Characterization of children and adolescents with acute lymphoblastic leukemia who presented without peripheral blood blasts at diagnosis

Abstract

Of 1003 children with acute lymphoblastic leukemia (ALL), 147 (14.7%) presented without peripheral blood blasts (PBB). While absence of PBB was not independently associated with survival outcomes when compared to those with PBB, patients without PBB had distinct genetic and clinical characteristics. Notably, we identified a novel genotype to phenotype relationship in that patients without PBB had a significantly higher incidence of hyperdiploid B-ALL, accounting for almost half of all patients without PBB (46.9% vs. 22.7%, P < 0.001). Further, absence of PBB was associated with decreased rates of leukemia involvement of the central nervous system (P < 0.001).

INTRODUCTION

Acute lymphoblastic leukemia (ALL) is the most common pediatric cancer (1). The degree of leukocytosis at presentation is a strong predictor of outcome, with white blood cell (WBC) counts of ≥50 × 10⁹/L being a common cut-off for defining higher risk patients who receive intensified chemotherapy (1, 2). While most patients with ALL present with blasts visible on the differential of a complete blood count (CBC), the characteristics and prognostic implications of ALL patients who present without morphologically detectable peripheral blood blasts (PBB) are not known. This study is the first to evaluate the clinical characteristics, presentation patterns, immunophenotypic and genetic subgroups, and outcomes of patients with ALL who presented without morphologically detectable PBB.

PATIENTS AND METHODS

Patients

With approval of the institutional review board, this retrospective study included children with newly diagnosed ALL who were treated with informed consent on the Total XV (1 to 18 years of age from 2000 to 2007; NCT00137111) (3) or Total XVI (0 to 18 years of age from 2007 to 2017; NCT00549848) (4) protocol at St. Jude Children’s Research Hospital. Clinical information was obtained from medical records and study databases. ALL genetic subtypes were determined as previously described (5). In this study, we defined PBB as being morphologically absent if they constituted <1% of cells in the presenting CBC.

Statistical analysis

Association of PBB percentage (<1% vs. ≥1%) with clinical characteristics was analyzed with Pearson’s chi-squared test. Duration of event-free survival (EFS) was defined as the time to the date of treatment failure for any reason (refractory disease, relapse, death, or second malignancy), and duration of overall survival (OS) was defined as the time to the date of death. Time was censored at the date of last patient contact if no events occurred. EFS and OS distributions were estimated by Kaplan–Meier methods and compared using the log-rank test. The cumulative incidence of refractory disease/relapse was estimated with the Kalbfleisch–Prentice estimator and tested by Gray’s method. Multivariable analysis was performed using the Wald Chi-Squared test.

RESULTS

Patients

Of 1006 patients treated on Total XV (n = 408) or Total XVI (n = 598), three started treatment at a referring center and their diagnostic CBCs lacked a differential. For the remaining 1003 patients, the median WBC count was 10.7 × 10⁹/L (range: 0.2 to 1014.0 × 10⁹/L) and the median blast percentage was 42.0% (range: 0.0% to 99.0%) (Table S1). In 147 of the 1003 patients (14.7%), blasts were less than 1% in the peripheral blood at diagnosis and classified as those without PBB.

Diagnostic features in patients without PBB

Compared with patients with PBB at diagnosis, those without PBB had lower presenting WBC counts (P < 0.001) (only one patient with a WBC count exceeding 50.0 × 10⁹/L), B-ALL rather than T-ALL (P < 0.001), and central nervous system (CNS) 1 status (P = 0.001) (Table 1). However, four of the 147 patients (2.7%) without PBB had CNS3 disease (three patients with B-ALL, one with T-ALL).

TABLE 1.

Patient characteristics based on peripheral blood blast percentage

Variables	All n = 1003 (%)	Peripheral blood blast %		P value
		<1% n = 147 (%)	≥1% n = 856 (%)
Age at diagnosis, n (%)				0.133
<1 year	12 (1.2)	0 (0.0)	12 (1.4)
1–9.9 years	722 (72.0)	114 (77.6)	608 (71.0)
≥10 years	269 (26.8)	33 (22.4)	236 (27.6)
Sex, n (%)				0.701
Male	574 (57.2)	82 (55.8)	492 (57.5)
Female	429 (42.8)	65 (44.2)	364 (42.5)
Ancestry, n (%)				0.368
European	756 (75.4)	115 (78.2)	641 (74.9)
African	150 (15.0)	18 (12.2)	132 (15.4)
Native American	26 (2.6)	6 (4.1)	20 (2.3)
Other	71 (7.1)	8 (5.4)	63 (7.4)
Lineage, n (%)				<0.001
B cell	836 (83.3)	138 (93.9)	698 (81.5)
T cell	167 (16.7)	9 (6.1)	158 (18.5)
WBC count at diagnosis, n (%)				<0.001
<50 × 109/L	757 (75.5)	146 (99.3)	611 (71.4)
≥50 × 109/L	246 (24.5)	1 (0.7)	245 (28.6)
WBC count at diagnosis (× 109/L)				<0.001 *
Median (min, max)	13.0 (0.2, 1014.0)	2.80 (0.2, 65.0)	17.5 (0.6, 1014.0)
CNS status, n (%)				0.001
CNS1	647 (64.5)	115 (78.2)	532 (62.1)
CNS2	267 (26.6)	26 (17.7)	241 (28.2)
CNS3	32 (3.2)	4 (2.7)	28 (3.3)
TLP with blasts	56 (5.6)	2 (1.4)	54 (6.3)
Not available	1 (0.1)	0 (0.0)	1 (0.1)
Protocol, n (%)				0.181
Total XV	407 (40.6)	67 (45.6)	340 (39.7)
Total XVI	596 (59.4)	80 (54.4)	516 (60.3)
Genetic subtype, n (%)				<0.001
Hyperdiploid	263 (26.2)	69 (46.9)	194 (22.7)	(<0.001)†
ETV6-RUNX1	205 (20.4)	26 (17.7)	179 (20.9)	(0.438)
T-ALL	147 (14.7)	8 (5.4)	139 (16.2)	(<0.001)
PAX5alt	44 (4.4)	1 (0.7)	43 (5.0)	(0.014)
TCF3-PBX1	39 (3.9)	2 (1.4)	37 (4.3)	(0.105)
KMT2A rearranged	36 (3.6)	2 (1.4)	34 (4.0)	(0.149)
DUX4 rearranged	31 (3.1)	3 (2.0)	28 (3.3)	(0.607)
BCR-ABL1–like	27 (2.7)	3 (2.0)	24 (2.8)	(0.786)
BCR-ABL1	23 (2.3)	2 (1.4)	21 (2.5)	(0.560)
ETP	20 (2.0)	1 (0.7)	19 (2.2)	(0.340)
Hypodiploid	16 (1.6)	4 (2.7)	12 (1.4)	(0.274)
ETV6-RUNX1–like	12 (1.2)	0 (0.0)	12 (1.4)	(0.232)
iAMP21	12 (1.2)	2 (1.4)	10 (1.2)	(0.692)
Other	128 (12.8)	24 (16.3)	104 (12.1)	(0.180)
MRD on day15/19 # , n (%)				<0.001
<1%	731 (72.9)	123 (83.7)	608 (71.0)
≥1%	258 (25.7)	19 (12.9)	239 (27.9)
Not available	14 (1.4)	5 (3.4)	9 (1.1)
MRD at end of induction, n (%)				0.046
<0.01%	840 (83.7)	131 (89.1)	709 (82.8)
≥0.01%	150 (15.0)	14 (9.5)	136 (15.9)
Not available	13 (1.3)	2 (1.4)	11 (1.3)
Treatment risk, n (%)				<0.001
Low	458 (45.7)	97 (66.0)	361 (42.2)
Standard/high	545 (54.3)	50 (34.0)	495 (57.8)
Outcome (5-year), % (standard error)
Event-free survival	88.3 (1.1)	93.5 (2.2)	87.4 (1.2)	0.015
Overall survival	93.3 (0.8)	97.9 (1.2)	92.5 (1.0)	0.035
CIN	8.1 (0.9)	5.1 (1.9)	8.6 (1.0)	0.116

Abbreviations: WBC, white blood cell; CNS, central nervous system; TLP, traumatic lumbar puncture; ALL, acute lymphoblastic leukemia; ETP, early T-cell precursor; iAMP21, intrachromosomal amplification of chromosome 21; MRD, minimal residual disease; CIN, cumulative incidence of refractory disease/relapse

^*t-test, all other analyses were performed with Pearson’s chi-squared test.

^#MRD was tested on day 15 in Total XVI and on day 19 in Total XV. The “Not available” category is excluded in the test.

^†P-values in parenthesis are derived from 2 x 2 table (<1% PBB, ≥1% PBB vs. the subtype, not the subtype) by the Fisher’s exact test for the incidence of each leukemia subtype.

With regard to genetic subgroups, patients without PBB had a significantly higher incidence of hyperdiploid ALL (46.9% vs. 22.7%, P < 0.001) and lower incidence of T-ALL (5.4% vs. 16.2%, P <0.001) and PAX5alt (0.7% vs. 5.0%, P = 0.014), but a similar incidence of ETV6-RUNX1 rearrangement (17.7% vs. 20.9%, P = 0.438) when compared to patients with PBB (Table 1, Figure 1). Infrequently, high-risk genetic subtypes including hypodiploid (n = 4), BCR-ABL1–like (n = 3), BCR-ABL1 (n = 2), KMT2A-rearranged (n = 2), and iAMP21 (n = 2) were seen among patients presenting without PBB.

Figure 1: ALL Genotype distribution based on Peripheral Blood Blasts.

In this study, 856 ALL patients had ≥1% blasts on a peripheral blood smear at diagnosis, while 147 patients had <1% blasts. Between these two groups, the distribution of ALL genotypes was significantly different (P < 0.001). Among the 147 patients with <1% peripheral blood blasts, 138 had B-ALL and almost half (46.9%, 69/147) had hyperdiploid B-ALL, which is a significantly greater proportion when compared to patients with ≥1% peripheral blood blasts (22.7%, 194/856; P < 0.001). T-ALL and PAX5alt were significantly lower for patients with <1% peripheral blood blasts (P < 0.001 and P = 0.014, respectively). (Not all genotypes displayed, see Table 1 for complete details.)

Presenting symptoms and CBC findings in patients without PBB

In the 147 patients without PBB, common symptoms at presentation were fever (n = 96, 65.3% of patients), fatigue (n = 72, 49.0%), lymphadenopathy (n = 69, 46.9%), bone/joint pain (n = 51, 34.7%), petechiae/bruising/bleeding (n = 34, 23.1%), and pallor (n = 26, 17.7%) (Figure S1), with most patients having two or more of these signs and symptoms (n = 113, 76.9%).

Regarding the CBC findings (Table S2), 129 of the 147 patients (87.8%) had decreased counts for at least one cell type: 112 (76.2%) had neutropenia (absolute neutrophil count < 1.5 × 10⁹/L); 111 (75.5%) had anemia (hemoglobin < 11.0 g/dL); and 87 (59.2%) had thrombocytopenia (platelet counts < 150 × 10⁹/L). At least two cell types were affected in 114 patients (77.5%), and all three cell types were decreased in 67 patients (45.6%). Isolated anemia (n = 8, 5.4%), isolated neutropenia (n = 4, 2.7%), and isolated thrombocytopenia (n = 3, 2.0%) were uncommon events. Of the three patients with isolated thrombocytopenia, one had lymphadenopathy and fever whereas the other two had petechiae/bruising/bleeding and bone pain.

Of the 18 patients (12.2%) who had a normal CBC at diagnosis, six had isolated lymphadenopathy, four had a mediastinal mass (with three also having lymphadenopathy), four had bone/joint pain (with three also having fever), two had both lymphadenopathy and bone pain, one had a testicular mass, and one had lymphadenopathy plus hepatosplenomegaly.

Treatment outcomes in patients without PBB

Patients without PBB were less likely than were those with PBB to have minimal residual disease (MRD) ≥ 1% on day 15 (in Total XVI) or day 19 (in Total XV) (12.9% vs. 27.9%, P < 0.001) or to have MRD ≥ 0.01% at the end of induction (9.5% vs. 15.9%, P = 0.046) and have higher percentage of low-treatment risk (P < 0.001) (Table 1). Among patients who lacked PBB at diagnosis, ten experienced events (details in Table S3). When compared to patients with PBB, patients without PBB had a better 5-year EFS (93.5% ± 2.2% vs. 87.4% ± 1.2%, P = 0.015) and OS (97.9% ± 1.2% vs. 92.5% ± 1.0%, P = 0.035) (Table 1 and Figure S2). However, multivariable analysis that used ALL genotype, treatment risk, and PBB as variables found no statistically significant survival benefit for absence of PBB (P = 0.202 for EFS and P = 0.211 for OS) and, among hyperdiploid B-ALL patients, PBB was not prognostic when adjusted for treatment risk (P = 0.581 for EFS and P = 0.893 for OS) (Table S4). Furthermore, there was no statistically significant difference in the cumulative incidence of refractory disease or relapse (5.1% ± 1.9% vs 8.6% ± 1.0%, P = 0.116) (Figure S2)

DISCUSSION

In this study, 147 of 1003 pediatric patients with ALL (14.7%) presented with less than 1% of PBB. Among these 147 patients, the absence of PBB did not independently affect outcomes. However, we found that patients without PBB had distinct genetic and clinical characteristics, including associations with B-ALL, hyperdiploidy, and CNS1 status. Moreover, we described the pattern of symptoms and cytopenias observed in patients without PBB to serve as a guide for clinicians evaluating patients with suspected leukemia without PBB.

In patients without PBB, presenting symptoms were similar to those in a meta-analysis of pediatric patients with newly diagnosed ALL, most of whom presumably had circulating blasts (6). While clinical judgement is required to know when to obtain a CBC, we found that typically at least two of the three cell lines were decreased in the CBC. Involvement of only a single cell line (or even a normal CBC) at diagnosis is uncommon, and these patients typically have additional suggestive signs and symptoms.

Occasionally, clinicians have difficulty differentiating immune thrombocytopenic purpura (ITP) from ALL in patients with isolated thrombocytopenia. The American Society of Hematology currently recommends that steroids should be used as first-line treatment without a preceding bone marrow examination for pediatric patients who present with typical features of ITP and require treatment (7). However, the impact of steroid pretreatment on ALL remains a concern (8). We found that it is rare for patients with ALL to present with isolated thrombocytopenia without PBB, occurring in only three of the 1003 study patients (0.3%). However, these patients had additional symptoms such as bone pain or fever and lymphadenopathy. Thus, a thorough history and examination when evaluating presumed ITP can help clinicians avoid potential misdiagnosis for the rare cases of ALL that present with isolated thrombocytopenia on the CBC.

There was no significant difference in age distribution at diagnosis based on PBB, which is noteworthy as age is an important prognostic marker in B-ALL (2). It is tempting to attribute the decreased CNS involvement directly to the absence of PBB. However a report identified CNS invasion occurring secondary to leukemic blasts transiting from the vertebral or calvarial bone marrow directly into the subarachnoid space along the outside of bridging vessels, and not from the bloodstream (9), which may explain why four patients without PBB had CNS3 disease.

In this study, we identified a novel genotype-to-phenotype correlation wherein a significantly higher proportion of patients without PBB had hyperdiploid B-ALL. In contrast, the proportion of patients with another favorable genotype, ETV6-RUNX1, did not differ according to whether the patients had PBB. However, patients without PBB should not be assumed to have favorable genotypes, as other unfavorable ALL genetic subtypes, such as hypodiploid, BCR-ABL1, BCR-ABL1–like, KMT2A- rearranged, iAMP21, and T-ALL, were observed in some patients without PBB.

Although patients without PBB had lower MRD levels during induction than did those with PBB, EFS and OS were not significant in multivariable analysis and there was no significant difference in the incidences of refractory disease/relapse. Therefore, it is important to use a comprehensive risk classification that considers traditional clinical factors (e.g., patient age and WBC counts), genetic subtype, and treatment response (e.g., MRD) when devising an individualized therapeutic plan, regardless of the percentage of PBB at diagnosis (5).