Touch Imprint Cytology in Tumor Tissue Banks for the Confirmation of Neoplastic Cellularity and for DNA Extraction Anita Mangia, PhD; Annalisa Chiriatti, PhD; Patrizia Chiarappa, PhD; Maria Angela Incalza, PhD; Giovanni Antonaci, MD; Brunella Pilato, PhD; Giovanni Simone, MD; Stefania Tommasi, PhD; Angelo Paradiso, MD
● Context.—Learning the characteristics of frozen tissue samples stored in tumor banks for biological studies remains a problem. Objective.—To assess the use of touch imprint cytology on fresh tissue samples as a rapid and reliable method of determining the presence and quantity of neoplastic cells before freezing. Design.—Touch imprint cytology was performed on 259 specimens of operable breast cancer. Touch imprints were prepared from fresh tissue specimens before freezing samples for storage. Each tumor sample was imprinted on a glass slide and stained with hematoxylin-eosin. Tumor cellularity was quantified as negative, poor, moderate, or rich. Results.—A significant correlation was found between samples with a tumor size greater than 2 cm and high tumor cellularity (P ⴝ .03; 2 test). Furthermore, 35% of ductal tumors showed higher tumor cellularity compared with lobular tumors (P ⬍ .001; 2 test). No association was found between lymph node status and tumor grade. When samples for which more than 2 imprints were available
were examined, tumor cellularity among imprints of the same sample showed an overall agreement of 0.67 (P ⬍ .001; statistic). It was also determined that the higher the cellularity, the higher the agreement. Our data also showed concordance of 0.87 (P ⬍ .001; statistic) between touch imprint cytology imprints and histologic sections from contiguous tumor. Moreover, 11 randomly selected samples underwent DNA extraction, polymerase chain reaction, and sequencing to verify the feasibility of DNA analyses. We found that DNA from touch imprint cytology was amplifiable and suitable for direct sequencing. Conclusions.—Touch imprint cytology may represent an important step in the quality control of tumor cellularity of breast cancer specimens designed to be stored in tumor biobanks and a valid method for assessing the suitability of such tissue for further biomorphologic and biomolecular applications. (Arch Pathol Lab Med. 2008;132:974–978)
F
and the paraffin block can then be considered available for research. Alternatively, QC can be performed on the fixed sections taken for diagnostic review. With this method, it is critical that the tissue taken for research be adjacent to the diagnostic sections used for QC. Fifteen percent of specimens2 collected for research are not suitable for their intended use. For example, a specimen collected as tumor may be necrotic and be unsuitable for research. Similarly, a specimen collected as uninvolved tissue may contain tumor in the lymphatic vessels, making the specimen unsuitable for biological analysis. Regardless of the method used, it is imperative that some form of QC be performed to confirm the diagnosis of the original specimen intended for research use. It is estimated that in 2004,3 as many as 10% of clinical laboratory tests were based on the analysis of DNA, RNA, or proteins from tumor, and control tissues. The reliability of test results requires high-quality tissues that are handled according to standard operative procedures and that are analyzed for morphology and cellularity. We studied the use of TIC to determine the presence and quality of neoplastic cells and the quality of the extracted DNA. We specifically wanted to determine wheth-
or decades, touch imprint cytology (TIC) has been used as a diagnostic tool for breast cancer.1 The simplicity, speed, and cost-effectiveness of the technique, along with its ability to maximize cell recovery from very small tissue pieces, make TIC a valuable resource for virtually every aspect of experimental and diagnostic medicine. In particular, TIC seems ideal for quality control (QC) of tumor cellularity on a tissue sample. The QC of specimens for tissue banks can be based on a routine hematoxylin-eosin (H&E) section taken adjacent to the tissue processed for research. After review, the H&E-stained slide can be retained as a permanent record,
Accepted for publication November 28, 2007. From the Clinical Experimental Oncology Laboratory (Drs Mangia, Chiriatti, Chiarappa, Incalza, Pilato, Tommasi, and Paradiso) and the Cytopathology Unit (Dr Simone), National Cancer Institute, Bari, Italy; and the Department of Internal Medicine, Immunology and Infectious Disease, University of Bari Medical School, Bari, Italy. The authors have no relevant financial interest in the products or companies described in this article. Reprints: Anita Mangia, PhD, Clinical Experimental Oncology Laboratory, National Cancer Institute, Via Hanheman, 10, 70126, Bari, Italy (e-mail:
[email protected]). 974 Arch Pathol Lab Med—Vol 132, June 2008
Touch Imprint Cytology in Tumor Tissue Banks—Mangia et al
Table 1.
Classification of Tissue Samples After Surgical Resection No. (%) of Cases
Candidates for tissue bank storage Touch imprint cytology performed
553 259 (47)
Unsuitable samples 294 (53) Small tumor size* 139 (25) HCV⫹/HBsAg⫹† 100 (18) Organizational problems 55 (10) * pT1a–pT1b. † HCV indicates hepatitis C virus; HBsAg, hepatitis B surface antigen.
er TIC could represent both an inexpensive and a rapid method for maximizing cell recovery when very small pieces of tissue are available, and a potential source of DNA for use in further molecular investigations. MATERIALS AND METHODS Preparation and Analysis of TIC During 2005, 890 patients with a first diagnosis of operable breast cancer underwent surgery at our institute. Five hundred fifty-three breast cancers (62%) were sampled and frozen for tissue bank storage; 337 (38%) could not be used because of small tumor size (23%) or management problems (15%). Touch imprint cytology was performed on 259 (47%) of 553 patients. The other 294 (53%) samples were unsuitable for the following reasons: organizational problems (10%), small tumor size (25%), or positive for hepatitis C virus (HCV⫹) or for hepatitis B surface antigen (HBsAg⫹) (18%) (Table 1). The main pathologic characteristics of the tumors are summarized in Table 2. A total of 259 consecutive patients with operable breast cancer were enrolled in our study. Within a few minutes of surgical resection, the pathologic material was analyzed macroscopically, and contiguous fragments of tumor tissue were used for morphohistologic determinations and TIC assays before being stored at ⫺80⬚C. Each specimen was imprinted using gentle pressure on a glass slide, which was then air dried and stained with H&E (Figure 1, a and b) to ensure preservation of the morphology for TIC and good correspondence with an H&E-stained section from contiguous tumor. Tumors with a sufficient quantity of patholog-
Table 2.
Figure 1. Touch imprint cytology from a fresh human breast cancer sample with rich tumor cellularity (hematoxylin-eosin, original magnifications ⫻200 [a] and ⫻400 [b]).
Relationship Between Pathologic Characteristics of Tumors and Tumor Cellularity by Touch Imprint Cytology (TIC) Tumor Cellularity by TIC, No. (%) of Cases
Variable
Total No.
Negative (n ⴝ 13)
Poor (n ⴝ 83)
Moderate (n ⴝ 78)
Rich (n ⴝ 85)
P Value by 2
Tumor size, cm ⱕ2 ⬎2
112 147
5 (4) 8 (5)
45 (40) 38 (26)
35 (32) 43 (29)
27 (24)* 58 (40)*
.03
Differentiation G1 G2 G3 Not recorded
28 119 81 31
1 (3) 4 (3) 2 (3)
11 (39) 38 (32) 19 (23)
8 (29) 37 (31) 25 (31)
8 (29) 40 (34) 35 (43)
.56
Lymph node status Negative Positive Biopsy
105 131 23
3 (3) 7 (5)
41 (39) 35 (27)
35 (33) 39 (30)
26 (25) 50 (38)
.58
72 (31) 3 (16) 3 (37)
82 (35)† 0 (0)† 3 (37)
⬍.001
Histology Ductal 232 9 (4) 69 (30) Lobular 19 4 (21) 12 (63) All other subtypes 8 0 (0) 2 (24) * Association between tumor size and tumor cellularity. † Ductal tumor cellularity with respect to lobular tumor cellularity. Arch Pathol Lab Med—Vol 132, June 2008
Touch Imprint Cytology in Tumor Tissue Banks—Mangia et al 975
Table 3. Agreement of Cellularity Between Different Touch Imprints From the Same Tumor* Cellularity of First Touch Imprint
Cellularity of Repeated Touch Imprints Negative
Negative Poor Moderate Rich * Overall agreement
Poor
Moderate
Rich
9 3 2 0 88 22 1 28 79 0 4 44 ⫽ 0.67 (P ⬍ .001; statistic).
0 4 26 97
Table 4. Agreement Between Cellularity of Touch Imprint Cytology and Histologic Section* Cellularity of the Touch Imprint Cytology
Cellularity of the Histologic Section Negative
Negative Poor Moderate Rich * Overall agreement ⫽
Table 5. Sample No.
239A 249A 237A 238A 182A 224A 244A 206A 225A 201A 175A
Poor
Moderate
0 1 7 0 79 4 0 4 70 0 0 10 0.87 (P ⬍ .001; statistic).
Rich
5 0 4 75
Quality and Quantity of the Extracted DNA Tumor Cellularity
DNA, ng/L
DNA Ratio, A260/A280
⫹⫹/⫹⫹⫹ ⫹⫹⫹ ⫹⫹/⫹⫹⫹ ⫹⫹ ⫹⫹⫹ ⫹⫹ ⫹⫹⫹ ⫹⫹⫹ ⫹⫹ ⫹⫹⫹ ⫹⫹⫹ Mean
26.285 31.657 35.507 41.125 45.867 58.607 59.500 67.9 72.730 136.39 203.15 70.79255
1.32 1.58 1.12 1.57 1.46 1.52 1.65 1.7 1.29 1.82 1.16 1.47
ic tissue were used to obtain more than one imprinted slide, which allowed for different assays. The touch imprints were stained with H&E and analyzed using a microscope. The areas of highest density were chosen, and necrotic areas were avoided. Tumor cellularity was assessed by 2 investigators in at least 10 different fields (⫻400 magnification) and was recorded as negative (no tumor cells despite adequate material), poor (⬍10 cells per field), moderate (10–50 cells per field), or rich (⬎50 cells per field). The touch imprints were also compared with the H&E-stained histologic sections from formalin-fixed, paraffin-embedded primary tumor specimens taken contiguous to the tumor sample that were designated for the tissue bank and were analyzed by a pathologist. These procedures satisfy the requirements of a modern tissue bank, and they should work on prospective and consecutive series of samples. Tumor classification was performed according to the World Health Organization criteria.
DNA Extraction Eleven randomly selected samples underwent DNA extraction. From each specimen, 4 touch imprints were prepared. Among these, 1 was stained with H&E to evaluate the percentage of tumor cellularity, while the others underwent DNA extraction. The imprints were rehydrated by incubating for 5 minutes in 70% ethanol-OH, 5 minutes in 50% ethanol-OH, and then 10 minutes in distilled water. Samples were scraped from the slide with 976 Arch Pathol Lab Med—Vol 132, June 2008
Figure 2. a and b, Results of 0.8% agarose gel electrophoresis of genomic DNA extracted from 11 (8 ⫹ 3) samples using touch imprint cytology. Figure 3. Results of 0.8% agarose gel electrophoresis of amplicons from 11 samples extracted by touch imprint cytology.
a scalpel blade, and DNA was extracted according to the QIAamp DNA micro kit protocol (Qiagen, Valencia, Calif). DNA quality and quantity were evaluated spectrophotometrically and on 0.8% agarose gel in Tris/boric acid/EDTA buffer.
Polymerase Chain Reaction and Sequencing To evaluate the suitability of TIC for further molecular studies, exon 20 of the BRCA1 gene was investigated for mutation analysis. Polymerase chain reaction was performed using 150 ng of DNA4 extracted by TIC. Furthermore, the amplified exon was used for direct automatic sequencing (377, Applied BioSystems, Foster City, Calif), as described elsewhere.4 We performed all of the analyses at the Clinical Experimental Oncology Laboratory of the National Cancer Institute of Bari, Italy. The Laboratory is ISO 9001-2000 certified (DNV Certificate No.: CERT-17885-2006AQ-BRI-SINCERT). Touch Imprint Cytology in Tumor Tissue Banks—Mangia et al
Figure 4. a and b, Examples of electropherogram obtained from 1 sample. No mutations were present.
Statistical Analysis The statistical association between the pathologic characteristics of the tumors and tumor cellularity was assessed using the 2 test. In addition, we used the (kappa) statistic as a measure of the agreement between different imprints from the same tumor sample and between TIC and histologic sections in evaluating tumor cellularity. Differences were considered to be significant when the P value was less than .05. Statistical analyses were performed using SPSS statistical software (SPSS, Inc, Chicago, Ill).
RESULTS Using TIC, a significant association was found between samples with a tumor size of greater than 2 cm and high tumor cellularity (40% in tumors ⬎2 cm vs 24% in tumors ⱕ2 cm; P ⫽ .03; 2 test). Furthermore, 35% of ductal carcinomas versus 0% of lobular carcinomas showed high tumor cellularity (P ⬍ .001; 2 test). No association was found between lymph node status and tumor grade. When cellularity was evaluated from different imprints (up to 5) obtained from the same sample, an overall agreement of 0.67 (P ⬍ .001; statistic) was found between the Arch Pathol Lab Med—Vol 132, June 2008
first touch imprint and the other imprints; the higher the cellularity, the higher the agreement (Table 3). Moreover, concordance of 0.87 (P ⬍ .001; statistic) between tumor cellularity evaluated with TIC and tumor cellularity of the corresponding histologic section was also demonstrated (Table 4). Only specimens with moderate or high tumor cellularity were used for DNA extraction. The mean yield of DNA obtained from TIC was 70.8 ng/L (range, 26.3–203.1 ng/ L), with a ratio A260/A280 (range, 1.12–1.82) (Table 5) on 0.8% agarose gel (Figure 2, a and b). The DNA was mostly of high molecular weight (more than 20 kb), even when a smear was present. The DNA integrity of sample 175A was not evaluated, probably because its high concentration caused failure of evaluation. Highly efficient polymerase chain reaction was obtained from all samples (Figure 3), and these amplicons were suitable for BRCA1 exon 20 mutational analysis performed on 4 samples (225A, 238A, 244A, 249A). As shown in Figure 4, a and b, high-quality electropherograms were obtained from all samples, even if no mutations were present. Touch Imprint Cytology in Tumor Tissue Banks—Mangia et al 977
COMMENT In our study, touch imprints from 259 breast cancer samples were used to evaluate the reliability of TIC in assessing the quantity of tumor cellularity and the quality of extracted DNA. This technique is simple and cost-effective and appears to be as reliable as other methods, such as frozen section preparation; however, TIC uses less tissue and is less time consuming. The data in Table 2 suggest that the percentage of tumors unsuitable for TIC due to negative neoplastic cellularity was not significantly different for small (ⱕ2 cm) and large (⬎2 cm) tumors. We preliminarily evaluated the probability of finding tumor cells in tissue that appeared normal on gross examination. We found tumor cells in grossly normal and adjacent tissues only occasionally and in a low percentage absolutely. In this pilot study, small pT1a-pT1b samples did not reach our laboratory because they were directly fixed in formalin by the pathologist. Our preliminary results led to the hypothesis that TIC could be used on tumors of any size before fixation. Furthermore, tissue from hepatitis patients was excluded from the analysis to limit any biological risk to the technicians. In an ongoing study, our tissue banking is now being performed in HCV⫹ and HBsAg⫹ cases. We confirmed that taking touch imprints from breast tumor samples constitutes a feasible way of obtaining fresh material to evaluate tumor cellularity. We found a strong correlation between tumor cellularity evaluated using TIC and tumor cellularity determined from sections cut from a contiguous sample (Table 4). Touch imprint cytology can be used as a rapid, reliable, alternative method to histology in evaluating breast tumor cellularity. Slides can be archived for further consultation before making molecular determinations on corresponding frozen samples. Since the late 1960s, TIC preparations have been used as an alternative to frozen section examination of tissues that are submitted for intraoperative diagnosis,5,6 with up to 100% accuracy.7 These data were confirmed in a recent study that demonstrated the accuracy of intraoperative imprint cytology for sentinel lymph node evaluation in the treatment of breast carcinoma.8 Previous studies have already compared histologic diagnosis performed from TIC imprints to that performed from standard histologic samples, and demonstrated the reliability of the cytotechnique.9,10 Moreover, Khanna et al11 compared TIC, fine-needle aspiration cytology, and tru-cut needle biopsy with tissue biopsy and found that imprint cytology had a sensitivity of 98.4% and a specificity of 100%. We also confirm the possible use of TIC for molecular determinations. In fact, the originality of our work consists in the possibility of using TIC to extract DNA qualitatively suitable for even the most sophisticated molecular analysis. Touch imprint cytology has already been used for different in situ techniques, such as telomerase activity determination,12 fluorescent in situ hybridization,13 and ploidy analysis.14,15 Furthermore, Kovach et al16 demonstrated the possibility of detecting and analyzing p53 point mu-
978 Arch Pathol Lab Med—Vol 132, June 2008
tations directly and without previous DNA extraction in cell clusters derived from touch imprints of adenocarcinomas of the human breast, demonstrating the feasibility of using TIC for molecular determinations. Finally, our study has demonstrated for the first time the ability to extract DNA of good quality and in good yield from touch imprints of breast cancers, even though the 0.8% agarose gel demonstrated a slight degradation. The resulting quantity of DNA was sufficient and suitable for further applications in which the presence of ‘‘clear’’ DNA is necessary, such as mutational screening by sequencing. A second important result of the present study is that all of the analyses can be performed without consuming precious tumor material. In conclusion, our data indicate that TIC may represent an important step in the QC of breast cancer tissues for research and biobanking activity. Moreover, TIC appears to be a good tool for analysis of tumor cellularity and for DNA extraction. The authors are grateful to Nicola Paciolla, MSc, and to Mrs Vincenza Rubini for their technical collaboration. We thank Mr Baldassarre Stea for his valid statistical support and Filippo Menolascina, BSc, of the Bioinformatics Research Group (NCI-Bari) for his valuable support. We would also like to thank Silvana Valerio, MS, for her assistance in the preparation of this manuscript. References 1. Helpap B, Tschubel K. The significance of the imprint cytology in breast biopsy diagnosis. Acta Cytol. 1978;22:133–137. 2. Grizzle WE, Aamodt R, Clausen K, LiVolsi V, Pretlow TG, Qualman S. Providing human tissues for research. Arch Pathol Lab Med. 1998;122:1065– 1076. 3. Qualman SJ, France M, Grizzle WE, et al. Establishing a tumor bank: banking, informatics and ethics. Br J Cancer. 2004;90:1115–1119. 4. Tommasi S, Crapolicchio A, Lacalamita R, et al. BRCA1 mutations and polymorphisms in a hospital-based consecutive series of breast cancer patients from Apulia, Italy. Mutat Res. 2005;15;578:395–405. 5. Geelhoed GW, Silverberg SG. Intraoperative imprints for the identification of parathyroid tissue. Surgery. 1984;96:1124–1131. 6. Suen KC, Wood WS, Syed AA, Quenville NF, Clement PB. Role of imprint cytology in intraoperative diagnosis: value and limitations. J Clin Pathol. 1978; 31:328–337. 7. Silverberg SG. Imprints in the intra-operative evaluation of parathyroid disease. Arch Pathol. 1975;99:375–378. 8. Chicken DV, Kocjan G, Falzon M, et al. Intraoperative touch imprint cytology for the diagnosis of sentinel lymph node metastases in breast cancer. Br J Surg. 2006;93:572–576. 9. Aryya NC, Khanna S, Shukla HS, Tripathi FM, Shukla VK. Role of rapid imprint cytology in the diagnosis of skin cancer and assessment of adequacy of excision. Indian J Pathol Microbiol. 1992;35:108–112. 10. Paulose RR, Shee CD, Abdelhadi IA, Khan MK. Accuracy of touch imprint cytology in diagnosing lung cancer. Cytopathology. 2004;15:109–112. 11. Khanna AK, Singh MR, Khanna S, Khanna NN. Fine needle aspiration cytology, imprint cytology and tru-cut needle biopsy in breast lumps: a comparative evaluation. J Indian Med Assoc. 1991;89:192–195. 12. Chieco P, Bertaccini A, Giovannini C, Stecca BA, Martorana G. Telomerase activity in touch-imprint cell preparations from fresh prostate needle biopsy specimens. Eur Urol. 2001;40:666–672. 13. Moore JG, To V, Patel SJ, Sneige N. HER-2/neu gene amplification in breast imprint cytology analyzed by fluorescence in situ hybridization: direct comparison with companion tissue sections. Diagn Cytopathol. 2000;23:299–302. 14. Hayes SJ, Hinchliffe SA, Pope JD, et al. Ploidy analysis on Wilms’ tumor touch imprints using ethidium bromide and automated image analysis integrated confocal laser scanning microscopy. Virchows Arch. 1995;427:101–104. 15. Mainguene C, Choquenet C, Deplano C, et al. DNA ploidy by image cytometry in urothelial carcinomas: comparison of touch imprints and paraffinembedded biopsies from 31 patients. Anal Quant Cytol Histol. 1997;19:437–442. 16. Kovach JS, McGovern R, Cassady J, Swanson S, Sommer S. Direct sequencing from touch preparations of human carcinomas: analysis of p53 mutations in gastric carcinoma by touch preparation. J Natl Cancer Inst. 1991;83: 1004–1009.
Touch Imprint Cytology in Tumor Tissue Banks—Mangia et al