Proficiency Testing for Laboratories Performing Fluorescence In Situ Hybridization With Chromosome-Specific DNA Probes

Transcription

1 Proficiency Testing for Laboratories Performing Fluorescence In Situ Hybridization With Chromosome-Specific DN Probes James T. Mascarello, PhD; rthur R. rothman, PhD; Keri Davison, S; Gordon W. Dewald, PhD; Marille Herrman, MD, PhD; Danette McCandless, MD; Jonathan P. Park, PhD; Diane L. Persons, MD; Kathleen W. Rao, PhD; Nancy R. Schneider, MD, PhD; Gail H. Vance, MD; Linda D. Cooley, MD; for the Cytogenetics Resource Committee of the College of merican Pathologists and merican College of Medical Genetics Objective. To assess laboratory performance, use, and limitations in the joint College of merican Pathologists and merican College of Medical Genetics proficiency testing program for laboratories performing cytogenetic tests based on fluorescence in situ hybridization (FISH). Data Sources. Eight proficiency surveys dealing with FISH detection of microdeletions or microduplications, aneuploidy in interphase cells, gene amplification, and neoplasm-specific translocations. Participating laboratories used their own DN probes (commercial or home-brew), hybridization methods, and analytic criteria to answer clinical questions about cases represented by slides included in the survey materials. They also described their test results according to the International System for Human Cytogenetic Nomenclature (ISCN) and answered supplementary questions relating to their experience with the subject test systems. Data Extraction. In addition to evaluating diagnostic accuracy, we evaluated survey use, laboratory experience, Fluorescence in situ hybridization (FISH) using chromosome-specific probes has become an important cytogenetic tool in the evaluation of many congenital (con- ccepted for publication June, 22. From the Genetic Services, Children s Hospital, San Diego, Calif (Dr Mascarello); the Cytogenetics Laboratory, University of Utah Medical Center, Salt Lake City, Utah (Dr rothman); the College of merican Pathologists, Northfield, Ill (Ms Davison); the Division of Laboratory Genetics, Mayo Clinic, Rochester, Minn (Dr Dewald); the Department of Pathology, University of Pennsylvania School of Medicine, Philadelphia, Pa (Dr Herrman); the Department of Medical & Molecular Genetics, Indiana University Medical Center, Indianapolis, Ind (Drs McCandless and Vance); the Department of Pathology, Dartmouth- Hitchcock Medical Center, Lebanon, NH (Dr Park); the Department of Pathology and Laboratory Medicine, University of Kansas Medical Center, Kansas City, Kan (Dr Persons); the Department of Pediatrics, University of North Carolina, Chapel Hill, NC (Dr Rao); the Department of Pathology, University of Texas Southwestern Medical Center, Dallas, Tex (Dr Schneider); and the Section of Medical Genetics and Molecular Medicine, Children s Mercy Hospital, Kansas City, Mo (Dr Cooley). Each author is or was a member or consultant of the CP-CMG Cytogenetics Resource Committee. variation in methodologic approach, and the practicality of using ISCN nomenclature for describing test results. Synthesis and Conclusions. With the exception of one challenge, at least 8% of the participants reached the correct diagnostic conclusion. In the sole exception, there was still a consensus of 91.7% of participants with the same (albeit erroneous) diagnostic conclusion. The overall outstanding performance of participating laboratories clearly shows the reliability of current FISH methods. Despite the fact that a large number of laboratories reported little or no experience with the specific test systems, the overwhelming majority performed very well. This result shows that the program s strategy of targeting classes of abnormalities (vs a single abnormality associated with a specific disease) did not put at a disadvantage participants who did not routinely perform all of the potential tests in the class. The extraordinary variation in ISCN descriptions submitted by participants showed that the existing system for human cytogenetic nomenclature is not suitable for facile communication of FISH test results. (rch Pathol Lab Med. 22;126: ) stitutional) disorders, hematologic malignancies, and some solid tumors. The Cytogenetics Resource Committee (CyRC) of the College of merican Pathologists and the merican College of Medical Genetics (CP-CMG) assessed the demand for proficiency testing and the feasibility of providing test materials in 3 pilot surveys that were offered in 199 and ased on the favorable response to these pilot surveys, the CyRC began offering regular proficiency surveys for chromosome-specific FISH in October To date, the CyRC has completed and evaluated 8 proficiency surveys. The purpose of this report is to review laboratory performance, use, and limitations of the surveys offered in this program. MTERILS ND METHODS Cells used in the CYF and CYG FISH surveys were obtained from blood, lymphoblastoid cell lines, and amniocyte cultures. Reprints: James T. Mascarello, PhD, Genetic Services, Children s Hospital, 32 Children s Way, San Diego, C ( jmascarello@ chsd.org). 148 rch Pathol Lab Med Vol 126, December 22 Proficiency Testing for Laboratories Performing FISH Mascarello et al

2 Survey 1997 CYF 1998 CYF 1998 CYG 1999 CYF 1999 CYG 2 CYF 2 CYG 2 CYH Table 1. Fluorescence In Situ Hybridization Surveys: Use and Participant Performance Description Microdeletion of chromosome 22 in DiGeorge/velocardiofacial syndrome CR-L gene fusion in chronic myeloid Microdeletion of chromosome 7 in Williams syndrome PML-RR gene fusion in acute promyelocytic Prenatal detection of aneuploidy in uncultured amniocytes TEL-ML1 gene fusion in acute lymphoblastic Microdeletion and duplication of chromosome 1 HER-2/neu gene amplification in carcinoma of the breast No. of Survey Kits Returned/Shipped 1/ /134 19/121 18/136 17/124 13/ /11 3/3 Test Condition 98% Neoplastic 28% Neoplastic 92% Neoplastic 19% Neoplastic Trisomy 21 Triploidy % Neoplastic 2% Neoplastic Duplicated No amplification mplified Correct Interpretations, % For some hematologic disease challenges, normal and abnormal cells were mixed to test the participants ability to detect abnormal cells at varying frequencies. The slides for the CYH survey were sections of paraffin-embedded tissue from 2 patients with invasive carcinoma of the breast. ll slides used in the CYF and CYG FISH challenges were prepared at the Mayo Clinic using standard procedures. To control the quality of the slides, cells were dropped on dry slides with temperature and humidity maintained at C and %, respectively. Slides were stored at 2 C until they were shipped to participants using 2-day priority mail but were not aged by any other process. The slide sets for 2 CYH were prepared at the Cleveland Clinic and consisted of 3 serial sections. To help participants focus their analyses on areas of abnormal histology, 1 of the 3 sections was intended for hematoxylin-eosin staining. Participants were instructed to use the same DN probes, hybridization protocols, and scoring criteria that they would use for comparable clinical studies. In addition to a diagnostic interpretation, participants were asked to report the source of their probes, the number of cells analyzed, the percentage of normal and abnormal cells, and, except for CYH, the FISH nomenclature (International System for Human Cytogenetic Nomenclature [ISCN] ) that described their test results. In 6 surveys, they were also asked to answer supplementary questions regarding their experience with the FISH test that was the subject of the challenge. RESULTS Table 1 provides details on the nature of the 8 FISH surveys as well as the number of participants and the frequency of correct interpretations. Note that other interpretations included not only incorrect diagnoses but also interpretations indicative of equivocal results or technical problems. More than 92% of participants were from the United States; the rest were from Canada, South Korea, Singapore, Japan, razil, Spain, ustralia, Cyprus, Germany, and a few other countries. Except for the recent survey dealing with HER-2/neu gene amplification in breast cancer, the number of participants has been relatively constant and at a level that makes survey production economically practical. We could not measure the adequacy of the sample materials included in the FISH surveys except to say that reports of broken or otherwise inadequate materials were very rare. The fact that some laboratories purchased the survey kit but did not return result forms is more likely to be a function of the laboratory not performing the test included in the survey than a problem with the sample materials. The challenges can be grouped into 4 categories: detection of microdeletions or microduplications in metaphase cells, detection of the presence of cells with neoplasia-related translocations in an interphase cell population, enumeration of chromosomes in interphase cells, and identification of gene amplification in tissue sections from breast carcinoma. ased on their FISH test results, participants were asked to choose from a list of possible diagnostic conclusions. eginning in 1999, responses to these questions were reviewed and graded according to the standards for nonregulated analytes. Surveys 1997 CYF, 1998 CYG, and 2 CYG challenged participants ability to detect the presence or absence of microdeletions responsible for velocardiofacial syndrome, Williams syndrome, and Prader-Willi syndrome, respectively. The deletions that give rise to these disorders remove DN sequences that can be targeted by fluorescent DN probes. Cells without the deletion in question have a fluorescent signal on both homologous chromosomes, whereas cells with the deletion have a signal on only one chromosome. ecause absence of a FISH signal can have an artifactual basis, most of the laboratories that participated in these surveys used probe cocktails that simultaneously targeted the potentially deleted region and a control locus located on the same chromosome but outside the deleted region. In the 1997 CYF and 1998 CYG surveys, participants were told the specific diagnosis being considered. oth surveys included slides from one affected and one unaffected individual. In neither survey did any participant report a normal result in a patient known to have a microdeletion (ie, false-negative test results). In contrast, 2 participants in 1997 CYF and 1998 CYG interpreted their results as indicative of a deletion in the normal sample (ie, false-positive results). In an attempt to encompass more of the cytogenetic testing process, 2 CYG participants were asked to select a FISH probe on the basis of their evaluation of photographs rch Pathol Lab Med Vol 126, December 22 Proficiency Testing for Laboratories Performing FISH Mascarello et al 149

3 of G-banded chromosomes that were included in the test kit. The photographs and 1 of the 2 slide sets were from a patient with the chromosome 1 microdeletion commonly found in patients with Prader-Willi syndrome. The second slide set was from an individual who had a duplication of the same region. Remarkably, every laboratory that reported results selected a DN probe suitable for evaluating the Prader-Willi syndrome region and correctly interpreted the first slide set as being abnormal. On the other hand, only 12 laboratories (8.3%) concluded that there was an abnormality in the second slide set, and only half of this number correctly recognized the duplication. lthough it is not possible to know why nearly 92% of the participants reported normal results, it is clear that the test materials and the test result form itself had a role. Duplication of the Prader-Willi syndrome region is reliably detected only in interphase cells or in the extended chromosomes of prophase or prometaphase. 3 Unfortunately, the slide set contained very few prophase or prometaphase cells. Moreover, ungraded questions in the test result form implied that interpretations for both slide sets needed to be based on the analysis of mitotic cells. Survey 1999 CYG was intended to simulate the prenatal diagnosis of aneuploidy in amniocytes. Chromosome-specific probes can be used to enumerate chromosomes in amniocytes prepared directly from amniotic fluid. For example, amniocytes from a male fetus with Down syndrome (trisomy 21) when tested with probes for chromosomes X, Y, 13, 18, and 21 will exhibit 1 probe signal for the X, 1 probe signal for the Y, 2 probe signals for the 13, 2 probe signals for the 18, and 3 probe signals for chromosome 21. To distinguish the signal of one chromosome from another, multiple slides and, in most laboratories, cocktails of probes labeled with different colors are used. ecause amniotic fluid is an impractical source of cells for proficiency testing, this survey relied on amniocyte cultures. The presence of a homogeneous cell population and the absence of the debris normally present in amniotic fluid made these slides substantially less challenging than slides used in routine clinical service. One of the cultures from which slides were prepared was initiated with cells from a pregnancy with trisomy 21. The other was from a pregnancy with triploidy. Tabulation of the results from this survey was complicated by the fact that participants were allowed to choose any combination of probes and target chromosomes that they felt were appropriate for the clinical information and consistent with the tests offered by their laboratories. Consequently, some participants tested for only a subset of the common aneuploidies (sex chromosomes, 13, 18, and 21). Every laboratory that tested for trisomy 21 in the Down syndrome case correctly reported the trisomy (ie, no falsenegative results). Every participant recognized that the second case was also abnormal, but 4 participants interpreted their results as consistent with a diagnosis other than triploidy. One reported trisomy 21, 1 reported a double trisomy, and 2 reported a triple trisomy. Surveys 1998 CYF, 1999 CYF, and 2 CYF challenged participants to detect interphase cells with translocations commonly seen in chronic myeloid (CML), acute promyelocytic, and pediatric acute lymphoblastic, respectively. Such testing relies on probe pairs that target the 2 translocation breakpoints and that can be distinguished on the basis of signal color. cells exhibit 2 signals for each of the probes, and all of the signals are usually well separated. Different probe combinations have somewhat different signal patterns for cells that contain translocations but, at a minimum, all produce at least one signal for which the alternate colors are closely opposed. With a mix that contains green and red probes, this often leads to a yellow signal. These surveys included slides with a high frequency of abnormal cells, slides with no abnormal cells, and slides with a relatively low frequency of abnormal cells. Participants were asked to determine whether their FISH results were consistent with translocation-bearing cells being present or absent in challenges of the 1999 and 2 surveys. In the CR-L survey, one challenge involving a low frequency of neoplastic cells was intended to simulate a patients with CML after treatment. Thus, one of the alternative interpretations was consistent with treated CML. In this survey, and in the PML-RR survey, participants were given response options that we have collectively lumped under uninterpretable test results. These options were used 26 times in the CR-L challenges and 1 times in the PML-RR challenges. Eight participants in the CR-L survey provided interpretations indicative of problems in the quantification of neoplastic cells. Three reported results consistent with treated CML for the high-concentration sample, and reported results consistent with untreated CML for the low-concentration sample. lthough the percentage of correct responses far exceeded 8%, these challenges did generate false-negative as well as false-positive results. Two participants reported neoplastic cells in the normal sample from the PML-RR survey, and one participant reported neoplastic cells in the normal sample from the TEL-ML1 survey. Three participants reported cells with the CR-L translocation in the normal sample from the CML survey. Two of these reported that the number of translocation-bearing cells was consistent with treated CML, while the other reported the number being consistent with untreated CML. ll false-negative results (no abnormal cells detected) in these 3 surveys came from the low-concentration samples. There was 1 such report in the CR-L survey, 4inthePML-RR survey, and 7 in the TEL-ML1 survey. There were no false-negative results for the high-concentration TEL-ML1 sample or for the high-concentration CR-L sample. However, there was 1 false-negative for the high-concentration PML-RR sample. The CYH 2 survey required participants to evaluate HER-2/neu amplification status in 2 sets of slides prepared from paraffin-embedded tissue sections. HER-2/neu amplification status has bearing on the prognosis and treatment of carcinoma of the breast. 4 FISH with a DN probe for the HER-2/neu gene produces 2 signals in normal cells and, usually, many times this number of signals in cells with HER-2/neu amplification. The survey kit contained one set of slides with HER-2/neu amplification and one without. ll of the 3 laboratories that participated in the survey correctly identified the presence or absence of amplification in both cases. In addition to providing diagnostic interpretations, FISH survey participants were also asked to provide answers to a variety of questions relating to the FISH methods used in their laboratories. The subjects of these questions included source of probes, number of cells analyzed, annual test volume, availability of written protocols, and availability of validated normal/abnormal ranges. Probes 146 rch Pathol Lab Med Vol 126, December 22 Proficiency Testing for Laboratories Performing FISH Mascarello et al

4 Table 2. FISH Proficiency Test Performance Is Not Readily Predicted From Laboratory QC Systems or Laboratory Experience* Microdeletions Leukemia Translocation Prenatal Diagnosis DG/VCF Williams CML PL LL neuploidy No. of laboratories Mean No. of tests per year % With tests per year % With 1 1 tests per year % With written QC system * Voluntary responses to questions regarding QC activities and previous year s experience. The column refers to laboratories that gave correct interpretations for all of the challenges in the survey. The column refers to laboratories that were unable to interpret or incorrectly interpreted at least one challenge in the survey. The data that describe the previous year s experience include experience only with the specific test addressed by the survey. FISH indicates fluorescence in situ hybridization; QC, quality control; DG/VCF, DiGeorge/velocardiofacial syndrome; CML, chronic myeloid ; PL, acute promyelocytic ; and LL, acute lymphoblastic were available from commercial sources for all of the survey challenges and, with rare exception, participating laboratories took advantage of these sources. The number of cells evaluated varied with the diagnostic question being asked. In the first microdeletion challenge (DiGeorge/velocardiofacial syndrome), participants were told to examine at least 2 metaphase cells, and most examined exactly that number. Similar guidance was not given in subsequent surveys, so there was considerable variation in the number of cells analyzed. In the second microdeletion survey (Williams syndrome), participants examined from as few as cells to as many as 1, but the modal number of cells examined was 2. The modal number of cells examined in the challenges was 2, with a range of 2 to 1. For the prenatal aneuploidy detection challenge, there was a bimodal distribution for the number of cells examined (per probe). Many laboratories examined approximately interphase cells, and many examined approximately 1 interphase cells. The range of cells examined (per probe) in this survey extended from a low of 1 to a high of 1. We examined whether performance on these proficiency tests was correlated with experience and with the presence/absence of systems for quality control. Table 2 shows responses to supplementary questions that have bearing on this question and compares laboratories that correctly interpreted all challenges in a particular survey with laboratories that could not interpret or incorrectly interpreted one or more challenge. Laboratories with all correct interpretations appeared to be more likely to have written quality control plans and, for most challenges, appeared to average more cases (of the same test) in the previous year, but the differences were not statistically significant (by Fisher exact test and Mann-Whitney U test, respectively). However, the table clearly illustrates that most laboratories had little or no experience with the specific test systems that were the subjects of the challenges and that, despite this fact, the overwhelming majority performed very well. In the CP-CMG surveys that deal with traditional cytogenetic analysis, participants use the ISCN to communicate their test results. This nomenclature is then interpreted by the CyRC to determine whether the submitted results are acceptable. Recognizing that it might be difficult to achieve a greater than 8% consensus response with ISCN FISH nomenclature, the CyRC, instead, asked participants to select their responses from a coded table of possible interpretations and then describe these results with nomenclature that was not graded. The frequency of the most common nomenclature string ranged from a high of 7% for the 1997 DiGeorge/velocardiofacial microdeletion survey to a low of 6.4% for the 1999 survey simulating aneuploidy in uncultured amniocytes. Differences in probe selection, lack of uniformity in the specification of the probe s target band, syntax errors, and uncertainty about how to describe situations not anticipated by the crafters of the 199 version of ISCN all contributed to the variation in nomenclature. COMMENT FISH is a powerful new technology that gives cytogenetic laboratories the means to detect chromosome abnormalities that are difficult or impossible to detect with traditional methods and the ability to evaluate cell types not previously amenable to cytogenetic analysis. lthough FISH methods have been in widespread use for over 1 years, the surveys described here represent the first largescale effort to provide proficiency testing for these methods. The sustained level of participation suggests that the program successfully met the FISH proficiency testing needs of most laboratories. Except for the recent survey dealing with HER-2/neu amplification, the number of participating laboratories was over 1. In the most recent microdeletion survey, a record number of 14 laboratories participated. Not all cytogenetic laboratories perform FISH testing. However, based on the number of FISH kits shipped, at least 4% of the (approximately 29) North merican laboratories participating in the CP-CMG standard cytogenetics survey are also participating in one or more FISH surveys. Survey result forms were returned by a great majority of participants, indicating that the test materials were of adequate quality and that the survey logistics were effective. Kit failures occurred rarely, and when they did occur, were easily resolved by replacement materials. s with any proficiency testing process, this one had limitations that directly bear on its validity. The most serious involve the nature of the test materials. In all of the surveys, participants received prepared slides. Problems that a laboratory might have with preanalytic variables or in preparing slides suitable for analysis would not be evident with this survey material. ll of the translocation surveys included artificial mixtures of cells from rch Pathol Lab Med Vol 126, December 22 Proficiency Testing for Laboratories Performing FISH Mascarello et al 1461

5 2 or more patients. These were unlikely to differ significantly from slides prepared from a single specimen because mixing was performed before fixation. On the other hand, there is no question that the cultured cells used for the prenatal diagnosis simulation were less challenging for FISH study than cells obtained directly from amniotic fluid would have been. The fact that a large proportion of the participating laboratories reported little or no experience with the FISH tests included in the surveys raises the question of how appropriate the surveys were. The CyRC concluded that the alternative of regularly offering a different proficiency test for every diagnosis amenable to FISH testing was impractical. Moreover, the skill that one obtains with one FISH test is largely transferable to other FISH tests. Thus, the committee chose to separate the surveys on the basis of whether they applied to congenital anomalies or to the acquired anomalies of neoplasia and on the basis of whether the survey materials were traditional cytogenetic preparations or sectioned tissue more commonly found in histology laboratories. The validity of this approach is borne out by the fact that greater than 8% participant consensus was reached for the diagnostic conclusion on all challenges. This approach is also supported by the published experiences of a consortium of laboratories comprising the Great Lakes Regional Genetics Group. 6,7 lthough the only methodologic difference between some FISH tests is the DN probe, this difference can have a significant financial consequence. Commercially available probes are relatively expensive reagents that are typically sold in minimum amounts suitable for 1 to 2 reactions. The cost of the probe wasted could easily exceed the cost of the survey subscription if a laboratory does not routinely offer a particular test and, thus, has to buy probe solely for proficiency testing. The accuracy of any FISH test is related to the number of cells scored and the analytic sensitivity of the probes used. ased on the expected sensitivity, the merican College of Medical Genetics has provided guidelines 8 for the number of cells that should be included in FISH studies for microdeletions, enumeration of chromosomes in interphase cells from prenatal samples, and translocation analysis in interphase cells from neoplasms. The modal number of cells studied by participants in challenges corresponding to all 3 of these categories was equal to or higher than these recommendations, but there were some notable exceptions. The number of cells examined by a minority of laboratories was as low as one fifth of the recommendation and as high as times the recommendation. lthough the number of diagnostic errors in the FISH surveys was small, there was no evidence that such errors were a consequence of analyses that included fewer than the recommended number of cells. When reporting cytogenetic test results, most laboratory directors provide a detailed textual description and a description that uses standard nomenclature for human cytogenetics (ISCN ).The conventions included in the standard nomenclature allow a succinct and precise portrayal of the karyotype and any anomalies it might contain. Thus, this nomenclature has been the basis for graded responses in surveys for traditional cytogenetics. 9 The nomenclature was expanded to include FISH in 199. Importantly, the expansion incorporated variation reflecting probe selection, but could not foresee reagents and applications that have been developed over the last 6 years. To determine whether these factors would make ISCN problematic for communicating proficiency test results, the CyRC included an ungraded question on all surveys (except CYH) that asked participants to describe their results using ISCN nomenclature. Not surprisingly, a greater than 8% consensus was not achieved for the nomenclature in any challenge. Moreover, for some challenges, unique responses outnumbered responses given by multiple participants. The results of this exercise and the implications that they have for widespread use of ISCN in FISH will be presented in greater detail elsewhere. However, for purposes of the CP-CMG surveys, it is clear that the current nomenclature system is not well suited to communicating FISH proficiency test results. The CyRC continues to look for ways to expand the repertoire of FISH surveys and to offer challenges that more closely simulate clinical practice. Observations made during traditional cytogenetic studies frequently raise suspicions that are later resolved by FISH testing. s yet, neither the standard CY cytogenetic surveys nor the FISH surveys have been able to satisfactorily capture this aspect of routine clinical cytogenetic service. We have begun exploring participant interest in image-based surveys that could be offered via the Internet. We have also begun banking apheresed cells from patients with hematologic malignancies and lymphoblastoid cell lines from patients with congenital chromosome abnormalities. Our hope is that, in the future, participants might be able to select challenges from a menu of slide-based and electronic media based materials that more closely match their service repertoire. References 1. Dewald GW, rothman R, utler MG, et al. Pilot studies for proficiency testing using fluorescence in situ hybridization with chromosome-specific DN probes: a College of merican Pathologists/merican College of Medical Genetics Program. rch Pathol Lab Med. 1997;121: International Standing Committee on Human Cytogenetic Nomenclature. In: Mittelman F, ed. n International System for Human Cytogenetic Nomenclature. asel, Switzerland: S. Karger G; Watson MS, uchanan PD, Cohen MM, et al. Technical and clinical assessment of fluorescence in situ hybridization: an CMG/SHG position statement. I. Technical considerations. Genet Med. 2;2: Schnitt SJ. reast cancer in the 21st century: neu opportunities and new challenges. Mod Pathol. 21;14: Shahangian S. Proficiency testing in laboratory medicine: uses and limitations. rch Pathol Lab Med. 1998;122: Dewald G, Stallard R, l Saadi, et al. multicenter investigation with interphase fluorescence in situ hybridization using X- and Y-chromosome probes. m J Med Genet. 1998;76: Dewald GW, Stallard R, lsaadi, et al. multi-center investigation with D-FISH CR/L1 probes. Cancer Genet Cytogenet. 2;1: Standards and Guidelines for Clinical Genetics Laboratories. 2nd ed. ethesda, Md: merican College of Medical Genetics; Hoeltge G, Dewald GW, Palmer CG, et al. Proficiency testing in clinical cytogenetics: a 6-year experience with photographs, fixed cells, and fresh blood. rch Pathol Lab Med. 1993;117: rch Pathol Lab Med Vol 126, December 22 Proficiency Testing for Laboratories Performing FISH Mascarello et al