The Impact of Bone Marrow Hematogones On Umbilical Cord Blood Transplant Outcomes in Acute Myeloid Leukemia Patients

Abstract

Early after umbilical cord blood transplant (UCBT), patients show marked differences in bone marrow (BM) hematogone percentages. Little is known about whether these differences are clinically relevant. We hypothesized that early recovery of hematogones may be associated with improved transplant outcomes. BM aspirates were assessed from 88 patients with acute myeloid leukemia (AML) by two independent reviewers at Day 21 and 100 after UCBT. Inter-observer variability for BM hematogone percentages at these time points showed correlation coefficients of 0.83 and 0.98, respectively (p=<0.01 for both). A high percentage of hematogones at D21 was associated less acute graft vs. host disease (GVHD) grade 3-4 (p=0.01). At day 100, a high percentage of BM hematogones was associated with improved overall survival (p=0.02) and lower treatment related mortality (p=<0.01). This study shows that BM hematogone percentages may be useful prognostic indicators in AML patients following UCBT and should be routinely reported in BM differential counts.

INTRODUCTION

Hematogones were first described in the 1930’s as lymphoid-appearing cells present in the sternal aspirates of children undergoing evaluation for an assortment of neoplastic, congenital, infectious and autoimmune diseases. (1, 2) Although their biological significance was unclear, authors speculated on their role in hematopoiesis, with some considering hematogones to be undifferentiated primordial cells (“hemocytoblasts”) (1-3). Eventually the name hematogone, from hematogonia (“blood-maker” in Latin), was adopted. (1, 2) Since that time, numerous studies document that they represent B-lymphocyte precursors which reside in the marrow and that undergo an orderly maturation sequence to give rise to mature B cells.(4-10) By flow cytometry hematogones have been characterized into several stages using CD10, CD19, CD20, CD22, CD34 and CD38 (9). Although this method is very effective in characterizing the maturational spectrum of hematogones, the wide variation in flow cytometry protocols limits the determination of cell percentages, therefore assessment by morphology is standard for all current classification systems (11).

Hematogones may be increased in numerous neoplastic as well as non-neoplastic conditions including marrow regeneration and immune deficiencies. (4, 6-10, 12-21) One situation where increased percentages of hematogones are commonly observed is in the post-umbilical cord blood (UCB) transplant setting (17, 22, 23). UCB has been shown to have high numbers and increased generative capacity of B lymphocyte progenitors (24). For unknown reasons, elevated hematogone percentages are noted in some, but not other UCB transplant patients. Whether there is a prognostic association between the presence of BM hematogones and clinical outcomes in UCBT is unknown. This lack of information may explain why hematogones are not routinely reported in BM differential counts and are, instead, combined with other lymphocytes.

To date, no study has assessed the reproducibility of morphologic detection of hematogones in the BM of transplant recipients or whether hematogone percentages are associated with post-transplant outcomes. Considering that hematogones give rise to B cells that provide protection against pathogens and that in non-transplant patients, the percentage of hematogones in the BM is inversely correlated with the percentage of leukemic blasts in the marrow (9, 25), we hypothesized that increased hematogones would be associated with superior transplant outcomes.

To address this hypothesis, we assessed the percentage of BM hematogones in a cohort of 91 consecutive AML patients treated with allogeneic single or double umbilical cord blood transplant (UCBT). Recognizing that any associations between hematogones and clinical outcomes are only useful if they can be reproducibly recognized, we also determined the correlation between two independent observers in the assessment of the percentage hematogones in the marrow aspirates at D+21 and 100 after transplantation.

METHODS

Patients

Ninety one consecutive patients with AML undergoing myeloablative UCB transplantation were studied. We included patients who had undergone UCB transplantation at the University of Minnesota between 02/1999 and 07/2008. Patients were required to have a bone marrow biopsy with no evidence of relapse at the time of the marrow analysis. Hematogone determination were performed on day 21 and 100 post-UCB transplantation (i.e., D21 or 100) samples. Patients who relapsed between D21 and D100 were censored for the D100 hematogone determination. Because of the morphological similarity between hematogones and leukemic lymphoblasts, only patients with AML were included in this analysis. Of 91 eligible AML patients receiving UCBT during the study period, 3 were excluded due to early post-transplant relapse (n=2) or lack of availability of any interpretable slides for review (n=1). A summary of exclusion criteria is provided in Supplemental Table 1. Of the 88 remaining patients, 85 had interpretable slides available for review at day 21. Sixty six cases were available for review at both D21 and D100, while 19 cases were reviewed at only D21 and 3 cases were reviewed only at D100. Overall, 154 cases from 88 patients (88/91) fit the above criteria and had material adequate for differential counts.

Morphologic Analysis

Wright-Giemsa stained direct and concentrate bone marrow aspirate smears and trephine touch imprint preparations were each reviewed independently by a faculty hematopathologist (VJD) and pathology resident (TJH). The specimens used to perform differential counts were, in order of preference, direct smears, touch preparations, and concentrate smears. In general, touch preparations were used in cases in which aspirates were hemodilute or not available. Concentrate smears were used in cases with hypocellular aspirates. Prior to the beginning of the study, both reviewers concurrently examined several cases noted to have high percentages of hematogones and came to agreement on morphologic features useful in distinguishing hematogones from mature lymphocytes. A five hundred cell differential count, including hematogones, was performed in each case, except in instances where low cellularity precluded a full 500 cell count. In those cases, a 50-300 cell differential count was performed. A 50 cell differential count was performed in one case only, on a concentrate smear. Hematogones had condensed and homogenous nuclear chromatin without significant clumping, high nuclear to cytoplasmic (N/C) ratios with a thin rim of eccentric basophilic cytoplasm, and no cytoplasmic vacuoles, inclusions or granules. When present, nucleoli were indistinct. In contrast, mature lymphocytes had coarse, clumped chromatin, lower N/C ratios and more abundant cytoplasm with variable cytoplasmic inclusions and granules. (Figure 1)

Figure 1. Morphological appearance of hematogones (A) and lymphocytes (B).

A) Hematogones had condensed and homogenous nuclear chromatin without significant clumping, high nuclear to cytoplasmic (N/C) ratios with a thin rim of eccentric basophilic cytoplasm, no cytoplasmic vacuoles, inclusions or granules and no distinct nucleoli (open arrow). B) In contrast to the hematogones pictured opposite, lymphocytes had coarse, clumped chromatin and more abundant cytoplasm (closed arrow).

Interobserver Variability in the Morphological Assignment of Hematogones

In 146/154 cases (94.8%) the hematogone percentages reported by the two observers were within 3.6%. The remaining 8 cases (5.2%) were jointly reviewed and a consensus was reached after which interobserver variability for percent hematogones reported at days 21 and 100 was calculated using Pearson’s correlation. Correlation coefficients at days 21 and 100 were 0.83 and 0.98, respectively (p=<0.01 for both) (26). For the below analysis correlating hematogone percentage with transplant outcomes, the average percentage from the two reviewers was used.

Flow cytometric immunophenotyping

The clinical laboratory database for all patients with slides available for microscopic review was searched in order to determine whether flow cytometric immunophenotyping was performed on the marrow aspirate and whether a percentage value for hematogones was reported at the time of the original study. Flow cytometry was performed on bone marrow aspirates that were collected in sodium heparin. After lysis of erythrocytes (ammonium chloride), the cells were washed twice with phosphate buffered saline and stained according to manufacturers’ recommendations with a panel of monoclonal antibodies directly conjugated to fluorochromes (B D Biosciences, San Jose, CA). Data were acquired on a FACSCaliber (BD) using CELL Quest Pro and analyzed using FACS Diva software.

Assessment of Patient Absolute Lymphocyte Count (ALC)

The medical records were reviewed for patients and absolute lymphocyte counts were obtained as previously described (27). Briefly, ALCs were collected from the medical record at D30 (+3 days). ALC were recorded as missing if the WBC was <0.5 ×10⁹/L as no manual differential counts were performed during severe leukopenia. Correlation between average percentage hematogone and ALC was performed using chi square analysis.

Statistical Analysis

Data on transplant patient characteristics, post-transplantation complications and outcomes were prospectively collected by the Biostatistical Support Group at the University of Minnesota using standardized collection procedures. Details regarding graft selection, conditioning regimens and GVHD prophylaxis have been previously described. (28-30) Patients and disease characteristics were summarized using descriptive statistics.

Statistical comparisons of these variables between high and low hematogone percentage were completed by a nonparametric Wilcoxon test for continuous factors and Pearson’s chi-square test for categorical factors. All patients were followed longitudinally until death or last follow up. The endpoints included neutrophil and platelet recovery, overall survival (OS), disease free survival (DFS), transplant related mortality (TRM), risk of relapse, acute GVHD (aGVHD), and chronic GVHD (cGVHD). Kaplan-Meier (31) was used to estimate neutrophil recovery, OS and DFS. Cumulative incidence (32) was used to estimate platelet recovery, relapse, TRM, aGVHD, and cGVHD. Statistical comparison of neutrophil, overall survival and disease free survival between groups was completed by the Log-Rank test. The proportional hazards model of Fine and Gray (33) was used to assess the independent factors on platelet recovery, relapse, TRM, aGVHD, and cGVHD. Factors included in multivariate analysis were: hematogone percentage (low vs. high), donor number (single vs. double), recipient age (<21 vs. ≥ 21), recipient CMV status (negative vs. positive), total, post-thaw CD34 (<0.50 vs. ≥0.50), total, post-thaw CFU (<0.042 vs. ≥ 0.042), and total nucleated cell (TNC) (<0.38 vs. ≥0.38). Backwards stepwise method was used to decide the final model. Groups with value of p ≤ 0.05 were considered to be statistically different.

RESULTS

Percentage Marrow Hematogones and Transplant Outcomes

Patients were grouped into quartiles based on the average percentage of BM hematogones (determined using morphology by two reviewers) at D21 and D100 after transplantation. In univariate analysis, those in the lowest quartile had differing outcomes compared to patients in the upper three quartiles (not shown). Based on these findings, patients were grouped using the twenty-fifth percentile (0% at 21 days after UCT and 0.9% at 100 days after UCT) as a cutoff. There were no demographic differences between patients who had a low or high percentage of hematogones at D21 (supplemental table 2) or Day 100 (table 1)

Table 1.

Patient demographics based on percentage of hematogones in the marrow at day 100 after UCBT.

Factor	n (%)	Hematogones at day 100 n (%)
		= 0.9	>0.9	p-value
Number of UCB units				.20
Single	21 (32%)	3 (5%)	18 (27%)
Double	45 (68%)	13 (20%)	32 (48%)
Gender				.55
Male	37 (56%)	10 (15%)	27 (41%)
CMV status				.53
Positive	41 (62%)	11 (17%)	30 (45%)
HLA (engrafting unit)				.96
4/6	29 (44%)	7 (11%)	22 (33%)
5/6	28 (42%)	6 (10%)	22 (33%)
6/6	8 (12%)	2 (3%)	6 (10%)
Disease status				.60
CR1	27 (41%)	8 (12%)	19 (29%)
CR2	36 (54%)	7 (11%)	29 (44%)
CR3	3 (5%)	1 (2%)	2 (3%)
Conditioning
MA	66 (100%)	16 (24%)	50 (76%)	--
Total nucleated dose (×107/kg)	0.41 (0.14-2.80)	0.42 (0.23-1.47)	0.40 (0.14-2.80)	.64
CD3 cell dose/kg (×106/kg)	0.14 (0.04-0.56)	0.13 (0.04-0.51)	0.15 (0.04-0.56)	.82
CD34 cell dose/kg (×106/kg)	0.50 (0.13-27.53)	0.46 (0.14-1.59)	0.50 (0.13-27.53)	.54
Recipient Age	21 (0-44)	22 (2-44)	21 (0-44)	.66

HLA = Human leukocyte antigen; values correspond to number of matched allelic loci for HLA-A, HLA-B and HLA-DR between recipient and engrafting donor. CMV = cytomegalovirus. CRN = In Nth complete remission at time of transplant. MA = myeloablative.

At day +21 after transplant the percentage of marrow hematogones varied from 0 to 10.8% (n=91). In assessing these percentages, we observed that patients in the lowest quartile (corresponding to those with hematogones=0%) were more likely to develop grade III-IV aGVHD (38% [95% CI = 20 – 57%] vs. 22% [95% CI = 11 – 32%] (p=0.05)). This association was confirmed in multivariate analysis where patients with hematogones in the lowest quartile were more likely to develop severe aGVHD (p=0.01) (table 2). Based on multivariate analysis, hematogone percentages at day 21 were not associated with OS, DFS, TRM, neutrophil engraftment, platelet engraftment or cGVHD (table 2). On multivariate analysis patients in the upper three quartiles (with hematogone percentages >0) had a relative risk of relapse of 8.48 (95% CI = 0.82 - 88.06) compared to those with hematogones=0 (p = 0.07) (table 2). Patients who had a high bone marrow hematogone percentage at D21 tended to have a high percentage at D100 as well. More specifically, of the 66 patients who had hematogone analysis at both D21 and D100, 47 patients at D21 had a hematogone percentage of >0. Forty of these (85%) remained in the high hematogone group at D100 (p=0.01).

Table 2.

Multivariate analysis examining the percentage hematogones detected in the BM at day 21 after UCBT.

Outcome	Variable	RR (95% CI)	p value
DFS	Donor number		0.04
	1	1
	2	0.13 (0.02-0.90)
Relapse	Hem at Day 21
	=0	1.00
	>0	8.48 (0.82-88.06)	0.07
	Donor number
	1	1.00
	2	0.33 (0.12-0.95)	0.04
ANC	Hem at Day 21
	=0	1.00
	>0	0.95 (0.43-2.09)	0.89
	CD34
	<0.50	1.00
	>=0.50	3.70 (1.29-10.62)	0.01
TRM	Hem at Day 21
	=0	1.00
	>0	0.45 (0.15-1.33)	0.15
aGVHD III-IV	Hem at Day 21
	=0	1.00
	>0	0.30 (0.15-0.59)	0.01
	Donor number
	1	1.00
	2	5.44 (1.41-20.92)	0.01
	Receiver CMV
	Negative	1.00
	Positive	5.97 (2.03-17.58)	<0.01
	CFU
	<0.0419	1.00
	>=0.0419	0.28 (0.10-0.74)	0.01
CGVHD	Hem at Day 21
	=0	1.00
	>0	1.90 (0.47-7.74)	0.37

RR = Relative risk. ANC = Absolute neutrophil count. Hem = Percentage hematogones. CD34 = CD34 cell dose of transplanted cord blood unit(s). TRM = Transplant-related mortality. aGVHD III-IV = acute graft vs. host disease, grades 3-4. CMV = cytomegalovirus. CFU = colony forming units. CGVHD = chronic graft vs. host disease.

At day +100 after transplant, the percentage of hematogones varied from 0 to 29.6% (n=66). At this time the percentage of hematogones was more strongly associated with transplant outcomes. In univariate analysis, patients with high hematogone percentages (i.e., upper three quartiles (>= 0.9%)) had a 3 year OS of 76% (95% CI = 61 – 85%) compared to 49% (95% CI = 61 - 85%) for those with hematogone percentages in the lowest quartile (<0.9%) (p=0.02) (Figure 2A). As shown in table 3, this was confirmed in multivariate analysis, where patients with a high percentage of hematogones in the marrow were 5-fold more likely to survive (RR=0.20 (95% CI = 0.05 - 0.76), p=0.02). This improvement in OS was due to a reduction in TRM. In univariate analysis, patients with a high hematogone percentage (>= 0.9%) at day 100 showed a 1 year TRM rate of 2% (95% CI = 0 – 6%) compared to 31% (95% CI = 9 – 54%) (p=<0.01) (Figure 2B). This was again confirmed in multivariate analysis (table 3), where increased hematogone percentages (>= 0.9%) at day 100 were associated with a relative risk of TRM of 0.03 (95% CI = 0 – 0.34), p=<0.01). Hematogone percentages at day 100 were not associated with DFS, relapse, neutrophil engraftment, platelet engraftment, or cGVHD on multivariate analysis (table 3).

Figure 2. BM hematogones at day 100 are associated with transplant outcomes.

A) Probably of 1 year OS based on the percentage of hematogones in the BM at day 100 after transplantation. B) Cumulative incidence of TRM based on the percentage of hematogones in the BM at day 100 after transplantation.

Table 3.

Multivariate analysis examining the percentage hematogones detected in the BM at day 100 after UCBT.

Outcome	Variable	RR (95% CI)	p value
Overall Survival	Hem at Day 100
	<0.9	1.00
	>=0.9	0.20 (0.05-0.76)	0.02
	Donor number
	1	1.00
	2	0.19 (0.04-0.84)	0.03
	Transplant Age
	<20.5	1.00
	>=20.5	4.89 (1.11-21.51)	0.04
Relapse	Hem at Day 100
	<0.9	1.00
	>=0.9	2.84 (0.32-24.91)	0.35
TRM	Hem at Day 100
	<0.9	1.00
	>=0.9	0.03 (0-0.34)	<0.01
	HLA (engrafting unit)
	4/6	1.00
	5/6	0.61 (0.05-7.13)	0.69
	6/6	0	<0.01
CGVHD	Hem at Day 100
	<0.9	1.00
	>=0.9	0.44 (0.11-1.69)	0.23
	Transplant Age
	<20.5	1.00
	>=20.5	3.61 (1.18-11.06)	0.02

RR = relative risk. Hem = percentage hematogones. TRM = transplant-related mortality. HLA = human leukocyte antigen; values correspond to number of matched allelic loci for HLA-A, HLA-B and HLA-DR between recipient and engrafting donor. CGVHD = chronic graft vs. host disease.

Morphologic Correlation with FACS Assessment of Hematagones

Review of the laboratory database showed that prior to 2007, flow cytometric immunophenotyping was rarely performed on bone marrow aspirates of post-UCB transplant AML patients at our center. Flow cytometric immunophenotyping included a percentage hematogones in the final report for 11 of the flow cytometry cases. As shown in figure 3, hematogone determination by flow cytometry was correlated with that determined by morphology (p=0.05) in the limited cases where data was available.

Figure 3. Correlation of bone marrow hematagones with flow cytometry.

Shown is the correlation for the 11 cases that had both morphological determination of hematogone percentage by morphology and flow cytometry where percentage of hematogones were reported.

Marrow Hematogone Percentage is Not Associated with Absolute Lymphoycte Count

We have recently determined that rapid lymphocyte recovery after UCB transplant is associated with improved transplant outcomes (27). In that study, patients treated with myeloablative transplant showed superior outcomes (OS, PFS and TRM) if their absolute lymphocyte count (ALC) was >200 ×10⁶/L at D+30. We hypothesized that ALC and percentage marrow hematogones would be correlated at D21 and/or D100. As shown in table 4, ALC at D30 was available in the medical record for 61 patients at D21 and 56 patients at day 100. Surprisingly, there was no correlation between ALC and hematogone percentage in the marrow at either D21 or D100. Thus, hematogone percentage is independent of ALC (p=0.86 and 0.12, respectively), but both are associated with UCB transplant outcomes (above and (27)).

Table 4.

Association of D21 and 100 hematogone count with ALC

	HemD21 = 0	HemD21>0	p-value
ALC D30< 200	5 (27%)	13 (72%)	.86
ALC D30 >=200	11 (26%)	32 (74%)
	HemD100 = 0.9	HemD100>=0.9	p-value
ALC D30< 200	5 (33%)	10 (67%)	.12
ALC D30 >=200	6 (15%)	35 (85%)

DISCUSSION

We investigated whether the percentage of marrow hematogones as determined by morphology was associated with UCB transplant outcomes in a population of uniformly treated adults and children with AML. Our results show that, increased percentages of hematogones in the BM at day 21 were associated with lower rates of grade III-IV aGVHD, likely suggesting that severe aGVHD interferes with B cell reconstitution. At day 100, high percentages of hematogones were associated with improved 3 year OS, due to lower 1 year TRM. While recent studies demonstrate a relationship between the presence of hematogones detected by flow cytometry and improved OS in AML patients receiving chemotherapy, multivariate analysis could not confirm those findings. (34) Our study demonstrates that increased numbers of hematogones are an independent predictor of OS and TRM in a subset of AML patients undergoing UCB transplantation. We specifically chose AML patients for this study, due to the concern for morphologic similarity between hematogones and lymphoblasts. Thus, additional studies are needed to investigate whether increased marrow hematogones are associated with similar outcomes in other diseases, including acute lymphoblastic leukemia or lymphoma.

In this study only a limited number of cases were reviewed by flow cytometry considering that BM flow cytometry for AML was not standard of care at the time of the study. However, of the limited cases, there was adequate correlation between morphology and flow cytometry. Based on this, and because the relative quantification of marrow constituents using flow cytometry is problematic, we chose to assess hematogones using morphology in this study. Current concerns for using flow cytometry include that some cell types, such as erythroid precursors, are lost when a lysis step is incorporated. This would artificially increase hematogone percentages. Likewise, various gating strategies may include cells which do not represent hematogones, or alternatively may exclude true hematogones. This would artificially increase or decrease the calculated hematogone percentage, respectively. In addition, samples submitted for flow cytometry often represent the second aspirated specimen and are hemodilute, artificially lowering hematogone percentages. Thus, we chose to perform morphologic assessment as the “gold standard” for quantification of marrow hematogones. As flow cytometry becomes more standardized across institutions, future studies should be performed to determine whether this modality is also associated with transplant outcomes.

While we provide evidence that marrow hematogones are associated with UCB transplant outcomes, there are limitations to this study. First, this is a single institution study which may introduce not only patient bias, but also subjectivity in the assessed morphologic parameters. Another potential limitation relates to specimen preparation. All direct smears were prepared at the bedside during the marrow aspirate procedure by hematology technicians and were stained by hand. This method, in our experience, facilitates a more accurate assessment of the bone marrow cellular composition. In contrast, anticoagulated specimens processed by automated machine stainers may preclude accurate assessment of the cellular constituents, limiting the applicability of these findings.

One of the largest causes of mortality in patients undergoing UCBT is infections, predominantly during the lymphopenic period. (35) Although we demonstrate that increased percentages of marrow hematogones correlate well with clinical outcomes, it is not clear that their increase predicts immunity or that immunity from infection is the explanation for improved outcomes. In fact in a subset of patients, there was no correlation of BM hematogone percentages with serum immunoglobulin levels at 100 or 365 days after transplantation (not shown). Likewise, detailed B cell immunological reconstitution was not routinely performed on this cohort and so how this data is related to mature B cell recovery will be the subject of additional studies.

In summary, AML patients undergoing UCBT with increased BM hematogones have less acute GVHD, less TRM, and improved OS. Additional prospective studies are needed in other diseases since these results could become a useful tool to predict post-transplant outcomes.