Prognostic significance of blast immunophenotype on first post-induction bone marrow biopsy

Acute myeloid leukemia (AML) is a life-threatening disease that requires urgent treatment. The standard of care for initial treatment is induction chemotherapy, most commonly with a backbone including a cytarabine and an anthracycline.¹ The goal of this regimen is to achieve a complete morphologic remission. The response to this course of chemotherapy is initially assessed by a bone marrow performed 14 to 21 days from the start of chemotherapy. However, the value and timing of this first assessment is often debated.² If there is hypoplasia, as defined by NCCN guidelines as cellularity <20% and blasts <5%, then no further therapy is typically given and the patient’s marrow is reassessed at time of hematologic recovery to determine remission status. NCCN guidelines suggest re-induction if there is “significant cytoreduction with low % residual blasts” at the day 14–21 marrow, but they also suggest repeating a bone marrow biopsy in 5–7 days if the physician is not sure about the presence of residual AML.³ A similar approach is noted by Cheson et al who recommend repeating the bone marrow biopsy if blasts are 5–20% in the bone marrow biopsy, as in some patients the increased blasts disappear in the next bone marrow biopsy.⁴

Multiparameter flow cytometry to detect residual disease or measurable residual disease (MRD) is now performed in many institutions as part of the assessment for complete remission at time of expected hematologic recovery. However, the referenced guidelines do not incorporate phenotypic abnormalities of the blasts as a point to consider in the first post-induction bone marrow assessment at day 14–21. There are only a few studies in the literature that have investigated this question, and some have used relatively sophisticated flow cytometric panels that are geared toward MRD assessment and have MRD-level limits of detection for phenotypically abnormal blasts.^5,6 In this study, we investigated whether blast immunophenotypic abnormalities identified by routine, non-MRD-level multiparameter flow cytometry on the first bone marrow assessment had a better predictive value for progression when compared to the manually counted absolute blast percentage on the marrow aspirate.

We identified patients with AML that had received induction chemotherapy between June 2016 and January 2021 at UPMC Hillman Cancer Center. From these patients, we included in the analysis anyone who did not receive subsequent therapy between their initial post-induction bone marrow assessment and the next bone marrow biopsy. Patients that did not have an aspirate with acceptable quality for flow cytometry were excluded from the analysis. All patients included in this study were induced with cytarabine (100 mg/m²/day/IV for 7 days) and idarubicin (12 mg/m²/day/IV days 1–3). For patients with FLT3 mutations midostaurin was added. Prognostic factors at diagnosis of AML were evaluated using the ELN 2017 stratification⁷ and response to induction chemotherapy was evaluated with established criteria.^4,7 The study was approved by the University of Pittsburgh Institutional Review Board.

Flow cytometric analysis of bone marrow aspirate material was performed using FACSCanto II (2016–2020) and FACSLyric (2020–2021) flow cytometers (BD Biosciences, San Jose, CA). The following antibody:fluorophore combinations were assessed: Tube 1—CD36:FITC (CB38, Pharmingen), CD123:PE (9F5, BD Biosciences), CD64:PerCP-Cy5.5 (10.1, BD Biosciences), CD33:PE-Cy7 (P67.6, BD Biosciences), CD34:APC (8G12, BD Biosciences), CD14:APC-H7 (MφP9, BD Biosciences), HLA-DR:V450 (L243, BD Biosciences), CD45:V500 (HI30, BD Biosciences); Tube 2—CD15:FITC, (W6D3, BioLegend) CD56:PE (MY31, BD Biosciences), CD117:PerCP-Cy5.5 (104DR, BD Biosciences), CD13:PE-Cy7 (L138, BD Biosciences), CD11b:APC (D12, BD Biosciences), CD16:APC-H7 (3G8, Pharmingen), HLA-DR:V450 (L243, BD Biosciences), CD45:V500 (HI30, BD Biosciences); Tube 3—CD16/57:FITC (NKP15/HNK-1, both BD Biosciences), CD7:PE (M-T701, BD Biosciences), CD4:PerCP-Cy5.5 (SK3, BD Biosciences), CD3:PE-Cy7 (SK7, BD Biosciences), CD56:APC (NCAM16.2, BD Biosciences), CD8:APC-H7 (SK1, BD Biosciences), CD2:V450 (S5.2, BD Biosciences), CD45:V500 (HI30, BD Biosciences); Tube 4—kappa:FITC (TB28–2, BD Biosciences), lambda:PE (1–155-2, BD Biosciences), CD5:PerCP-Cy5.5 (L17F12, BD Biosciences), CD19:PE-Cy7 (J3–119, Beckman Coulter), CD10:APC (HI10a, BD Biosciences), CD38:APC-H7 (HB7, BD Biosciences), CD20:V450 (L27, BD Biosciences), CD45:V500 (HI30, BD Biosciences); Tube 5—CD7:FITC (M-T701, BD Biosciences), CD13/33:PE (L138/P67.6, both BD Biosciences), CD19:PerCP-Cy5.5 (SJ25C1, BD Biosciences), CD56:APC (NCAM16.2, BD Biosciences). For all tubes, 30,000 events were collected. In two cases, fewer tubes were performed due to low cellularity.

Blasts were quantified by flow cytometry using forward and side light scatter properties and/or by expression of immature markers such as CD34 or CD117. Blasts were determined to be qualitatively phenotypically abnormal using a combined leukemia-associated phenotype and different-from-normal approach, with common phenotypic abnormalities including blasts being positive for CD7, CD56, or CD36; negative or abnormally dim positive for HLA-DR; negative for CD33 or CD13; abnormally bright positive for CD123 or CD34; abnormally bright positive or negative for CD117; positive for TdT; or combinations of these abnormalities.

Bone marrow aspirate slides were stained using the Wright-Giemsa method. Blasts were quantified by manual cell counting of at least 100 cells on the aspirate, when possible, but in 30% of the cases fewer cells were counted due to low aspirate cellularity. The bone marrow core biopsy was stained using hematoxylin and eosin and was used to visually estimate bone marrow cellularity. In cases with suboptimal aspirate material, the blast percentage was estimated from the core biopsy and immunohistochemical staining for CD34 and/or CD117.

128 newly diagnosed AML patients (median age 58, IQR 42.5 – 65.5 years) were treated with intensive induction chemotherapy. 18 (14.3%) had favorable-risk cytogenetics, 84 (66.7%) patients had intermediate-risk cytogenetics, and 24 (19%) patients had unfavorable-risk cytogenetics. Table 1 summarizes the characteristics of the included AML patients. 46 (35.9%) patients had NPM1 mutations, and 4 (3.1%) patients had TP53 mutations. 23 (18%) patients had secondary or therapy-related AML.

Table 1.

Characteristics of patients included in the analysis (n= 128). WBC is measured in 10⁹/L, Hgb in g/dL, PLT in 10⁹/L, LDH in IU/L.

	Patients
Age (median, IQR)	58 (42.5 – 65.5)
Sex
Male	64 (50%)
Female	64 (50%)
Cytogenetics
Favorable	18 (14.3%)
Intermediate	84 (66.7%)
Poor	24 (19%)
Secondary or Therapy-related AML
Yes	23 (18%)
No	105 (82%)
WBC (median, IQR)	50.5 (22 – 78.5)
Hgb (median, IQR)	8.8 (7.65 – 10.4)
PLT (median, IQR)	55 (29 – 103)
LDH (median, IQR)	411 (209 – 641.5)
Mutations
NPM1	46 (35.9%)
Double CEBPA	11 (8.6%)
IDH1	10 (7.8%)
IDH2	14 (10.9%)
FLT3-ITD	29 (22.7%)
TP53	4 (3.1%)

From the 128 patients, 115 (89.8%) patients had a day 14 bone marrow biopsy assessment, and 13 (10.2%) patients had a day 21 bone marrow assessment. Combining both groups, 93 (72.7%) patients had <5% blasts in the bone marrow biopsy and 34 (26.6%) patients had 5–20% blasts. One patient had >20% blasts. From all patients, 23 (18%) patients had flow cytometric abnormalities and 105 (82%) patients did not. Flow cytometric abnormalities included: increased side scatter (1 case), abnormally bright positive CD123 (6 cases), negative to abnormally dim positive HLA-DR (8 cases), negative or abnormally dim positive CD33 (2 cases), negative or abnormally bright positive CD117 (5 cases), abnormally bright positive CD34 (4 cases), abnormally bright positive CD36 (1 case), positive CD7 (6 cases), negative CD13 (1 case), positive CD56 (3 cases) and positive TdT (1 case). 116 (90.6%) patients achieved a remission following this bone marrow biopsy without the need of any further therapy. 109 patients achieved a complete remission (CR), 5 patients achieved complete remission with incomplete hematologic recovery (CRi) and 2 patients achieved a morphologic leukemia-free state (MLFS). 12 patients progressed.

From the patients with <5% blasts, 91.4% had no aberrant phenotypic abnormalities versus 58.8% of patients with 5–20% blasts. Table 2 details the positive predictive value (for progression) of blast number and/or aberrant flow. 5–20% blasts in the bone marrow have approximately the same PPV as aberrant flow markers (26.5% vs 26.1%). The PPV of aberrant flow decreases in the subgroup with <5% blasts (12.5%) but increases in the subgroup of 5–20% blasts (35.7%). Still, most patients with 5–20% blasts with aberrant markers achieved a remission.

Table 2.

Positive predictive value (PPV) of morphology or multiparametric flow cytometry (MFC) in predicting progression. Positive MFC was defined as the presence of any aberrant blasts. P-value was derived from a Fisher’s exact test.

	Remission	Progression	PPV	p-value
<5%	90	3	n/a
5–20%	25	9	26.5%
Negative MFC	99	6	n/a
Positive MFC	17	6	26.1%
<5%, negative MFC	83	2	n/a
<5%, positive MFC	7	1	12.5%
5–20%, negative MFC	16	4	20%	0.435
5–20%, positive MFC	9	5	35.7%

In this study, we found that the positive predictive value of aberrant flow cytometric markers on the first post-induction bone marrow is low. Thus, clinicians should refrain from interpreting an abnormal blast immunophenotype at day 14–21 as residual disease that mandates reinduction therapy. Patience and repeat bone marrow examination can spare patients from repeat induction chemotherapy which may lead to prolonged neutropenia, infections and death.⁸ We also have to note that our study joins a large pool of studies showing limited value of day 14 bone marrow biopsy as both morphology and MFC are poor predictors of residual disease.²

Our findings are quite similar to a previous report from the University of Washington that used a highly sensitive MFC assay designed for MRD assessment.⁵ The main reason behind our motivation to perform this analysis was that flow cytometry not as sensitive as the one used in the University of Washington report could have a higher predictive value in predicting progression. In other words, if residual aberrant blasts were identified by a less sensitive method, it could reflect a higher disease burden. A higher disease burden in turn would be more likely to predict progression. However, the results of our study were similar to Chen et al: Their study showed a PPV of 49% for morphology and 46% for MFC, whereas our study showed 26.5% and 26.1% respectively.

Some limitations of this study warrant comment. Our results can only be applied to newly diagnosed AML patients treated with intensive induction chemotherapy. In addition, six patients that had 5–20% blasts and were re-induced immediately were excluded from this analysis. There may have been factors that convinced clinicians that these patients truly had actionable residual disease, and their exclusion from this retrospective study could potentially skew the overall predictive value of MFC at day 14/21. We considered differential counts to be adequate in these frequently hypocellular aspirates when at least 100 cells were counted, but it should be noted that counts of at least 500 cells are generally considered to be optimal in assessment of bone marrow blast percentage, and so the blast percentages in this study (and in day 14/21 marrows generally) may have decreased accuracy. Additionally, our flow cytometric method is unique to our institution and thus the extrapolation of our results to other centers should be made with caution. Furthermore, the bone marrow aspirate is often hemodiluted which can decrease the probability that aberrant blasts will be detected. Finally, our study did not assess the association of aberrant blasts during aplasia and event-free survival. Previous reports have found that aberrant blasts during aplasia are associated with worse event-free survival. ⁶

To summarize, this is the first study to our knowledge that examines the importance of blast immunophenotype on the first post-induction bone marrow biopsy using real-world flow cytometry that is not set up for MRD detection. We showed that most patients with 5–20% blasts in the bone marrow biopsy and an aberrant blast immunophenotype still achieve a complete remission. Thus, clinicians should refrain from interpreting an abnormal immunophenotype at day 14–21 bone marrow biopsy as residual disease that mandates reinduction therapy.