The Value of Next-Generation Sequencing in the Screening and Evaluation of Hematologic Neoplasms in Clinical Practice

Victoria Northrup, MSc; Allison Maybank, MSc; Nancy Carson, PhD; Tarek Rahmeh, MD


Am J Clin Pathol. 2020;153(5):639-645. 

In This Article


Among the 178 cases, a total of 95 (53%) were found to have either a gene variant and/or gene fusion by NGS and are considered positive, for a total of 189 gene variants and eight gene fusions reported. The percentage of positive cases was significantly higher in bone marrow samples (72% [31/43]) compared with peripheral blood samples (47% [64/135]) (P = .0016).

There was a wide variation of the number of variants detected in the 95 positive cases (range, one to eight variants per case) with a mean (SD) of 2.1 (1.57) variants per case. In terms of pathogenicity, 67% (64/95) of positive cases harbored known pathogenic variants. A total of 29% (28/95) of positive cases were found to have at least one variant that had not been previously reported in the literature (novel variant), with a total of 39 novel variants, representing 21% (39/189) of all variants reported in this cohort. Seventy-four percent (29/39) of novel variants detected were deemed to be pathogenic, considering that they are of the same mutational categories as other known pathogenic variants in those genes Table 1. The remaining 26% (10/39) of novel variants were determined to be VUSs, based on either limited information or complete lack thereof in the medical literature, COSMIC database for somatic mutations, and ClinVar archives or due to conflicting reports on pathogenicity between different resources (Supplementary Table 1; all supplemental materials can be found at American Journal of Clinical Pathology online). Overall, 34% (32/95) of all positive cases contained at least one VUS and 18% (34/189) of all variants reported in this cohort of patients were VUSs. Among known pathogenic variants, the most common were those involving JAK2 (12/108 or 11%), TP53 (12/108 or 11%), SF3B1 (10/108 or 9%), and NRAS (9/108 or 8%). Among novel variants, by far the most common were those involving the TET2 gene (18/36 or 50%), followed by ZRSR2 (4/36 or 8%) and ASXL1 (4/36 or 8%).

Of the 32 cases with a VUS reported, 16 (50%) had no other gene variants that were pathogenic, with VUS being the only aberrant NGS finding (Supplementary Table 2). In the remaining 16 cases, at least one pathogenic variant was also identified.

Cytogenetic (CG) karyotyping was done at some point in time in 75 cases. Karyotyping was considered concurrent if the study was done within a month or less prior to or after the NGS study date, with 50 studies meeting this criterion. All other karyotypes were done within a year or less, prior or after, the date of NGS, with the exception of one case in which the only karyotype available was done within 27 months prior to NGS. Karyotyping was positive in only 42% (21/50) of concurrent studies and 39% (29/75) of all studies. Karyotyping failed in two of the concurrent studies and was considered suboptimal in four due to eight or fewer metaphases available for evaluation. All patients with positive karyotyping, with the exception of one, had positive NGS results (95% or 20/21 of concurrent cases and 97% or 28/29 of all positive cases). In contrast, only 45% (21/47) of positive NGS cases had positive concurrent karyotypes, and only 38% (28/74) of all positive NGS cases had a positive karyotype at some point (concurrent or nonconcurrent). When comparing the sensitivity of NGS and cytogenetics, NGS had a much higher sensitivity (91.7% vs 55.8%, P = .001), with both having similar specificities (99% vs 93%). Positive karyotyping results were considered concordant if the chromosomal findings mirrored the genetic aberrancies detected by NGS, partially concordant if either karyotyping or NGS demonstrated an additional finding not mirrored in the other, and unrelated if the aberrancies involved completely different chromosomes and/or genes. A comparison of NGS and CG results is summarized in Table 2.

We stratified cases based on concurrent CBC and peripheral blood smear findings, the two being strong indicators of the hematological state of the patient. Table 3 lists the various peripheral blood presentations with their frequencies, as well as the frequency at which each of these presentations yielded positive NGS results. No CBC was available for review in 21 of the 178 cases, being all referred in from other medical centers.

Hematology consultation reports were reviewed to determine the clinical indication for ordering NGS. Table 4 lists these clinical indications, their frequencies and percentages of all cases, and the frequency at which each of these indications was positive by NGS.

NGS in our cohort provided evidence of pathogenic variants that supported the diagnosis of MDS in 11 cases of anemia and/or cytopenia in which cytogenetic studies were either negative or inadequate (Table 2). In five of these cases, NGS was performed on peripheral blood with no bone marrow sample obtained. In six additional cases, cytogenetic studies revealed chromosomal abnormalities that are diagnostic of MDS, and thus the NGS findings were complementary to cytogenetics and provided mainly prognostic value. In three patients, only VUSs in the DNMT3A gene were detected. In two known MDS cases, NGS provided prognostic information, and in a third case of a patient with MDS who developed progressive thrombocytosis, a combination of JAK2 and U2AF1 pathogenic variants was identified, confirming evolution into MDS/MPN and indicating worse prognosis, respectively.

In all 14 de novo AML cases, NGS provided prognostic information in addition to potential molecular targets for future minimal residual disease (MRD) assessment. In patients with posttreatment assessment of AML, NGS provided genetic evidence of MRD in two cases, in which the bone marrow was morphologically negative and cytogenetics were inadequate. In five other cases, NGS provided complementary evidence of recurrent and/or persistent AML plus prognostic information in which the bone marrow was morphologically positive.

NGS showed evidence of a known pathogenic variant and fusion genes that are typical for MPN and was essential in making the diagnosis in seven cases. In four additional cases, no bone marrow was done at the time of the study, but there were pathogenic genetic abnormalities consistent with MPN, an indication of a neoplastic process, warranting further hematologic investigation. Among the 11 patients with known MPN who were being followed up, seven were positive for known pathogenic variants consistent with MPN, including one patient with primary myelofibrosis with a JAK2 variant in addition to SRSF2 and TP53 pathogenic variants signaling worse outcome and high risk of disease progression. Another case with essential thrombocythemia with peripheral blood changes suggestive of evolving into MDS/MPN, in addition to a CALR variant, had a pathogenic SF3B1 variant that is typical for MDS with ringed sideroblasts.

NGS showed pathogenic variants, which constituted further evidence of chronic myelomonocytic leukemia, in four cases that had abnormal bone marrow morphology and one in which the bone marrow appeared normal. CG studies on the above cases were either negative or inadequate.

Clinical information was available in 12 of the 16 cases that harbored only VUS with no evidence of pathogenic variants. In six of these cases, the VUSs were believed to be an incidental finding, with no evidence of a causative relationship with the clinical presentation. The diagnostic or prognostic significance of VUS in the remaining six cases could not be determined (Supplementary Table 2 and Supplementary Table 3). In the 16 cases in which VUSs were identified in conjunction with pathogenic gene variants, diagnostic decision making and final patient disposition were based on the latter rather than the VUS. While this does not exclude involvement of these VUSs in disease etiology or progression, it is difficult to determine whether they have an independent pathogenic role in the presence of the other known pathogenic variants in these cases.

The replacement of the AmpliSeq Myeloid Panel with the Oncomine Myeloid Panel in 2018 allowed the detection of variants in genes that were not part of the previous panel in five of the 33 cases that were tested with the Oncomine panel. In three of these cases, the information provided had diagnostic and/or prognostic clinical value. In the other two cases, the gene variants detected were novel and of undetermined clinical significance, and they had no impact on the management of the patient (Supplementary Table 4).