Cancer Genetics 207 (2014) 316e325

Population-based characterization of the genetic landscape of chronic lymphocytic leukemia patients referred for cytogenetic testing in British Columbia, Canada: the role of provincial laboratory standardization Alina S. Gerrie a,1, Steven J.T. Huang a,b,1, Helene Bruyere b, Chinmay Dalal a, Monica Hrynchak c, Aly Karsan d, Khaled M. Ramadan e, Adam C. Smith d,f, Christine Tyson c, Cynthia L. Toze a,1, Tanya L. Gillan b,*,1 a

Leukemia/BMT Program of BC, Vancouver General Hospital and British Columbia Cancer Agency, University of British Columbia, Vancouver, Canada; b Pathology and Laboratory Medicine, Vancouver General Hospital, University of British Columbia, Vancouver, Canada; c Cytogenetics Laboratory, Royal Columbian Hospital, New Westminster, Canada; d Cancer Genetics Laboratory, Pathology and Laboratory Medicine, British Columbia Cancer Agency, University of British Columbia, Vancouver, Canada; e Division of Hematology, St. Paul’s Hospital, University of British Columbia, Vancouver, Canada; f Instituto de Pesquisa Pele
Pequeno Princı´pe, Curitiba, Brazil Detection of recurrent chromosome abnormalities by fluorescence in situ hybridization (FISH) is an essential component of care in chronic lymphocytic leukemia (CLL) patients. In the province of British Columbia (BC), Canada, population 4.6 million, CLL patients receive uniform evaluation and therapy with FISH testing performed in three jurisdictions. The aims of this study were to (i) validate CLL-FISH testing among the BC cytogenetic laboratories to ensure standardization of results and (ii) characterize population-level CLL-FISH abnormalities by pooling provincial data. From 2004 to 2011, 585 consecutive patients underwent pretreatment CLL-FISH testing at laboratory A (50.1%), laboratory B (32.3%), or laboratory C (17.6%). For validation purposes, 26 CLL-FISH abnormalities were tested by each laboratory’s protocol, with 91% result concordance. Discordant results involved percent abnormalities at or near cutoff values; therefore, a 10% universal cutoff was established when pooling results. Applying the universal cutoff to the provincial cohort, CLL-FISH abnormalities were detected in 74.9%: 54.9% 13q-, 18.8% þ12, 8.5% 11q-, and 7.7% 17p-. In this large population-based cohort of patients referred for CLL-FISH testing, frequencies of abnormalities detected by FISH analysis were highly consistent with those reported in single-institution and clinical trial populations. Provinces or districts that work together to care for CLL patients can effectively pool data with appropriate laboratory validation to ensure standardization of results. Keywords Chronic lymphocytic leukemia, cytogenetic, fluorescence in situ hybridization, validation, population-based ª 2014 Elsevier Inc. All rights reserved.

Presented in abstract form at the XIV International Workshop on CLL, Houston, TX, October 28, 2011. Received February 15, 2014; received in revised form August 14, 2014; accepted August 21, 2014. * Corresponding author. E-mail address: [email protected] 1 ASG and SJH, and CLT and TLG contributed equally to this article. 2210-7762/$ - see front matter ª 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.cancergen.2014.08.006

Chronic lymphocytic leukemia (CLL) remains the most common leukemia in North America and Europe, largely affecting individuals over 60 years of age (1,2). Small lymphocytic lymphoma (SLL) shares similar clinical and pathological characteristics with CLL and is approached and managed in the same manner as CLL. Together, they are classified as “CLL/SLL” by the World Health Organization 2008 (3). Population-based analyses of the incidence of CLL

Provincial CLL FISH standardization and SLL include a United States Surveillance, Epidemiology and End Results (SEER) report that found if the incidence of SLL were added to that of CLL, the combined incidence of CLL/SLL by tumor registry data was 5.13 per 100,000 person years 1993e2004 (4). A Canadian provincial report from Manitoba, Canada, combined data from a centralized flow cytometry facility and the provincial tumor registry and found an age-adjusted incidence of 7.52 per 100,000 persons (5). The clinical course of CLL/SLL is highly variable, as survival can range from months to more than 20 years from the time of diagnosis (6). Traditional prognostic markers, such as age, lymphocyte doubling time, and CD38 positivity, fail to accurately predict outcome, particularly in the early stages of disease, when most patients are diagnosed (7,8). Recurrent CLL-specific chromosomal abnormalities detected by fluorescence in situ hybridization (FISH), found in over 80% of individuals with CLL/SLL, are well-described prognostic markers that can better predict overall survival (OS) and the time to first treatment (7). Deletions of 17p13 involving the TP53 locus (17p-) and at 11q22 involving the ATM locus (11q-) have been demonstrated to confer an unfavorable prognosis (7,9,10); trisomy 12 (þ12) confers an intermediate prognosis (7,9,10); and deletion of 13q14 (13q-) predicts a more favorable prognosis (7,11). Lack of all of these abnormalities (i.e., a normal FISH result) confers an intermediate prognosis (7). Translocations involving the immunoglobulin heavy chain (IGH ) gene locus [t(IGH )] at chromosome 14q32 may carry an unfavorable prognosis (12,13), whereas deletion of 6q21 involving the MYB locus (6q-) has been reported to carry an intermediate risk (14e16). Detection of these recurrent chromosomal abnormalities by FISH analysis is an essential component of the clinical care of CLL/SLL patients, guiding both prognostication and risk-adapted management strategies (17e19). In the province of British Columbia (BC), Canada, population 4.6 million (20), CLL patients receive uniform evaluation and therapy according to centrally derived standard protocols developed by the BC Cancer Agency (BCCA) (21). We have previously reported on outcomes in both the upfront and relapsed/refractory setting (22,23). Between 2004 and 2011, FISH testing in BC was performed at one of three cytogenetic laboratories, prior to therapy, to guide treatment decisions. National guidelines to validate and interpret FISH assays in clinical practice are based on the American College of Medical Genetics (ACMG) guidelines, which are endorsed by the Canadian College of Medical Geneticists (CCMG) (24); however, certain aspects of FISH testing are left to the laboratory directors’ discretion, including FISH probe design, scoring criteria, and statistical methods used to calculate normal cutoff values. The CLL Research Consortium (CRC) has highlighted the importance of FISH standardization within cooperative groups to ensure validity of clinical correlative studies that pool results from multiple laboratories (25). Prior studies have evaluated the prevalence of recurrent cytogenetic abnormalities in large, nonuniform groups of CLL patients (7,8,26e28). To our knowledge, no study has examined recurrent cytogenetic abnormalities in a large, population-based cohort of uniformly managed and treated CLL patients referred for FISH testing, in part because of variation in therapy among institutions and variation in FISH testing among multiple laboratories. Therefore, the

317 objectives of this present study were to (i) validate CLL FISH testing among the three BC cytogenetic laboratories to ensure standardization of results when pooling data and (ii) characterize CLL FISH abnormalities at a population-level by analyzing their prevalence across the entire province of BC.

Materials and methods FISH validation Provincial validation of FISH was performed first by conducting interviews with each laboratory to evaluate methods including protocols, probe design, probe vendor, number of cells analyzed, number of scorers, and establishment and calculation of normal cutoff values. All of the protocols, cutoffs, and additional details were compiled and reviewed. Second, each laboratory selected specimens for evaluation by the other laboratories, with the conditions that the specimens should have more than one FISH abnormality and at least one specimen from each laboratory should include an abnormality that was at or near the cutoff value. Specimens were coded for blinded studies and sent to the other laboratories for FISH processing and analysis as per the evaluating laboratory’s clinical standard operating procedures. Each laboratory tested for 13q-, þ12, 11q-, 17p-, and IGH rearrangements (Labs B and C also tested for 6q-). Deletions in 13q14.3 were identified as either monoallelic (13q- 1) or biallelic (13q-2) loss of the 13q14.3 locus, or a combination of both.

Patients Results of the first FISH test from each untreated CLL patient referred to one of three cytogenetic laboratories in BC (Labs A, B, and C) from January 2004 to December 2011 were collected and entered into the BC Provincial CLL Database. A diagnosis of CLL was confirmed according to consensus guidelines (18) by an experienced hematopathologist after review of peripheral blood (PB) morphology and PB/bone marrow (BM) immunophenotyping. The three cytogenetic laboratories each receive distinct patient populations: Lab A receives samples from throughout the province on generally treatment-naive patients; Lab B is the referral laboratory for the Leukemia/Bone Marrow Transplant (BMT) Program of BC, treating the highest risk patients and those with advanced or aggressive disease (23); and Lab C receives generally treatment-naive patients from community-based internists, hematologists, and oncologists. Interphase FISH was performed on unselected PB or BM-aspirate specimens at one of the three laboratories according to standard operating procedures and manufacturers’ instructions. Patients positive for the CCND1/IGH translocation, which is typically diagnostic of mantle cell lymphoma, were excluded from the study. All BC CLL patients receive uniform evaluation and therapy based on centrally derived BCCA era-specific protocols (21,22). In this cohort, among 403 treated patients, first-line therapy was fludarabine (F)ebased for the majority (66%) with rituximab (R) or cyclophosphamide (C) (FR 162 patients, purine analogue alone 85, FCR 19, FC 1), 27% received alkylator-based therapy (chlorambucil 68 patients,

318 CVP-R 32, CVP 6, C alone 1, CHOP 1), and the remaining received other regimens (steroids alone 14, R alone 8, interferon 1, radiation 1, unknown 1). Variations in first-line therapy reflect the era of treatment; protocol-specified alkylator therapy in the setting of active, uncontrolled autoimmune complications; and patient comorbidities. Clinical and laboratory data were collected from institutional databases, with supplemental information obtained from hospital, clinical, and individual physicians’ records. This study was approved by the University of BC, BCCA, and Fraser Health clinical research ethics boards.

Statistical analysis Baseline clinical and laboratory characteristics were compared among laboratories using the Fisher exact test for categorical variables and the KruskaleWallis test for continuous variables (nonparametric testing was used for baseline characteristics, including reporting of medians instead of means, due to the nonnormality of the data). Populationbased cytogenetic abnormalities were tabulated using descriptive statistics. A comparison of prevalence of abnormalities reported in the literature to the BC cohort was made using the two-sample binomial test of proportions. Data were analyzed using STATA version 12.1 (College Station, TX). A type I error rate of 0.05 was assumed for all statistical tests.

Results Provincial CLL FISH validation Each laboratory processes approximately 100e200 CLL FISH specimens per year and follows the CCMG-endorsed ACMG standards and guidelines for FISH analyses (24). All laboratories were accredited through the national Diagnostic Accreditation Program for clinical laboratory services and participated in regular proficiency testing through surveys from the College of American Pathologists and/or the Canadian Quality Management ProgramdLaboratory Services (QMPLS). Each laboratory used copy number probes hybridizing to the ATM gene locus located at chromosome 11q22.3, the D13S319-D13S25 locus at 13q14.3, and the TP53 gene locus at 17p13.1 and a chromosome 12 centromeric probe (Table 1). Labs B and C also included a copy number probe for assessment of the MYB gene locus at 6q23.3 and a dual-color break-apart IGH probe for the detection of IGH rearrangements. If a positive result was seen with the IGH break-apart probe, reflex testing was performed for the t(11;14)(q13;q32) involving CCND1 and IGH and the t(14;18)(q32;q21) involving IGH and BCL2. Although Lab A does not routinely test for the t(11;14) with its FISH panel, it does so on a case-by-case basis if mantle cell lymphoma has not been previously ruled out (e.g., by immunohistochemical staining for cyclin D1 on diagnostic bone marrow or lymph node samples). Lab A used the commercially available CLL FISH Probe Kit (Abbott Laboratories, Abbott Park, IL), whereas Labs B and C used a commercially available CLL FISH panel (Cytocell, Cambridge, UK). Normal cutoff values were determined for each probe independently by each laboratory as per

A.S. Gerrie et al. recommended guidelines and discretion of the laboratory directors (24) and are listed in Table 2. Labs A and C determined cutoffs by calculating the mean plus three standard deviations, whereas Lab B used the mean plus two standard deviations. Each laboratory evaluated 26 clinically reported FISH abnormalities among nine specimens. Of the 78 possible observations (26 clinically reported abnormalities  3 evaluations for the three labs), 71 abnormalities (91%) were correctly identified, as shown in Table 3. Four potential false positive results (5.1%) and three potential false negative results (3.8%) were noted. The percentage of abnormal cells in all potential false positive and false negative observations were at or near the laboratory cutoff values (Table 3). In one case, sample 7, Lab C correctly identified 13q-1; however, the percent of abnormal cells was low at 10.5% (compared with 30% and 40% for Labs A and B, respectively). Lab C did not identify any 13q-2 abnormalities. On repeat testing, Lab C demonstrated 17.5% 13q-1 and 12.5% 13q-2, and the difference was thought to be due to differences in probe intensity.

Patient baseline characteristics There were 585 consecutive patients who underwent pretreatment CLL FISH testing at one of the three BC cytogenetic laboratories from 2004 to 2011. Patient characteristics and comparisons of laboratories are shown in Table 4. The median age at the time of first FISH test was 63 years (range 25e97 y) and 66% of patients were male. The median time from diagnosis to first FISH test for the whole cohort was 0.5 years (range 0-21.8 y). Significant differences in characteristics among laboratories are shown in Table 4 and include Rai stage at diagnosis (more Rai stage 3e4 in Lab B, P Z 0.033), time from diagnosis to FISH test (longer time interval at Lab B, P Z 0.006), and specimen type tested (fewer BM specimens at Lab C, P Z 0.001). Most patients with an abnormal FISH result had one FISH abnormality (57%), with fewer patients having two or three or more abnormalities (19% and 2%, respectively). There were no significant differences in prevalence of FISH abnormalities among laboratories (using each laboratory’s specific cutoff value) except for 17p-, which was more frequently detected in Lab A than in Labs B and C (14.0% vs. 6.3% and 4.9% respectively, P Z 0.004) and the t(IGH ), which was more frequently detected in Lab B than in Lab C (13.4% vs. 1.9%, P < 0.001). For the entire cohort, the prevalence of recurrent cytogenetic abnormalities and classification by hierarchy are shown numerically in Table 5. Using each laboratory’s specific cutoff value, the overall prevalence of FISH abnormalities were 9.9% 17p-, 8.9% 11q-, 19.7% þ12, 56.6% 13q- (43.8% 13q1, 4.4% 13q-2, 8.4% both), with 23.2% having none of these four abnormalities. In addition, of 281 patients tested, 1.4% had 6q-, and of 290 patients tested, 9.3% had a t(IGH ).

Population-level CLL FISH abnormalities Based on our validation findings regarding false results at or near cutoff values, we implemented a conservative shared universal cutoff of 10% (abnormal results >10%), with the goal of eliminating clinically significant false positive results.

Provincial CLL FISH standardization Table 1

319

BC cytogenetic laboratory FISH probes Lab A

FISH panel FISH, abnormality 13q-

Labs B and C

Vysis CLL FISH Probe Kit

þ12

D13S319 (13q14.3) 13q34 CEP 12 (12 centromere)

11q-

LSI ATM (11q22.3)

17p-

TP53 (17p13.1)

6q-

Not tested

t(IGH ) t(14;18)

Not tested Not tested

t(11;14)c

Not routinely tested

a

Cytocell Panelb D13S319-D13S25 (13q14.3) D13S1825 (13qter) CEP 12 D12Z3 (12 centromere) ATM (11q22.3) D11Z1 (11 centromere) TP53 (17p13.1) D17Z1 (17 centromere) MYB (6q23.3) D6Z1 (6 centromere) IGH (14q32.33) IGH (14q32.33) BCL2 (18q21.3) IGH (14q32.33) CCND1 (11q13)

Abbreviation: CEP, centromere enumeration probe. a Abbott Laboratories, Abbott Park, IL. b Cambridge, UK. c Lab A does not routinely test for the t(11;14); however, it does so on a case-by-case basis if mantle cell lymphoma has not been ruled out previously (e.g., by immunohistochemical staining for cyclin D1 on diagnostic BM or lymph node samples).

We chose 10% as a cutoff since most false results were seen when abnormalities were detected in 10% or fewer cells, and based on the CLL Research Consortium FISH Standardization project, which also recommends a cutoff of 10% (25). The universal cutoff was applied to all probes except the dual-color dual fusion probes (for the t(11;14) and t(14;18)) and the D13S319 probe (for biallelic loss of 13q14), for which the cutoff remained at 1%. The prevalence of FISH

Table 2

abnormalities in our cohort was recalculated with the 10% universal cutoff value, and results between the individual laboratory cutoff and 10% or less were labeled as “equivocal” so as not to lose the information. There were 48 patients in our cohort who had an equivocal result. Slightly lower frequencies of FISH abnormalities were observed with the universal 10% cutoff than when individual laboratory cutoffs were used (Table 5). Using the universal cutoff, the

FISH scoring criteria and normal cutoff values for each BC laboratory Lab Aa

Lab Bb

Lab Cc

Number of nuclei scored Number of scorers Calculation of normal cutoff values

200 2 Mean þ 3 SD

100b 1 Mean þ 2 SD

200c 2 Mean þ 3 SD

13q- cutoff (%)

1R (7)

þ12 cutoff (%)

(6)

1R/1G (5) 1R/2G (7) (6)

1R/1G 1R/2G (10)

11q- cutoff (%)

1G (6)

17p- cutoff (%)

1R (6)

t(IGH ) cutoff (%) 6q- cutoff (%)

Not tested Not tested

1R/1G 1R/2G 1R/1G 1R/2G (11) 1R/1G 1R/2G

1R/1G 1R/2G 1R/1G 1R/2G (13) (12)

(6) (6.5) (7) (6.5) (4.5) (4)

(11)

(10) (9)

Abbreviations: R, red probe signal; G, green probe signal. a Lab A enumerates signals using single-color filters for the presence or absence of signals. Two scorers score 100 nuclei each. A third person scores an extra 100 nuclei if the interobserver delta is greater than 5% or if the result is within 0.5% of cutoff, and the results average is reported. The laboratory director may also indicate discretionary additional scoring if the interpretation of the result is unclear. b Lab B: For results at or near a cutoff, an extra 100 nuclei are scored and the two scorings are averaged to obtain a result. c Lab C cutoffs combine 1R/1G and 1R/2G signal patterns together. If result is normal, two people score 100 nuclei each. If result is abnormal, two people score 50 nuclei each. The results average is reported.

320 Table 3

A.S. Gerrie et al. Provincial laboratory validation results

Sample

FISH abn

Lab A % abn

Lab B % abn

Lab C % abn

1

þ12 t(IGH )

64 84.0a

59 70

2

13q-1 11q17pIGH1 13q-1 þ12

59 77 68 36 6.5 51

54 87 77 36 10

4

þ12

5

13q-1 11q-

3

Sample

FISH abn

74 68

6

7

39

67 87 80 27 9 60

67

59

71

9

13q13qþ12 11q17p13q13qþ12 17p13q11q17p13q11q17p-

89 88

86 80

85 82

8

x1 x2

x1 x2

Lab A % abn

Lab B % abn

Lab C % abn

31 48 56 20 5.5 30 15 50 10

33 44 74 23 8

34 43 71 30 6.5 10.5 0.0b

63 64 11 x1

41 74 17

40 5 42 3.0/2.0c 67 61 1.0/3.0c 32 76 7.0/4.0c

42 8 60 64 3.5 39 74 18.5

False positive result False negative result. Abbreviations: abn, abnormality; IGH1, deletion of one copy of the immunoglobulin heavy chain gene. a Lab A used Vysis t(IGH) probe (Abbott Laboratories, Abbott Park, IL) to test for a t(IGH ) in this study. b Repeat testing by Lab C demonstrated 1R2G 15%, 1R1G 2.5%, 2R2G (1 R signal small) 12.5% and the difference was thought to be due to differences in probe intensity. c Indicates 1R/1G and 1R/2G signal patterns, respectively.

prevalence of FISH abnormalities from each laboratory and €hner for the whole cohort was similar to that reported by Do et al. in 2000 (7) (Figure 1), except for a lower prevalence of €hner, P < 0.001) and 6q- (1.1% 11q- (8.5% BC vs. 18% Do €hner, P Z 0.002), and a higher prevalence of a BC vs. 6% Do €hner, P Z 0.008). When t(IGH ) (9.3% BC vs. 4% Do compared with a more recent clinical trial population of untreated CLL patients starting first-line therapy (29), our population had a similar prevalence of deletions 13q and 17p (P Z 0.53 and P Z 0.74, respectively); however, we found a higher incidence of trisomy 12 (18.8 % vs. 12.0%, P Z 0.001) and again a lower prevalence of deletion 11q (8.5% vs. 24.6%, P < 0.001) (data not shown).

Discussion We have demonstrated for the first time in a populationbased cohort of uniformly managed CLL patients referred for FISH testing that the distribution of CLL FISH abnormalities is consistent with that originally described by €hner et al. in 2000 (7). Of the 585 patients, a recurrent Do chromosome abnormality was detected by FISH in 74.9% using the universal cutoff, with the most common abnormality being 13q- (in 54.9% of patients), followed by þ12 (in 18.8%), 11q- (in 8.5%), and 17p- (in 7.7%). We chose to €hner compare our FISH frequencies to those reported by Do et al. as theirs was the seminal paper that evaluated these four recurrent FISH detectable chromosome abnormalities €hner’s cohort in CLL. We recognize, however, that Do included a mixture of pre- and posttreatment samples, and, therefore, we also compared our prevalence rates with a more recent clinical trial cohort about to start first-line therapy. As seen in our cohort, deletion of 11q (27,28,30)

and deletion of 6q (27,28) have been consistently reported at lower frequencies and, conversely, IGH translocations (12,30e32) and trisomy 12 (10,27) have been reported to occur at higher frequencies. These differences likely reflect variations in the populations studied, including whether they were clinical trial or population-based cohorts. A major challenge in genetic epidemiologic studies of CLL is the fact that large, population-based cohorts of patients have significant variability, including patient characteristics and genetic testing across different laboratories, for which it is difficult to control. Moreover, many important confounders are difficult to accurately ascertain, including prior therapies, socioeconomic status, and access to health care. Clinical trial populations may be able to record this type of information; however, these patients are highly selected and are generally not representative of the broad spectrum of the entire CLL population. We are therefore in a unique position in BC to analyze the genetic landscape of CLL on a population level. A publicly funded health care system ensures health care access for all CLL/SLL patients in BC to a network of oncologists and hematologists. BCCA cancer centers and community hematologists and oncologists are geographically dispersed throughout the province, thus decreasing bias related to geographic location. All BCCAapproved cancer therapies are available at no cost to patients through the provincial government, further decreasing bias related to income and socioeconomic status. In addition, all FISH testing is done at one of three laboratories in BC, with testing funded for all CLL patients through the provincial health care program. To pool our data effectively, it was crucial to have standardization among laboratories to minimize inconsistent results. The CLL Research Consortium (CRC) conducted a study to standardize FISH testing in CLL among the five CRC

Provincial CLL FISH standardization

321

Table 4 Baseline clinical and laboratory characteristics for BC CLL patients, with comparisons by laboratory and for the whole cohort (n Z 585) Characteristic Patient characteristic Male, no. (%) Median age at diagnosis, y (range) Rai stage at diagnosis, no. (%) Low risk (0) Intermediate risk (1e2) High risk (3-4) Complete blood count at diagnosis, median (range) Hemoglobin, g/L White blood cell count, 109 cells/L Lymphocyte count, 109 cells/L Platelet count, 109 cells/L CD38 positive,c no. (%) FISH test characteristic Median time from diagnosis to FISH, years (range) Median age at cytogenetic test, years (range) Specimen type, no. (%) PB BM Number of FISH abnormalities per specimen, no. (%) 0 1 2 3 Prevalence of recurrent FISH abnormalities, no. (%) Deletion 17p Deletion 11q Trisomy 12 Deletion 13q Monoallelic Biallelic Both monoalllelic and biallelic Deletion 6q t(IGH ) None of the above abnormalitiesd

Lab A (n Z 293) Lab B (n Z 189) Lab C (n Z 103) Whole cohort (n Z 585) No.a P valueb 200 (68) 61 (25e93)

161 (59) 94 (35) 17 (6)

112 (59) 60 (34e96)

87 (53) 61 (37) 17 (10)

72 (70) 59 (32e81)

384 (66) 61 (25e96)

585 585

0.08 0.87

57 (56) 43 (43) 1 (1)

305 (57) 198 (37) 35 (7)

538 538 538

0.033

142 (56e187) 19 (5e280)

139 (57e171) 16 (1e274)

141 (98e173) 18 (4e142)

140 (56e187) 18 (1e280)

508 517

0.36 0.60

13 (2e275)

10 (0e204)

12 (2e140)

12 (0e275)

510

0.84

201 (16e419) 70 (30)

201 (5e432) 44 (31)

192 (115e419) 28 (32)

200 (5e432) 142 (30)

507 469

0.96 0.92

0.3 (0e21)

1.3 (0e22)

0.7 (0e17)

0.5 (0e22)

585

0.006*

63 (25e93)

63 (34e97)

63 (33e88)

63 (25e97)

585

0.53

218 (74) 75 (26)

152 (80) 37 (20)

95 (92) 8 (8)

465 (80) 120 (22)

585 585

0.001*

61 167 59 6

(21) (57) (20) (2)

39 107 40 3

(21) (57) (21) (2)

28 61 13 1

(27) (59) (13) (1)

128 335 112 10

(22) (57) (19) (2)

585 585 585 585

0.53

41 29 51 180 134 15 31

(14.0) (9.9) (17.4) (61.4) (45.7) (5.1) (10.6)

(6.3) (6.9) (23.8) (51.9) (41.3) (1.6) (9.0)

5 10 19 53 44 8 1

(4.9) (9.7) (18.4) (51.5) (42.7) (7.8) (0.97)

58 52 115 331 256 26 49

(9.9) (8.9) (19.7) (56.6) (43.8) (4.4) (8.4)

585 585 585 585 585 585 585

0.004* 0.50 0.21 0.060 0.61 0.036* 0.010*

4 (1.4) 27 (9.3) 128 (21.8)

281 290 585

1 11q- þ/- other abnormalities except According to a modified FISH hierarchy proposed by Do 17p- > þ12 þ/- other abnormalities except 17p- or 11q- > 13q- without 17p-, 11q- or þ12 > none of 17p-, 11q-, þ12, or 13q-. Since not all laboratories tested for a t(IGH ) and 6q-, these abnormalities were not included when reporting the hierarchical classification. b Cutoffs were based on a conservative cutoff of 10%, which was determined from interlaboratory validation; see Table 3. c Analysis did not detect 13q-, þ12, 11q-, or 17p-.

Lab B is the referral center for the Leukemia/BMT Program of BC, seeing higher risk patients and those with relapsed/refractory CLL referred for consideration of allogeneic stem cell transplantation. Many of these patients were diagnosed with CLL prior to the routine availability of FISH testing and therefore had a long interval from diagnosis to their first FISH test. An additional difference between laboratories is that Lab A does not test for IGH translocations, which are found in 10e25% of patients in reported studies (12,13,31), leading to a weakness in our ability to estimate a true population-based prevalence of the t(IGH ). Another limitation in our ability to

compare characteristics across laboratories is that we do not have detailed clinical information at the time of the FISH test, but at diagnosis only, reflecting the noneclinical trial nature of this population-based analysis. In our validation study, specimens were selected on the basis of having multiple FISH abnormalities, with abnormal cell populations at or near cutoff values. These samples were selected precisely to test the interlaboratory variability and interpretation of difficult specimens using different probes and protocols. Of the 26 CLL-FISH abnormalities, we found that differences in percentage of abnormalities were

70

P = 0.98 60

Frequency (%)

50 40 30

P = 0.29

P < 0.001

20

P = 0.70

P = 0.008

P = 0.002

10 0

13q-

+12

11qLab A

Lab B

17pLab C

Whole Cohort

t(IGH)

6q-

Dohner

Figure 1 The prevalence of FISH abnormalities in BC CLL cohort, by laboratory and for the whole cohort (n Z 585) compared with the prevalence reported in the literature. The frequencies of CLL patients with 13q-, þ12, 11q-, 17p-, t(IGH ), or 6q- on their first FISHanalyzed sample shown by laboratory (using the universal cutoff of 10%) and for the whole cohort, compared with those of the cohort €hner et al. (7). Lab A, 293 patients; Lab B, 89 patients; Lab C, 103 patients; Do €hner, 325 patients. P values between the entire from Do €hner’s cohort were calculated using the two-sample binomial test of proportions. BC cohort and Do

Provincial CLL FISH standardization generally not more than 10%, except in one observation (sample 7) in which a difference of 20% was noted in the detection of 13q- 1. Of the 78 observations, there were four potential false positive results (5.1%) and three potential false negative results (3.8%). This rate is similar to the CRC’s findings of 3% and 2% for false positive and false negative results, respectively (25). One methodological difference between our study and the CRC study, which may account for the slight difference in concordance rates, is that the CRC study held a workshop for the technologists to assure close concordance in the intricacies of FISH signal scoring. Although we were unable to hold a similar workshop, results of our study were presented at the BC Inter-Hospital Cytogenetic seminar series and discussed with technologists and cytogeneticsts from all three participating laboratories. We note that the use of the wordings ‘false positive’ and ‘false negative’ may not be accurate, as we cannot be certain whether positive results at or near a cutoff are in fact genuine abnormalities, even if two out of three laboratories reported a positive result. A highly quantitative independent method such as quantitative PCR would be required to confirm findings. Variation in factors such as slide preparation, probe hybridization, scoring criteria, technologist variation, number of nuclei analyzed, and calculation of normal cutoff values may have contributed to the discordant results. The use of FISH probes from different manufacturers, which differ in their probe design and coverage (Lab A vs. Labs B and C, Table 1), is another possible explanation for the differences in results. This factor was highlighted by Smoley et al., who reported that the TP53 probe from Abbott Molecular has a tendency to have smaller signals than the other deletion probes, thus making it easier to miss the signal when scoring, thereby leading to more false positive results (25). In addition, the probe design by Abbott Molecular does not use a control probe for chromosome 17, making signal patterns for random loss of chromosome 17 and 17p- equivalent, which when combined may result in an erroneous false positive result near the cutoff. This may explain the two false positive 17p- results noted by Lab A (samples 6 and 7), which is the only laboratory in our study that used this probe. This may also explain why Lab A, which uses the Abbott Molecular TP53 probe, reported the highest number of patients with 17p-. In addition, the percentages of abnormal cells for all false results were reported at or near laboratory cutoff values. The false results involving 17p- were limited to Labs A and B, which had lower normal cutoffs for 17p- than Lab C. Therefore, the combination of variability in probe design and the established lower cutoff values were likely the most important contributing factors in the discrepancy of results observed in our validation study. Given this information, we pooled data from the three BC cytogenetic laboratories, using a universal cutoff value of 10% for all probes (>10% was considered abnormal). When we employed the universal cutoff, the variation in frequencies of þ12 and 17p- previously seen among laboratories was minimized. The greatest change was for 17p-: In Lab A, the frequency changed from14.0% to 9.6%, whereas Labs B and C had no change. Therefore, the higher incidence of 17pinitially observed in Lab A can be explained in part by the lower cutoff used by that laboratory for detection of 17p-. Overall, the FISH abnormality frequencies for the whole cohort dropped slightly but remained similar to those as

323 determined using individual laboratory cutoffs (Table 5 and Figure 1). The use of a 10% universal cutoff value is consistent with the previously reported results and recommendations by the CRC (25). Although the shared 10% cutoff will be used for research purposes only, it is important to consider the clinical implications. There is some suggestion that a low percentage of abnormal nuclei in CLL samples does not carry the same prognosis as a higher percentage for the same FISH abnormality (11,33). Particularly in the case of 17p-, a false positive result may lead to inaccurate counseling regarding prognosis and more aggressive or alternate therapy. Conversely, failing to identify a high risk genetic marker such as 17p- or 11q- may cause a false sense of reassurance for both patient and physician, and may lead to longer follow-up intervals and suboptimal therapy. It is therefore important for clinicians to understand when making treatment decisions involving CLL FISH results that are at or near laboratory cutoff values, clinical correlation, close monitoring of the patient for disease behavior, and repeat FISH testing in follow-up should be considered. The limitations of our study include the limitations associated with the BC Provincial CLL Database, which is a database containing retrospective data on historical cases in addition to prospective data collection since the database’s inception in 2008. Although BCCA provincial guidelines exist for hematologists/oncologists across the province, ultimately, following the guidelines is at the treating physician’s discretion. For this reason, FISH testing may be performed in a selected group of patients, such as those in whom the physician is considering more intensive therapy, in which the FISH result would significantly change disease management (e.g., 17p- in a young patient fit for consideration of allogeneic stem cell transplant). This may explain the short median time from diagnosis to FISH test (0.5 y, range 0e22 y) and the younger median age at diagnosis of our cohort (61 y, range 25e96 y), compared with the current estimates in the general population (72 y) (5). In conclusion, in this large population-based cohort of patients referred for CLL FISH testing, FISH abnormalities were detected in 74.9% of patients with frequencies highly comparable to previously reported single-institution and clinical-trial populations. Districts and provinces that work together in the care of CLL patients can effectively pool data with appropriate laboratory validation and standardization methods. A higher cutoff value when pooling results for research purposes will decrease false positive results. Although FISH analysis has been the gold standard for genetic prognostication in CLL, next generation sequencing technologies have uncovered some new players, notably small mutated TP53 subclones (34) and gene mutations within BIRC3, NOTCH1, and SF3B1, which appear to carry independent prognostic significance (35). Therefore, as we move forward into the era of genomics, clinical laboratories may consider validating and implementing mutational testing of BIRC3, NOTCH1, SF3B1, TP53, and possibly ATM, in addition to FISH analysis, as part of standard CLL patient care. FISH analysis, however, continues to play an important role in the management of CLL patients in BC, and the collection of FISH data combined with clinical information on a population-level is vital from a broad perspective, from counseling individual patients to guiding resource management and provincial therapeutic protocols.

324

A.S. Gerrie et al.

Acknowledgments A.S.G. was supported by the University of BC and Royal College of Canada Clinician Investigator Program. S.J.H. was supported in part by a University of BC Department of Pathology and Laboratory Medicine summer studentship. A Vancouver Coastal Health Research Institute grant and unrestricted research grants from Roche Canada and Genzyme Canada were used to assist with database and project development. The support of the Hematology Research and Clinical Trials Unit is acknowledged. The authors wish to thank Mary Joyce Chan, who assisted with database development, as well as the technologists and staff at BCCA, Royal Columbian Hospital, and Vancouver General Hospital Cytogenetic Laboratories for their excellent technical assistance and acknowledge the physicians of BC for their ongoing support, referrals, and long-term patient follow-up.

14.

15.

16.

17.

18.

References

19.

1. Matasar MJ, Ritchie EK, Consedine N, et al. Incidence rates of the major leukemia subtypes among US Hispanics, Blacks, and non-Hispanic Whites. Leuk Lymphoma 2006;47:2365e2370. 2. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2009. CA Cancer J Clin 2009;59:225e249. 3. Muller-Hermelink HK, Montserrat E, Catovsky D, et al. Chronic lymphocytic leukemia/small lymphocytic lymphoma. In: Swerdlow SH, Campo E, Harris NL, editors. World Health Organization classification of tumours of haematopoeitic and lymphoid tissues. Lyon, France: IARC; 2008. pp. 180e182. 4. Dores GM, Anderson WF, Curtis RE, et al. Chronic lymphocytic leukaemia and small lymphocytic lymphoma: overview of the descriptive epidemiology. Br J Haematol 2007;139:809e819. 5. Seftel MD, Demers AA, Banerji V, et al. High incidence of chronic lymphocytic leukemia (CLL) diagnosed by immunophenotyping: a population-based Canadian cohort. Leuk Res 2009;33:1463e1468. 6. Rozman C, Montserrat E. Chronic lymphocytic leukemia. N Engl J Med 1995;333:1052e1057. € hner H, Stilgenbauer S, Benner A, et al. Genomic aberrations 7. Do and survival in chronic lymphocytic leukemia. N Engl J Med 2000;343:1910e1916. 8. Oliveira AC, de la Banda E, Domingo-Domenech E, et al. Prospective study of clinical and biological prognostic factors at diagnosis in patients with early stage B-cell chronic lymphocytic leukemia. Leuk Lymphoma 2011;52:429e435. 9. Catovsky D, Richards S, Matutes E, et al. Assessment of fludarabine plus cyclophosphamide for patients with chronic lymphocytic leukaemia (the LRF CLL4 Trial): a randomised controlled trial. Lancet 2007;370:230e239. 10. Grever MR, Lucas DM, Dewald GW, et al. Comprehensive assessment of genetic and molecular features predicting outcome in patients with chronic lymphocytic leukemia: results from the US Intergroup Phase III Trial E2997. J Clin Oncol 2007; 25:799e804. 11. Van Dyke DL, Shanafelt TDTD, Call TGTG, et al. A comprehensive evaluation of the prognostic significance of 13q deletions in patients with B-chronic lymphocytic leukaemia. Br J Haematol 2010;148:544e550. 12. Cavazzini F, Hernandez JA, Gozzetti A, et al. Chromosome 14q32 translocations involving the immunoglobulin heavy chain locus in chronic lymphocytic leukaemia identify a disease subset with poor prognosis. Br J Haematol 2008;142:529e537. 13. Gerrie AS, Bruyere H, Chan MJ, et al. Immunoglobulin heavy chain (IGH@) translocations negatively impact treatment-free survival for chronic lymphocytic leukemia patients who have an

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

isolated deletion 13q abnormality. Cancer Genet 2012;205: 523e537. Stilgenbauer S, Bullinger L, Benner A, et al. Incidence and clinical significance of 6q deletions in B cell chronic lymphocytic leukemia. Leuk Off J Leuk Soc Am Leuk Res Fund, UK 1999;13: 1331e1334. Cuneo A, Rigolin GM, Bigoni R, De Angeli C, Veronese A, Cavazzini F, et al. Chronic lymphocytic leukemia with 6q- shows distinct hematological features and intermediate prognosis. Leuk Off J Leuk Soc Am Leuk Res Fund, UK 2004;18:476e483. Wang D-M, Miao K-R, Fan L, et al. Intermediate prognosis of 6q deletion in chronic lymphocytic leukemia. Leuk Lymphoma 2011; 52:230e237. € hner H, Stilgenbauer S. Genetics and risk-stratified Zenz T, Do approach to therapy in chronic lymphocytic leukemia. Best Pr Res Clin Haematol 2007;20:439e453. Hallek M, Cheson BD, Catovsky D, et al. Guidelines for the diagnosis and treatment of chronic lymphocytic leukemia: a report from the International Workshop on Chronic Lymphocytic Leukemia updating the National Cancer Institute-Working Group 1996 guidelines. Blood 2008;111:5446e5456. €hner H, et al. Importance of genetics in Zenz T, Mertens D, Do chronic lymphocytic leukemia. Blood Rev 2011;25:131e137. Government of British Columbia. BCStats Population Estimates, Annual Population. Available from: http://www.bcstats.gov.bc. ca/StatisticsBySubject/Demography/PopulationEstimates.aspx. Accessed on November, 2012. British Columbia Cancer Agency. Cancer Management Guidelines: Chronic leukemia. Available from: http://www.bccancer.bc. ca/HPI/CancerManagementGuidelines/Lymphoma/ChronicLeukemia. htm. Accessed on November, 2012. Gerrie AS, Toze CL, Ramadan KM, et al. Oral fludarabine and rituximab as initial therapy for chronic lymphocytic leukemia or small lymphocytic lymphoma: population-based experience matches clinical trials. Leuk Lymphoma 2012;53:77e82. Toze CL, Dalal CB, Nevill TJ, et al. Allogeneic haematopoietic stem cell transplantation for chronic lymphocytic leukaemia: outcome in a 20-year cohort. Br J Haematol 2012;158:174e185. Mascarello JT, Hirsch B, Kearney HM, et al. Section E9 of the American College of Medical Genetics technical standards and guidelines: fluorescence in situ hybridization. Genet Med 2011; 13:667e675. Smoley SA, Van Dyke DL, Kay NE, et al. Standardization of fluorescence in situ hybridization studies on chronic lymphocytic leukemia (CLL) blood and marrow cells by the CLL Research Consortium. Cancer Genet Cytogenet 2010;203:141e148. Juliusson G, Gahrton G. Chromosome aberrations in B-cell chronic lymphocytic leukemia. Pathogenetic and clinical implications. Cancer Genet Cytogenet 1990;45:143e160. Dewald GW, Brockman SR, Paternoster SF, et al. Chromosome anomalies detected by interphase fluorescence in situ hybridization: correlation with significant biological features of B-cell chronic lymphocytic leukaemia. Br J Haematol 2003;121: 287e295. Haferlach C, Dicker F, Schnittger S, et al. Comprehensive genetic characterization of CLL: a study on 506 cases analysed with chromosome banding analysis, interphase FISH, IgV(H) status and immunophenotyping. Leukemia 2007;21:2442e2451. Hallek M, Fischer K, Fingerle-Rowson G, et al. Addition of rituximab to fludarabine and cyclophosphamide in patients with chronic lymphocytic leukaemia: a randomised, open-label, phase 3 trial. Lancet 2010;376:1164e1174. Lu G, Kong Y, Yue C. Genetic and immunophenotypic profile of IGH@ rearrangement detected by fluorescence in situ hybridization in 149 cases of B-cell chronic lymphocytic leukemia. Cancer Genet Cytogenet 2010;196:56e63. Nowakowski GS, Dewald GW, Hoyer JD, et al. Interphase fluorescence in situ hybridization with an IGH probe is important

Provincial CLL FISH standardization in the evaluation of patients with a clinical diagnosis of chronic lymphocytic leukaemia. Br J Haematol 2005;130:36e42. 32. Haferlach C, Dicker F, Weiss T, et al. Toward a comprehensive prognostic scoring system in chronic lymphocytic leukemia based on a combination of genetic parameters. Genes Chromosomes Cancer 2010;49:851e859. 33. Tam CS, Shanafelt TD, Wierda WG, et al. De novo deletion 17p13.1 chronic lymphocytic leukemia shows significant clinical

325 heterogeneity: the M. D. Anderson and Mayo Clinic experience. Blood 2009;114:957e964. 34. Rossi D, Khiabanian H, Spina V, et al. Clinical impact of small TP53 mutated subclones in chronic lymphocytic leukemia. Blood 2014;123:2139e2147. 35. Rossi D, Rasi S, Spina V, et al. Integrated mutational and cytogenetic analysis identifies new prognostic subgroups in chronic lymphocytic leukemia. Blood 2013;121:1403e1412.