of the slide by the original observer (1 case). Discordances with PR were identified in 18 patients and its source attributed to technical and reading mistakes (7 each), neoadjuvant chemotherapy (4 cases), and signal heterogeneity (3 cases); some pairs having more than one factor. Her2 variations were observed in seven patients among CNB and EB. Variations were associated with tumor heterogeneity (3 cases), technical (3 cases), and reading errors (3 cases). Any discordance in diagnosis and HR/Her2 interpretation was more frequent if the number of tissue fragments by CNB sampling was less than three 29/31 (93%) versus 8/39 (21%), (p < 0.0001).
Discordances in HR and Her2 determinations were observed in 27 patients; the consequences of these variations were clinically irrelevant in 26 because hormonal treatment was granted using ER or PR results in CNB and Her2 treatment based on a positive result in either core biopsy or excision specimen. Serious implications were observed in one patient (Fig. 1).
Discussion
A concordance series of CNB and excisional resections of malignant breast tumors, -without excluding those submitted to the current neoadjuvant therapies, accurately reveal the day-to-day surgical pathology practice. None of the published articles5-16 (Table 1) is based on manual IHC and frequently do not mention the type of clones used. The present series include consecutive cases, even those receiving neoadjuvant therapies without complete pathologic response and, performing the IHC procedures manually. This is a common practice in countries with limited resources.
Table 1 Concordance of core needle and excisional biopsy interpretation of immunohistochemistry performed on infiltrative lesions
| Author | Countryn | n | IHC method | ER% | PR% | HER2% |
|---|---|---|---|---|---|---|
| Mann GB, et al., 20056 | Australia | 100 | A | 86 | 83 | 80 |
| Cahill RA, et al., 20067 | Ireland | 95 | U | 70 | 72 | 64 |
| Burge Ch, et al., 20068 | USA | 87 | A | 95 | 89 | 96 |
| Usami S, et al. 20079 | Japan | 111 | A | 95 | 88 | 88 |
| Wood B, et al., 200710 | Australia | 100 | U | 96 | 90 | 87 |
| Sutela A, et al., 200811 | Finland | 41 | A | 83 | 88 | 93 |
| Arnedos M, et al., 200912 | UK | 336 | U | 98 | 85 | 98 |
| Park SY, et al. ,200913 | Korea | 104 | U | 99 | 97 | 86 |
| Richer C, et al., 200914 | Germany | 164 | U | 75 | 75 | 54 |
| Tamaki K, et al., 201015 | Japan | 353 | A | 93 | 78 | 89 |
| Lorgis V, et al., 201116 | France | 175 | A | 84 | 78 | 98 |
| Ricci MD, et al., 201217 | Brazil | 69 | U | 95 | 87 | 78 |
| Present study | Mexico | 64 | M | 92 | 75 | 67 |
IHC: immunohistochemistry; A: automatic; M: manual; U: unknown.
Factors more frequently associated with discordances among both measurements were sampling and IHC omissions (pre-analytical) and interpretation of the slides (analytical). In contrast with previous studies5-17, tumor heterogeneity and neoadjuvant treatments were less related to discrepancies. Some information is available on the impact of tumor sampling in diagnosis and IHC interpretation in CNB. Patients with less than four tissue cores have higher probabilities of inadequate diagnosis/subcellular measurements, than those with more than four9,13,15,18. This finding move us to qualify the sampling we receive as adequate (≥ 4 cores) or inadequate (< 4 cores) in our surgical pathology report, to prospectively share possible disagreements in excisional specimens. As we did not find relevant clinical disagreements with HR in both measurements, only a warning was made to our technical staff to select and use in every run adequate positive/negative controls; a frequent omission observed in this review. Since hormonal blocking agents are prescribed using data from ER and PR, none of our tested patients with discordances was excluded of receiving this therapy.
A different picture was observed regarding Her2 with 33% of discordant evaluations in CNB and EB. A reading mistake excluded of specific blocking agent therapy to a patient in whom we observed the natural history of a Her2 type invasive ductal carcinoma. Reading mistakes, technical omissions, and heterogeneity in IHC expression contribute to Her2 discordances. Although within previous series concordance rates (Table 1)5-16, the observed Her2 discordances were too high for current pathology practice. As a consequence, all 1+, 2+, and 3+ Her2 IHC results will be submitted to chromogenic amplification assessments in this facility for the next 12 months, and a blind double check of the slides was advised for all breast samples. Mexico lacks of a national proficiency assessment programs of IHC and pathologist should contrast the IHC results with the morphology of the infiltrative lesion, i.e., consider features suggestive of Her2 discordances the presence of a positive result in a Grade 1 tumor. Recommendations sustained and shared in some previous reports14,19,20.
IHC automatization brings more interlaboratory consistency but without statistically significant differences in the estimation of nuclear HR percentage/intensity signal when compared with manual procedures21. Besides, the percentage of HR-positive cells assessment is more reliable than intensity signal interpretation, suggesting that combining percentage of positive cells and signal intensity as a score, lack of a practical advantage22. Budget availability is considered the main reason for IHC automation, our current cost of a manual slide is less than US 20 dollars, and it will rise more than the double with the marketed closed IHC platforms. We wager here for a model of morphological tests in sound with our pockets, in contrast with other scenarios23.
In conclusion, acceptable concordances were observed for diagnosis and HR, but worrisome results were identified for Her2 determination using handmade IHC protocols. The main sources of disagreements were analytical and poor sampling of suspicious breast lesions with fewer than three core needle biopsies.
