CAP Laboratory Improvement Programs Summary of Microsatellite Instability Test Results From Laboratories Participating in Proficiency Surveys Proficiency Survey Results From 2005 to 2012 Theresa A. Boyle, MD, PhD; Julia A. Bridge, MD; Linda M. Sabatini, PhD; Jan A. Nowak, MD, PhD; Patricia Vasalos, BS; Lawrence J. Jennings, MD, PhD; Kevin C. Halling, MD, PhD; and the College of American Pathologists Molecular Oncology Committee Context. The College of American Pathologists surveys are the largest laboratory peer comparison programs in the world. These programs allow laboratories to regularly evaluate their performance and improve the accuracy of the patient test results they provide. Proficiency testing is offered twice a year to laboratories performing microsatellite instability testing. These surveys are designed to emulate clinical practice, and some surveys have more challenging cases to encourage the refinement of laboratory practices. Objective. This report summarizes the results and trends in microsatellite instability proficiency testing from participating laboratories from the inception of the program in 2005 through 2012. Design. We compiled and analyzed data for 16 surveys of microsatellite instability proficiency testing during 2005 to 2012. Results. The number of laboratories participating in the microsatellite instability survey has more than doubled from 42 to 104 during the 8 years analyzed. An average of 95.4% of the laboratories correctly classified each of the survey test samples from the 2005A through 2012B proficiency challenges. In the 2011B survey, a lower percentage of laboratories (78.4%) correctly classified the specimen, possibly because of overlooking subtle changes of microsatellite instability and/or failing to enrich the tumor content of the specimen to meet the limit of detection of their assay. Conclusions. In general, laboratories performed well in microsatellite instability testing. This testing will continue to be important in screening patients with colorectal and other cancers for Lynch syndrome and guiding the management of patients with sporadic colorectal cancer. (Arch Pathol Lab Med. 2014;138:363 370; doi: 10.5858/ arpa.2013-0159-cp) Although most cases of colorectal cancer (CRC) are sporadic, approximately 2% to 3% are due to the hereditary CRC syndrome known as hereditary nonpolyposis CRC 1 or Lynch syndrome (LS). 1 Patients with LS have been found to have germline mutations in 1 of at least 4 Accepted for publication May 20, 2013. From the Department of Pathology and Division of Medical Oncology, University of Colorado, Aurora (Dr Boyle); the Department of Pathology & Microbiology, University of Nebraska Medical Center, Omaha (Dr Bridge); Roswell Park Cancer Institute, Buffalo, New York (Dr Sabatini); the Department of Pathology and Laboratory Medicine, North Shore University Health System, Evanston, Illinois (Dr Nowak); the College of American Pathologists, Northfield, Illinois (Ms Vasalos); Pathology and Laboratory Medicine, Northwestern University Feinberg School of Medicine, Ann and Robert H. Lurie Children s Hospital of Chicago, Chicago, Illinois (Dr Jennings); and the Department of Laboratory Medicine and Pathology, Mayo Clinic, Rochester, Minnesota (Dr Halling). Dr Sabatini is now with Molecular Diagnostics, North Shore University Health System, Evanston, Illinois. The authors have no relevant financial interest in the products or companies described in this article. Reprints: Theresa A. Boyle, MD, PhD, Department of Pathology and Division of Medical Oncology, University of Colorado Denver, 12801 E 17th Ave, Aurora, CO 80045 (e-mail: Theresa.Boyle@ ucdenver.edu). different DNA mismatch repair (MMR) genes known as MLH1, MSH2, MSH6, and PMS2. The lifetime risk for CRC in individuals with a germline mutation in one of these genes is approximately 80%. 2 The identification of a germline DNA MMR mutation in patients with CRC definitively establishes a diagnosis of LS for that individual and allows for presymptomatic diagnosis of at-risk family members. Individuals with a germline DNA MMR mutation can undergo surveillance for tumors that are part of the LS tumor spectrum including colorectal, endometrial, and gastric cancer. At least 1 study shows that regular colonic and gynecologic surveillance reduces mortality in MMR mutation carriers. 3 Tumors arise in patients with LS because biallelic inactivating mutations (a germline mutation and a somatic mutation) of these DNA MMR genes lead to defective DNA MMR, which then promotes tumorigenesis through mutations of oncogenes and tumor suppressor genes. One of the consequences of defective DNA MMR is a phenomenon known as microsatellite instability (MSI). Microsatellite DNAs are small repetitive sequences (mononucleotides, dinucleotides, and up to 7 repeats) that are scattered throughout the genome. Microsatellite DNAs are susceptible to changes in repeat length during replication but are normally repaired by the DNA MMR system. Defects in Arch Pathol Lab Med Vol 138, March 2014 Proficiency Survey Results From 2005 to 2012 Boyle et al 363
Figure 1. Enrollment for proficiency testing with the College of American Pathologists for microsatellite instability has increased over the years. At least 80% of responses needed to be received prior to compilation of the results. DNA MMR result in MSI, and this serves as a hallmark of defective DNA MMR. Microsatellite instability occurs in approximately 15% of all colorectal tumors but is a consistent feature of colorectal and other tumors in patients with LS. 4 Screening for LS in CRC patients generally begins with testing the tumor for evidence of defective DNA MMR, using MSI testing by polymerase chain reaction and/or DNA MMR protein immunohistochemistry (IHC). Tumors are generally examined for instability with a panel of 5 or more microsatellite DNA markers that compare microsatellite length in tumor cells versus normal cells. Tumors are classified as showing high levels of MSI (MSI-H phenotype) if 2 or more of the 5 markers (or 30% of the markers if more or fewer than 5 markers are used) exhibit instability, a microsatellite-stable (MSS) phenotype if none of the markers show instability, and an MSI-low (MSI-L) phenotype if only 1 of 5 or less than 30% of the markers show instability. 5 In a large study, Cicek et al 6 demonstrated that mononucleotide markers have higher sensitivity for an MSI-H phenotype than dinucleotide markers and confirmed that the appropriate cutoff for considering a case as positive for an MSI-H phenotype is the finding of instability in 30% or more of the markers. The finding of an MSI-H phenotype is consistent with the presence of defective DNA MMR in the tumor. Figure 2. Percentages of laboratories that use less than 5, exactly 5, or more than 5 markers for microsatellite instability testing. As per guidelines, most laboratories use 5 or more markers for microsatellite instability testing. 364 Arch Pathol Lab Med Vol 138, March 2014 Proficiency Survey Results From 2005 to 2012 Boyle et al
Figure 3. Percentages of laboratories that reported a result of microsatellite instability high (MSI-H), microsatellite instability low (MSI-L), microsatellite-stable (MSS), or not provided (NP) for each survey period. The percentages of laboratories that reported the correct result for each survey period are shown below the graph. It is important to note, however, that defective DNA MMR is observed in about 15% to 20% of sporadic colorectal tumors (ie, tumors not associated with LS) and that this does not result from germline DNA MMR mutations but rather is because of somatic hypermethylation of the MLH1 promoter. 7 Consequently, the finding of an MSI-H phenotype in a CRC increases the likelihood that the patient has LS but is not specific for LS. The definitive establishment of a diagnosis of LS requires the finding of a pathogenic germline mutation in one of the DNA MMR genes. Additional testing that can be offered to determine whether a patient with an MSI-H CRC is likely to have LS includes testing the tumor for DNA MMR protein expression using IHC, BRAF V600E point mutation analysis, and MLH1 promoter hypermethylation. 8 A BRAF V600E mutation is present in 60% of sporadic CRCs with defective MMR but is rarely present in CRCs from patients with LS. Consequently, the finding of a BRAF V600E point mutation in an MSI-H tumor that shows loss of MLH1 and PMS2 expression strongly suggests that the tumor is of sporadic origin. 9,10 In addition, MLH1 promoter hypermethylation is observed in sporadic MSI-H CRC with MLH1 loss but not in LS CRC with MLH1 loss. Thus, the absence of MLH1 promoter hypermethylation in an MSI-H tumor that shows loss of MLH1 expression strongly suggests that the tumor is occurring in a patient with LS. The presence of MSI-H tumors with loss of MSH2 and MSH6, MSH6 alone, or PMS2 alone almost always indicates that the patient has LS, and BRAF and MLH1 promoter hypermethylation testing is not indicated in these patients. Microsatellite instability testing has other clinical utilities in addition to identifying individuals with LS. Studies have shown that an MSI-H phenotype is a favorable independent prognostic indicator in CRC patients. 11 In addition, some reports indicate that MSI-H tumors may not be responsive to 5-fluorouracil based therapies. 12 Consequently, MSI testing is sometimes used to inform therapeutic decision making. SUMMARY OF SURVEY RESULTS The number of laboratories enrolled in the College of American Pathologists (CAP) MSI proficiency survey has more than doubled in the past 8 years, from 42 laboratories participating in the original survey, 2005A, to 104 in the most recent survey, 2012B (Figure 1). This increase reflects the increased utility and acceptance of MSI testing in evaluating patients with CRC. Beginning with the 2009A survey, the results were graded once responses were received from at least 80% of the enrolled laboratories. The correct result was defined by the consensus result of the laboratories, and the grading of the laboratories was defined by whether or not their result was the same as the consensus result. The majority of laboratories use 5 markers to assess for MSI, witnessing a growth from 20 of 32 laboratories (63%) reporting the use of 5 markers in the 2005A survey to 75 of 91 (82%) laboratories in the 2012B survey (Figure 2). Of the enrolled laboratories that do not use 5 markers, only 1 or 2 laboratories each year report using fewer than 5 markers, with the remainder using between 6 and 13 markers. Part of the reason most laboratories report the use of 5 markers may be that in 1998 the use of a 5-marker panel was recommended by a National Cancer Institute consensus report on MSI testing. 5 In 2007, Xicola et al 13 reported that a panel of 5 mononucleotide markers demonstrated significantly better sensitivity and specificity for the detection of defective DNA MMR than the original National Cancer Institute recommended panel, which included a combination of mononucleotide and dinucleotide markers. Another possible explanation for the dominant use of 5 markers is the increased use of the commercially available Promega panel of 5 mononucleotide markers (BAT25, BAT26, MONO-27, NR-21, and NR-24) for MSI analysis. This panel includes 2 additional pentanucleotide markers (penta C and penta D) for specimen identification. The use of this panel exhibited a dramatic increase from 9% (3 of 32) of laboratories in 2005A to 65% (55 of 85) of laboratories in 2012B. Arch Pathol Lab Med Vol 138, March 2014 Proficiency Survey Results From 2005 to 2012 Boyle et al 365
Figure 4. These 5 sets of electropherograms show the microsatellite instability (MSI) results from one laboratory that tested the specimen from the 2011B survey period. Each set represents the results of MSI testing from tumor tissue (top) and normal control tissue (bottom). A through D, 4 markers (BAT25 [A], NR-21 [B], BAT26 [C], and MONO27 [D]) showed MSI. E, One marker (NR-24) was stable. Overall, laboratories have performed well on the MSI surveys over the years, with an average correct classification rate of 95.4% from the 2005A through the 2012B surveys (Figure 3). The 2005B survey was complicated by 14 laboratories indicating that they could not provide results because of receiving an inadequate specimen. The 2005B proficiency survey summary explained that the reason the 14 laboratories encountered difficulties may have been an insufficient amount of tissue and that future surveys would aim to provide higher tissue quantity. Also of note, the 2011B survey had a lower correct classification percentage (78.4%) than other surveys. Specifically, of the 88 laboratory result responses in the 2011B survey, 69 correctly identified an MSI-H phenotype, whereas the remaining laboratories incorrectly classified the survey specimen as MSI-L or MSS (5 and 14 laboratories respectively). One possible explanation for misclassification of the case as an MSI-L or MSS phenotype rather than an MSI-H phenotype is failure to recognize the presence of subtle marker instability. The accompanying figure with example MSI electropherograms for the 2011B specimen (Figure 4, A through E) demonstrates clear-cut MSI of BAT25 and NR-21, subtle instability of BAT26 and MONO-27, and no evidence of instability for NR-24. Taken together, these results indicate instability at 4 of the 5 markers of the Promega 5-mononucleotide marker set, consistent with an MSI-H phenotype. Another possible explanation for misclassification of the case as an MSI-L or MSS phenotype by some laboratories is 366 Arch Pathol Lab Med Vol 138, March 2014 Proficiency Survey Results From 2005 to 2012 Boyle et al
Figure 5. The percentage agreement of individual marker results with the consensus result (microsatellite stable or unstable) for 8 commonly used microsatellite instability markers is shown for each survey period. that the tumor content in the sample used for MSI analysis may have been below the limit of detection of the assay used by some of the laboratories with incorrect responses. Interestingly, 6 of the 14 laboratories (43%) that stated they do not use microdissection before analysis misclassified the case as an MSS phenotype. In contrast, among the 75 laboratories that did use microdissection, only 13 (17%) misclassified the tumor as MSS (8 cases) or MSI-L (5 cases). The difference in the frequency of an MSI-H phenotype observed between the laboratories that used microdissection and those that did not was statistically significant (43% versus 17%, P ¼.03; Pearson v 2 ). An examination of the results of each individual microsatellite marker was initiated with the 2007B survey (Figure 5). The results of the individual markers were compared with the consensus result (microsatellite stable or unstable). The agreement between the individual marker result and the consensus result was greater than 80% for the commonly used markers with a few exceptions. The exceptions included agreement of less than 80% with the consensus result for D5S346 in 4 surveys, D17S250 in 3 surveys, and D2S123 in 2 surveys. Of note, these 3 markers (D5S346, D17S250, and D2S123) were all dinucleotide markers. The remaining markers were all mononucleotide Figure 6. Minimum tumor percentage required for microsatellite instability testing by laboratories. Arch Pathol Lab Med Vol 138, March 2014 Proficiency Survey Results From 2005 to 2012 Boyle et al 367
Figure 7. A through F, Estimates by laboratories of tumor percentage within entire section (red bars) and within a lesional area of tissue (blue bars) on hematoxylin and eosin slides for the 2010 2012 surveys (2010A [A], 2010B [B], 2011A [C], 2011B [D], 2012A [E], and 2012B [F]). Note that beginning with the 2011B survey, tumors with percentages within the entire section that were considered greater than 70% were grouped into one category. markers and none had less than 80% agreement with the consensus result for more than one survey. One survey question queried laboratories about the minimum percentage neoplastic cellularity that they require for MSI analysis. Although a rising percentage of laboratories over the years (25% in 2005A to 57% in 2012B) have reported a minimum percentage neoplastic cellularity requirement of between 11% and 40%, there is no clear consensus to date on what is required (Figure 6). The lack of consensus may be due to the neoplastic cellularity requirement being dependent on the availability of tumor enrichment in the laboratory and on the analytic sensitivity of the test method being performed. There also may be different perspectives about how to balance the risks of providing a false-negative result because of low neoplastic cellularity versus the risks of not analyzing a specimen that might have provided an adequate result. However, a study of serial dilutions of microdissected tumor cells from a specimen with known instability of the BAT25 allele revealed that the smallest concentration at which instability could be detected was approximately 10%. 14 Therefore, in a specimen with tumor cells heterozygous for an unstable allele, it may not be possible to detect MSI if the tumor cellularity is less than 20%. A different study comparing MSI testing of biopsies from 2 different areas of individual surgical resection specimens from 67 patients with colorectal carcinoma revealed concordant results in all paired biopsy samples (all but one biopsy contained at least 25% carcinoma). 15 Although criteria for specimen adequacy might account for the fact that small biopsies or suboptimal tissue may be the only available material, the risk for false negatives with testing of tissues with suboptimal adequacy should be included in the MSI report of results. Microsatellite instability surveys since the 2010A survey period have asked laboratories to estimate both the percentage of neoplastic cells within the lesional area and the percentage of neoplastic cells for the entire section of tissue on a hematoxylin and eosin stained glass slide. As expected, each year the average estimates of percentage neoplastic cellularity within the lesional area was higher than the average estimates of percentage neoplastic cellularity for the entire section on the slide (Figure 7, A through F). Overall, the results for the surveys showed a normal distribution. However, the range of estimates for percentage of neoplastic cells was wide for both the lesional area and the entire section. These results highlight the high interobserver variability for estimation of percentage neoplastic cellularity. In addition, the CAP sent 10 images of colon tissue specimens at 320 magnification as a part of the KRAS 2011B survey and reported the results of estimations of neoplastic cellularity by 194 laboratories. 16 In 5 of the 10 cases, more than 10% of laboratories overestimated the tumor cellularity. Overestimation was defined as the laboratory estimate of percentage tumor cellularity being 20% higher than the neoplastic cellularity determined by exact counting of each neoplastic and nonneoplastic cell. Overestimation of tumor cellularity is of concern because it could lead to false-negative results. There is a need for more research and education about percentage neoplastic cellularity assessment, especially because it is being used to determine specimen adequacy for molecular testing. The number of laboratories that use some form of tumor enrichment protocol (such as microdissection, where areas of higher tumor cellularity are microscopically identified and delineated) has gradually increased over the years from 78% (25 of 32) of laboratories in the 2005A survey to 88% (81 of 92) of laboratories in the 2012B survey (Figure 8). A small number of laboratories (up to 3 each survey period) have used a method other than manual microdissection, such as laser capture, to enhance neoplastic cellularity. Notably, 368 Arch Pathol Lab Med Vol 138, March 2014 Proficiency Survey Results From 2005 to 2012 Boyle et al
Figure 8. Percentages of laboratories using manual microdissection, laser capture microdissection, or no microdissection by survey period. The use of microdissection to enhance tumor cellularity has increased over the years. more than 10% of laboratories in each survey period did not use neoplastic cell enrichment approaches. These labs need to be certain that the technique they use to assess for MSI has the analytic sensitivity required to detect instability in all cases examined or to reject cases that fall below that limit of detection. Since the first MSI survey, 2005A, capillary electrophoresis was documented as the most popular technical approach used for MSI testing. However, in each survey period, 2 to 6 laboratories reported performing polyacrylamide gel electrophoresis or another method for MSI testing. Of interest, the use of laboratory-developed tests based on commercially available packages has increased from 12.5% (4 of 32) to 67.1% (57 of 85) of laboratories between the 2005A and 2012B survey periods (Figure 9). Laboratories were asked each year if they recommended MSI testing as a screening tool for LS, a prognostic indicator, and/or for use in treatment decisions. Virtually all of the laboratories reported screening for LS as an indication for MSI testing, about 40% have always considered prognosis as an indication, and an increasing fraction of laboratories included therapy as an indication (Figure 10). However, the laboratories that did not cite prognosis and/or therapeutic decisions as specific indica- Figure 9. Percentages of laboratories using commercially available packages for their laboratory-developed tests (LDT) versus those using non commercially available packages for microsatellite instability testing by survey period. The use of commercially available packages has increased over the years. Arch Pathol Lab Med Vol 138, March 2014 Proficiency Survey Results From 2005 to 2012 Boyle et al 369
Figure 10. Clinical indications for microsatellite instability (MSI) testing by survey period. Laboratories most commonly reported screening for Lynch syndrome as an indication for MSI testing, but many also reported guiding chemotherapy and assessing prognosis as indications for testing. tions for MSI testing may still recognize the significance of MSI testing for prognosis or therapy. COMMENT The performance of laboratories in each period relates to the nature of the specimen distributed but is also reflective of a general improvement of testing accuracy of the laboratories over the years. The importance of enhancement of tumor cellularity was highlighted by the finding in the 2011B survey that laboratories that microdissected specimens to enhance tumor cellularity were more likely to correctly determine the abnormalities present. The CAP also began offering a proficiency survey for MMR IHC in 2010. In the 2012 surveys, the same tissue was used in the MMR IHC survey as in the MSI polymerase chain reaction survey, such that the results for the 2 different surveys could be compared. In the 2012A survey, 98% of laboratories reported the MSI result as MSI-H and 91% of laboratories reported loss of immunostaining for both MLH1 and PMS2. In the 2012B survey, 100% of laboratories reported the MSI result as MSI-H, 95% of laboratories reported loss of MSH2, and 97% of laboratories reported loss of MSH6. The CAP intends to continue comparing the results of MSI testing with IHC testing and has changed the name of the MSI survey to the MMR survey to facilitate this effort. These results demonstrate an excellent performance of laboratories participating in the MSI proficiency surveys. This good performance of laboratories over the years is likely due, at least partly, to the educational nature of the CAP proficiency testing, which provides laboratories with an external mechanism to monitor the quality status of their testing. 17,18 References 1. Hampel H, Frankel WL, Martin E, et al. Screening for the Lynch syndrome (hereditary nonpolyposis colorectal cancer). N Engl J Med. 2005;352(18):1851 1860. 2. Vasen HF, Wijnen JT, Menko FH, et al. Cancer risk in families with hereditary nonpolyposis colorectal cancer diagnosed by mutation analysis. Gastroenterology. 1996;110(4):1020 1027. 3. Jarvinen HJ, Renkonen-Sinisalo L, Aktan-Collan K, Peltomäki P, Aaltonen LA, Mecklin JP. Ten years after mutation testing for Lynch syndrome: cancer incidence and outcome in mutation-positive and mutation-negative family members. J Clin Oncol. 2009;27(28):4793 4797. 4. Geiersbach KB, Samowitz WS. Microsatellite instability and colorectal cancer. Arch Pathol Lab Med. 2011;135(10):1269 1277. 5. Boland CR, Thibodeau SN, Hamilton SR, et al. A National Cancer Institute workshop on microsatellite instability for cancer detection and familial predisposition: development of international criteria for the determination of microsatellite instability in colorectal cancer. Cancer Res. 1998;58(22):5248 5247. 6. Cicek MS, Lindor NM, Gallinger S, et al. Quality assessment and correlation of microsatellite instability and immunohistochemical markers among population- and clinic-based colorectal tumors results from the Colon Cancer Family Registry. J Mol Diagn. 2011;13(3):271 281. 7. Herman JG, Umar A, Polyak K, et al. Incidence and functional consequences of hmlh1 promoter hypermethylation in colorectal carcinoma. Proc Natl Acad Sci USA. 1998;95(12):6870 6875. 8. Funkhouser WK Jr, Lubin IM, Monzon FA, et al. Relevance, pathogenesis, and testing algorithm for mismatch repair-defective colorectal carcinomas: a report of the Association for Molecular Pathology. J Mol Diagn. 2012;14(2):91 103. 9. Parsons MT, Buchanan DD, Thompson B, Young JP, Spurdle AB. Correlation of tumour BRAF mutations and MLH1 methylation with germline mismatch repair (MMR) gene mutation status: a literature review assessing utility of tumour features for MMR variant classification. J Med Genet. 2012;49(3):151 157. 10. Domingo E, Laiho P, Ollikainen M, et al. BRAF screening as a low-cost effective strategy for simplifying HNPCC genetic testing. J Med Genet. 2004; 41(9):664 668. 11. Popat S, Hubner R, Houlston RS. Systematic review of microsatellite instability and colorectal cancer prognosis. J Clin Oncol. 2005;23(3):609 618. 12. Sargent DJ, Marsoni S, Monges G, et al. Defective mismatch repair as a predictive marker for lack of efficacy of fluorouracil-based adjuvant therapy in colon cancer. J Clin Oncol. 2010;28(20):3219 3226. 13. Xicola RM, Llor X, Pons E, et al. Performance of different microsatellite marker panels for detection of mismatch repair-deficient colorectal tumors. J Natl Cancer Inst. 2007;99(3):244 252. 14. Trusky CL, Sepulveda AR, Hunt JL. Assessment of microsatellite instability in very small microdissected samples and in tumor samples that are contaminated with normal DNA. Diagn Mol Pathol. 2006;15(2):63 69. 15. Zauber NP, Sabbath-Solitare M, Marotta S, Perera LP, Bishop DT. Adequacy of colonoscopic biopsy specimens for molecular analysis: a comparative study with colectomy tissue. Diagn Mol Pathol. 2006;15(3):162 168. 16. Viray H, Li K, Long T, et al. A prospective, multi-institutional diagnostic trial to determine pathologist accuracy in estimation of percentage of malignant cells. Arch Pathol Lab Med. 2013;137(11):1545 1549. 17. Hammerling JA. A review of medical errors in laboratory diagnostics and where we are today. Lab Med. 2012;43(2):41 44. 18. Sciacovelli LS, Secchiero S, Zardo L, Plebani M. The role of external quality assessment. Biochem Med. 2010;20(2):160 164. 370 Arch Pathol Lab Med Vol 138, March 2014 Proficiency Survey Results From 2005 to 2012 Boyle et al