Limited Utility of Fluorescence In Situ Hybridization for Recurrent Abnormalities in Acute Myeloid Leukemia at Diagnosis and Follow-up

Abstract

Objectives

Acute myeloid leukemia (AML) is classified in part by recurrent cytogenetic abnormalities, often detected by both fluorescent in situ hybridization (FISH) and karyotype. The goal of this study was to assess the utility of FISH and karyotyping at diagnosis and follow-up.

Methods

Adult AML samples at diagnosis or follow-up with karyotype and FISH were identified. Concordance was determined, and clinical characteristics and outcomes for discordant results were evaluated.

Results

Karyotype and FISH results were concordant in 193 (95.0%) of 203 diagnostic samples. In 10 cases, FISH detected an abnormality, but karyotype was normal. Of these, one had a FISH result with clinical significance. In follow-up cases, 17 (8.1%) of 211 showed FISH-positive discordant results; most were consistent with low-level residual disease.

Conclusions

Clinically significant discordance between karyotype and AML FISH is uncommon. Consequently, FISH testing can safely be omitted from most of these samples. Focused FISH testing is more useful at follow-up, for minimal residual disease detection.

Acute myeloid leukemia (AML) is the most common acute leukemia in adults and is characterized by an aggressive clinical course and variable outcome due in part to clinical and biological heterogeneity. Cytogenetic studies are an important tool in the classification and risk stratification of AML into favorable, intermediate, and unfavorable risk groups.^1-3 Several well-known, recurrent cytogenetic abnormalities have prognostic significance. These include t(8;21)(q22;q22) [RUNX1-RUNX1T1], inv(16)(p13.1q22) or t(16;16)(p13.1;q22) [CBFB-MYH11], inv(3)(q21q26.2) or t(3;3)(q21;q26.2 [RPN1-MECOM], t(6;9)(p23;q34) [DEK-NUP214], and KMT2A (11q23) rearrangements. Thus, it is important to test for these at diagnosis of AML. In addition to cytogenetic analysis, testing for common single-gene mutations (FLT3, NPM1, KIT, and CEBPA) is recommended to aid in stratification of patients with newly diagnosed AML.² Many clinical laboratories supplement this with next-generation sequencing panels to identify mutations in other genes that may be of clinical significance.^4,5

It is common practice for laboratories to use interphase fluorescence in situ hybridization (FISH) panels in conjunction with karyotyping for the detection of the most prognostically significant recurrent AML-associated cytogenetic abnormalities, both in diagnostic specimens and during follow-up to monitor response to therapy. FISH is targeted toward specific abnormalities, and results can be evaluated in an automated fashion on interphase nuclei, allowing for examination of more cells than a traditional karyotype. In addition, because the probe size is much smaller than the resolution provided by the banding pattern in karyotype, FISH can detect cryptic abnormalities not revealed by conventional karyotype. Thus, it has higher analytic and, in certain circumstances, higher clinical sensitivity compared with karyotype.⁶ The rationale for employing FISH in AML diagnosis, therefore, is to aid in the detection of cryptic rearrangements or low-level abnormalities or to identify abnormalities in cases where adequate metaphase spreads are not available for analysis. However, some recent studies have called into question whether FISH actually provides additional diagnostic information compared with karyotype, especially considering its significantly higher cost.⁷

In light of these concerns and given the importance of genetic testing in AML, we sought to determine the utility of FISH testing for common recurrent genetic rearrangements in AML at diagnosis and in follow-up testing. We found that the usefulness of FISH is limited at diagnosis, except in very specific circumstances. However, targeted FISH testing may help identify low-level disease in follow-up samples. Thus, the use of FISH at diagnosis could be reduced without compromising the quality of patient care.

Materials and Methods

Case Identification and Classification

This study was approved by the Vanderbilt University Institutional Review Board. Bone marrow specimens obtained for diagnosis or follow-up of AML on adult patients between August 2010 and July 2014 were identified. Diagnosis of AML was based on World Health Organization criteria.^1,3 Cases were included if they included both a karyotype and an AML FISH analysis with at least one probe set from the AML FISH panel (described below).

The cases were divided into two groups based on available clinical history and previous testing results. The first group (n = 204) was composed of specimens (bone marrow biopsy specimens or peripheral blood samples) obtained to diagnose AML at the time of first presentation. Patient samples in this group had no history of AML. The second group (n = 217) consisted of follow-up bone marrow biopsy specimens in patients with a previous diagnosis of AML being monitored for disease progression and/or response to therapy.

Cytogenetic Analysis

Karyotype and FISH analyses were performed on each patient’s bone marrow or peripheral blood specimen in the Vanderbilt Cytogenetics Laboratory. Briefly, fresh bone marrow aspirates or peripheral blood specimens were subjected to overnight culture and harvested using standard cytogenetic methods. Routine karyotyping was performed following Giemsa-Pancreatin-Wright stain banding to obtain 400-band-level metaphase chromosomes. Karyotypes, including normal and abnormal findings, were reported according to the International System for Human Cytogenetic Nomenclature.⁸ If fewer than 20 metaphase cells were obtained, with normal karyotypes, the case was deemed inadequate.

FISH was performed on interphase nuclei according to standard methods using commercially available probes.⁹ The AML panel included probes for t(8;21) (RUNX1/RUNX1T1) and t(15;17) (PML/RARA), both dual-color dual-fusion probes, and KMT2A (11q23) and inv(16)/t(16;16) (CBFB), both dual-color break-apart probes (Vysis; Abbott Laboratories, Des Plaines, IL). Lower limits of analytical sensitivity (normal cutoffs) were calculated for each probe set in the laboratory using established recommendations and determined to be 1% for each of the probes in this study.¹⁰ Two trained technologists scored 200 cells per case for each probe set.

Results

Karyotype and FISH Results at Diagnosis

To determine the utility of performing FISH for AML-associated abnormalities at the time of initial diagnosis, we analyzed the karyotype and FISH results for bone marrow and peripheral blood specimens from 204 patients with a new diagnosis of AML who sought treatment from Vanderbilt University Medical Center over 4 years. The characteristics of these cases are shown in Table 1.

Table 1.

Characteristics of Diagnostic Cases

Characteristic	Value
No. of specimens	204
No. of individual FISH tests (probes or probe sets)	791
Patient age, median (range), y	63 (19-85)
No. of men/women (ratio)	124/80 (1.5)
Diagnosis, No. (%)
AML RCA	37 (18.1)a
AML NOS	88 (43.1)
AML MRC	67 (32.8)
tAML	11 (5.4)
MPAL	1 (0.5)

AML MRC, acute myeloid leukemia with myelodysplasia-related changes; AML NOS, acute myeloid leukemia not otherwise specified; AML RCA, acute myeloid leukemia with recurrent cytogenetic abnormalities; FISH, fluorescence in situ hybridization; MPAL, mixed-phenotype acute leukemia; tAML, therapy-related acute myeloid leukemia.

^aCytogenetic findings in AML RCA cases: t(8;21) in three (1%), inv(16) in 12 (6%), t(15;17) in 14 (7%), KMT2A in five (2.5%), t(6;9) in one (0.5%), inv(3) in two (1%).

Adequate karyotypes were obtained in 203 (99.5%) of 204 cases. One diagnostic peripheral blood sample had no analyzable metaphases, but FISH was positive for inv(16). Of the remaining samples, most (196/203, 96.5%) had FISH analysis performed for all four probes in the AML panel. Concordance between karyotype and FISH results is shown in Table 2. Clonal karyotype abnormalities were observed in 125 (61.5%) of 203 cases. In 97 of these cases, the karyotypic abnormality was also detected by FISH. In the remaining 28 cases with abnormal karyotype, FISH was uninformative because the probes did not include the region of karyotypic abnormality. The remaining 78 cases had a normal karyotype. Of these, 68 (33.5%) showed normal FISH results, for a total of 193 (95.0%) of 203 concordant cases.

Table 2.

Concordance of Karyotype and FISH at Diagnosis

Karyotype Result	Concordance With FISH Results, No. (%)
	Concordant	Discordant
Abnormal	125 (61.5)a	0 (0)
Normal	68 (33.5)	10 (5)

FISH, fluorescence in situ hybridization.

^aKaryotype-positive concordant cases include cases with abnormal FISH results that were consistent with the karyotype and normal FISH results if the karyotype abnormality was not included on the FISH panel.

The remaining 10 (5.0% of total) cases showed discordance, with normal karyotype and abnormal FISH results. Table 3 shows the details of these 10 discordant cases. In six (cases 1-6), the abnormal FISH findings were low level (<10%) losses or gains of a single locus. One case (case 7) showed extra copies of multiple probes, possibly due to a polyploid cell line that was not found on routine cytogenetic analysis. Follow-up karyotype analysis of this patient revealed a complex, abnormal karyotype with nonspecific chromosomal gains, losses, and rearrangements. Case 8 was found to have several abnormal fluorescence patterns for CBFB, including losses of either 5ʹ or 3ʹ CBFB signals in 4.7%, gain of the 5ʹ CBFB signal in 3%, and rearrangement of CBFB in 1.3% of cells. Case 9 showed an extra, intact copy of KMT2A at 11q23 in 74.5% of cells. On subsequent follow-up karyotype of this patient, a near-triploid clone was observed that was not identified at the time of diagnosis, and the presence of this clone was associated with morphologic evidence of disease. For each of these cases, the clinical relevance of these findings at diagnosis was uncertain and did not affect diagnosis, prognosis, future monitoring, or treatment. Clinical relevance was clear in one case (case 10) in which FISH for t(15;17) was positive, and subsequent metaphase FISH studies indicated that this case represented a cryptic rearrangement with insertion of RARA into PML.

Table 3.

Diagnostic Cases With Normal Karyotype and Abnormal FISH

Case No.	Blasts, %	FISH, %a	FISH Abnormality Detected
1	29	4.5	Loss of RARA (17q21)
2	20	2	Gain of RUNX1 (21q22)
3	45	9.5	Gain of RUNX1T1 (8q21.3)
4	31	7.5	Loss of RARA (17q21)
5	53	4.5	Loss of RARA (17q21)
6	68	7	Loss of RUNX1 (21q22)
7	49	64-72	Extra copies of multiple probes
8	65	9	Loss and gain of CBFB (16q22)
9	71	74.5	Gain of KMT2A (11q23)
10	86	37.5	PML/RARA (cryptic insertion of RARA into PML)

FISH, fluorescence in situ hybridization.

^aPercent cells positive for abnormal pattern out of 200 total cells analyzed.

Karyotype and FISH Results at Follow-up

Given the higher analytical sensitivity of FISH over karyotype,¹¹ it could theoretically have utility in monitoring response to therapy and for assessment of minimal residual disease in the context of other clinical and pathologic findings. To evaluate this hypothesis, we analyzed the results from 217 follow-up specimens over the same 4-year time period Table 4. Specimens in which both karyotype analysis and at least one FISH test was performed were included. Of these, six (2.8%) specimens exhibited culture failure or inadequate metaphases for complete cytogenetic analysis, likely related to hypocellularity following chemotherapy. Concordance between karyotype and FISH results for the remaining 211 specimens is shown in Table 4. Overall, concordant results were seen in 192 (91.0%) of the 211 cases with adequate karyotype. Most of these (119/192; 56.4%) were normal for both karyotype and FISH. The remaining concordant cases included 60 specimens with abnormal karyotype and corresponding abnormal FISH results and 13 specimens with an abnormal karyotype and normal FISH results in which the karyotypic abnormality was not detectable by the FISH probes used.

Table 4.

Concordance of Karyotype and FISH at Follow-up

Karyotype Result	Concordance With FISH Results, No. (%)
	Concordant	Discordant
Abnormal	73 (34.6)a	2 (0.9)
Normal	119 (56.4)	17 (8.1)

FISH, fluorescence in situ hybridization.

^aKaryotype-positive concordant cases include cases with abnormal FISH results that were consistent with the karyotype and normal FISH results if the karyotype abnormality was not included on the FISH panel.

The remaining 19 (9%) specimens were discordant for karyotype and FISH results. In two (0.9%) cases, a single abnormal metaphase was found by conventional cytogenetics, and the FISH assay did not identify the abnormality. Although single abnormal metaphase cells are not considered clonal, the abnormalities were consistent with those seen in the diagnostic study and were therefore reported.

Seventeen (8.1%) specimens from 14 different patients were discordant with abnormal FISH and normal karyotype. These results are summarized in Table 5. For each of these cases, the percentage of abnormal cells was below 10% (range, 1%-9%), suggesting that the cytogenetic abnormality may not have been detected by karyotype due to reduced analytical sensitivity. Five cases with normal cytogenetics had an abnormal FISH result showing a low-level CBFB rearrangement (1.5%-3%, cases 1-5). Two cases from one patient were positive for a KMT2A rearrangement (1% and 6%, cases 6 and 7) and one case (case 8) was positive for a t(10;21) rearrangement, resulting in an extra signal for RUNX1. A single case (case 9) was positive for a t(15;17) at 1%, and a second case (case 10) that was positive at diagnosis for t(15;17) showed an abnormal result with a loss of RARA in 4.7% of cells, the significance of which is uncertain, but reverse transcription polymerase chain reaction of the specimen was positive for a PML/RARA fusion product, indicating the specimen may have been positive for the rearrangement at a low level. Four of the remaining cases were follow-up marrows from two different patients with normal cytogenetics and similar low-level abnormal findings at diagnosis and follow-up (cases 11-14). Cases 15 and 16 showed low-level gains of RUNX1T1 and RUNX1 and relapsed with trisomy 8 and trisomy 21, respectively. The final case from the karyotype normal/FISH abnormal category was unusual in that FISH and karyotype were both normal at diagnosis. These patients do not generally get FISH analysis at follow-up, but in this case, bone marrow was sent for testing, and FISH revealed a 4% loss of RUNX1 and a 3% loss of KMT2A, both of uncertain clinical significance (case 17). No other follow-up specimens were received from this patient.

Table 5.

Follow-up Cases With Normal Karyotype and Abnormal FISH

Case No.	FISH Abnormality	FISH Abnormal, %a	Morphology	Time From Diagnosis, d	Clinical Follow-up
1	inv(16)	2	–	29	Relapse within 1 year, deceased
2	inv(16)	2.7	–	14	Successful BM transplant
3	inv(16)	1.5	+	14	Relapse at 1 year → remission
4	inv(16)	1.5	–	15	No evidence of relapse
5	inv(16)	3	–	15	Relapse at 1 year → remission
6	KMT2A b	1	–	93	Relapse within 1 year, deceased
7	KMT2A b	6	+	120	Relapse within 1 year, deceased
8	t(10;21)	1	–	120	Relapse within 1 year, deceased
9	t(15;17)	1	–	54	No evidence of relapse
10	Loss of RARA	4.7	–	47	No evidence of relapse
11	Loss of RARAb	3	–	32	No evidence of relapse
12	Loss of RARAb	4	–	122	No evidence of relapse
13	Loss of RUNX1b	6.5	–	31	Successful BM transplant
14	Loss of RUNX1b	4	–	83	Successful BM transplant
15	Gain of RUNX1T1	9	+	406	Relapse within 1 year
16	Gain of RUNX1	3.5	–	816	Relapse within 1 year, deceased
17	Loss of RUNX1/loss of KMT2A	4/3	–	18	No follow-up available

BM, bone marrow; FISH, fluorescence in situ hybridization; +, positive; –, negative.

^aPercent cells positive for abnormal pattern out of 200 total.

^bCases 6 and 7, same patient; cases 11 and 12, same patient; cases 13 and 14, same patient.

The time, in days, between diagnostic marrow sampling and follow-up testing is also indicated in Table 5. In general, the standard practice at our institution includes bone marrow testing at approximately 14 to 15 days postdiagnosis (end of induction) and again at 29 to 30 days, up to approximately 60 days (recovery). These end-of-induction and recovery bone marrow assessments include flow cytometry, morphologic examination, karyotype analysis, and FISH testing if a previous abnormality was detected. These were the time points for 10 of the 17 discordant, FISH-positive cases (cases 1-5, 9-11, and 17), and all but one showed no morphologic evidence of disease. The finding of a FISH-positive result in these specimens is therefore due to low-level residual disease. Of the remaining seven samples, five were drawn at approximately 90 days after diagnosis (postconsolidation), and two were distant follow-up specimens, drawn more than a year after diagnosis. For five of these samples, relapse occurred within a year of the abnormal FISH result. Thus, positive FISH results in these cases may have been early harbingers of relapse.

Analysis of morphology and flow cytometry data for these discordant specimens demonstrated no evidence of disease in 14 of the 17 cases, while three cases (cases 3, 7, and 15) showed evidence of abnormal cells or increased blasts. In seven of the 13 patients for whom we have follow-up testing, the discordant abnormal FISH result preceded a relapse event within 1 year of testing. Five of those patients had at least one follow-up specimen that was FISH positive but showed no morphologic signs of disease (see Table 5). Of the six patients who did not relapse, two had successful bone marrow transplants and the other four had FISH-positive specimens drawn at approximately 14 days postdiagnosis/induction, indicating residual disease rather than early signs of relapse.

Discussion

Clinical management for patients with hematologic malignancies relies on a combination of morphologic features, immunophenotypic data, and genetic test results. This information is used to classify a diverse group of diseases into distinct disease categories that guide diagnosis, prognosis, and treatment-related decisions. In recent years, genetic testing for AML has expanded and now routinely includes conventional karyotyping, FISH, single-gene molecular testing, and multigene panel sequencing. This testing, although costly, has become an integral part of disease management at diagnosis and for posttherapy monitoring. In the current environment of rising health care costs and concerns of test overutilization, it is important to determine if laboratory testing is cost-effective and of clear benefit to patients and clinical care providers.

The results of this study suggest that for most patients, karyotype and FISH are equally effective in the detection of recurrent cytogenetic rearrangements at diagnosis. However, these data confirm that karyotype has superior clinical sensitivity to FISH in that it detects all structural abnormalities when adequate metaphases are present as a well-known consequence of the unbiased genome-wide coverage of karyotype compared to targeted FISH panels. Taken together, these findings suggest that karyotype is a superior tool for identifying abnormalities at AML diagnosis and that FISH testing is largely redundant.

These findings are similar to those recently published by He et al,⁷ which also suggested that targeted FISH is not indicated or necessary for most diagnostic AML samples. Compared with He et al,⁷ we found a higher proportion of karyotype-negative/FISH-positive cases in our diagnostic cohort (5% vs 1.8%). However, the majority of these (9/10) were nonspecific and/or low level and thus did not affect the patient’s diagnosis, prognosis, or clinical management. Only one (<1%) discordant case had an abnormal FISH result with true clinical relevance at the time of diagnosis. This case had a cryptic RARA rearrangement that involved the insertion of RARA into PML that was not visible at the level of karyotype analysis. Because of this, and due to the need for rapid results for cases of suspected acute promyelocytic leukemia (APL), we continue to recommend FISH for PML/RARA rearrangement in diagnostic cases that present with a clinical and/or morphologic concern for APL. An additional consideration is KMT2A. In our study, six samples with KMT2A rearrangements were detected at diagnosis, including five cases of t(9;11) and one case of t(11;19). Although a visible translocation was observed in each of our cases by karyotype analysis, KMT2A rearrangements have been documented by Southern blot, FISH, or molecular methods in patients previously found to have a normal karyotype.^12,13 It has been recognized that 11q23 abnormalities may be difficult to detect due to subtle changes in banding that can be missed, especially in cases with poor morphology, and rarely, truly cryptic rearrangements have been documented.^6,14,15 Therefore, it is theoretically possible that some cases with KMT2A rearrangements may be missed if FISH is not performed, although none were detected in this study.

Of note, we have a lower culture failure and insufficient karyotype rate for diagnostic specimens than was reported in the He et al⁷ study. Our cytogenetics laboratory had only one (<1%) culture failure out of the 204 specimens received for AML diagnosis and no cases with inadequate karyotype compared with the reported six (2.4%) of 250 cases of culture failure and 24 (9.6%) of 250 cases with insufficient (<20) metaphases by He et al.⁷ This is likely due to differences in our patient populations and the time between specimen collection and culture setup. The Vanderbilt Cytogenetics Laboratory primarily serves patients seen at Vanderbilt hospitals and clinics; therefore, the cultures are routinely set up on the day of collection. This differs from a reference laboratory in which travel distance might preclude same-day culture setup, which is known to have an impact on successful culture of metaphases. The culture failures and inadequate karyotype cases still represent a small minority of all diagnostic cases but may require that laboratories continue to perform FISH studies in those circumstances for a full cytogenetic analysis.

One key addition to our study is evaluation of the utility of FISH and karyotype analysis in follow-up bone marrows. Although a higher culture failure/inadequate karyotype rate was observed (6/217, 2.8%), possibly reflecting the myelotoxic effect of chemotherapy, the overall results were similar to the diagnostic cohort, with a few exceptions. A higher proportion of follow-up samples (17/211, 8.1%) showed a normal karyotype and abnormal FISH results, as might be expected due to the superior analytical sensitivity of FISH compared with standard metaphase analysis at low levels of disease. Of the discordant samples, 10 showed an abnormal FISH result with clinical significance, and all of these were low-level rearrangements (1%-9%) that were detected at diagnosis by both FISH and routine cytogenetics. The remaining cases had low-level gains and losses of uncertain significance. Fewer cases (7.1% vs 14% in the follow-up and diagnostic time points, respectively) showed normal FISH and abnormal karyotype, which is likely due to the use of targeted FISH on follow-up marrow testing rather than use of the panel of four probe sets on all diagnostic cases, which is the current practice at Vanderbilt. It is important to note that in the absence of morphologic findings, no clinical decisions regarding treatment were made based solely on abnormal FISH results. Rather, the FISH results are used in conjunction with other available clinical and pathologic results, including flow cytometry and morphology, to determine if a patient may be showing signs of relapse, and closer monitoring may be warranted.

A similar study was designed to determine the utility of FISH testing in cases of myelodysplastic syndrome (MDS), and the results indicated that FISH at diagnosis and follow-up was rarely discordant from karyotype analysis.¹⁶ Based on the results of that study and other studies with similar conclusions,^17-19 evidence-based guidelines for karyotype and FISH testing in MDS were developed at Vanderbilt University Medical Center that employ MDS FISH only in diagnostic cases with inadequate karyotype and targeted FISH at follow-up. The use of these guidelines has resulted in a reduction in the number of tests performed both at diagnosis and at follow-up.²⁰

Our data suggest limited utility for FISH testing at diagnosis in cases of adult AML with adequate karyotype analysis. We expect that guidelines that limit AML FISH to cases with culture failure, those with inadequate karyotype (fewer than 20 metaphases) or suboptimal resolution (metaphases that do not achieve at least 400-band-level resolution), or cases of suspected APL will have several benefits. The most obvious benefit is a reduction in unnecessary laboratory testing that provides redundant information to clinicians and patients and the corresponding reduction in health care costs that would accompany the decreased testing. Another potential benefit to performing fewer FISH tests is a reduction in the reporting of “false-positive” results. Seven of the 10 discordant FISH-positive/karyotype-negative cases at diagnosis were low-level gains and losses that did not correlate with the presence of blasts, which could represent false-positive results. This type of result does not affect clinical care because the significance of these findings is unclear and may lead to additional and unnecessary follow-up testing. One possible way laboratories could reduce diagnostic FISH testing is to use AML FISH as a reflex test in the case of inadequate karyotype or culture failure. Since a cultured call pellet can be used for both karyotype and FISH analysis, no additional specimen collection or harvesting is necessary. At our institution, we have implemented a strategy to reduce unnecessary testing of bone marrow samples that we have described as a “grow and hold.” This includes culturing and harvesting of all samples upon receipt in the laboratory but a hold on karyotype and FISH test ordering until review of the bone marrow biopsy specimen by a hematopathologist is done to determine what, if any, cytogenetic testing is warranted.

In follow-up bone marrows, FISH for recurrent AML rearrangements has greater utility than at presentation, especially when focused on using FISH probes for rearrangements that were previously documented. In some instances, as we have documented, positive FISH results may precede morphologic evidence in cases of disease relapse, and FISH analysis of specimens without evidence of overt disease may be beneficial. However, due to the frequent finding of low-level FISH positivity at both the postinduction and the recovery time points without morphologic evidence of disease, the data suggest that FISH testing in these cases may not provide the clinical benefit necessary to warrant its use for these patients.

Evidence-based utilization of laboratory testing relies on analysis of accurate and quantifiable data to develop and support guidelines.^20,21 Taken together, this study and previous reports provide support for a more targeted role for FISH testing in diagnostic and follow-up bone marrow biopsy specimens for AML. In the future, similar studies will be needed to fully evaluate the use of cytogenetic and molecular testing in AML and other hematologic and solid malignancies.