Differential expression of hERG1A and hERG1B genes in pediatric acute lymphoblastic leukemia identifies different prognostic subgroups.

Acute lymphoblastic leukemia (ALL) is the most common malignancy of childhood, with 85% of cases being of B-cell lineage (B-cell precursor (BCP)-ALL) and 15% of T-cell lineage (T-ALL).[1] With wider use of intensive chemotherapy, the prognosis of childhood ALL has improved remarkably, and nearly 80% of BCP-ALL[2] patients can currently be cured. The prognosis of children with T-ALL has improved and has been reported to be similar to that for BCP-ALL (no differences in the 5-year event-free survival (EFS) rate).[3] However, long-term survival rates for pediatric T-ALL are still lower than those for BCP-ALL by up to 20%.[3] Systemic toxicity and chemoresistance are nowadays the main shortcomings of standard chemotherapy.[2]

Current interest focuses on identifying new specific molecular targets to be exploited either for risk stratification or for identification of novel, patient-tailored, therapeutic approaches that can improve therapy efficacy and reduce toxicity in pediatric ALL.

We have provided evidence that K+ channels encoded by the ether-a-gò-gò-related gene 1 (hERG1), hERG1 channels, besides exploiting a relevant role in cardiac physiology,[4] are often aberrantly expressed in human cancers including leukemias.[5, 6] In pediatric BCP-ALL, hERG1 channels sustain the development of chemoresistance,[7] as they modulate pro-survival signals triggered by the bone marrow microenvironment. In adult acute myeloid leukemias (AML), hERG1 regulates cell motility and transendothelial migration through an interplay with angiogenic signaling pathways. This effectively correlates with the worse prognosis in AML patients displaying high hERG1 expression.[5]

The hERG1 encoding gene shows two main alternative transcripts, hERG1A and hERG1B. hERG1B encodes a protein, hERG1B, with a unique N-terminus that justifies the peculiar biophysical features of hERG1B-sustained currents.[8] The two hERG1 isoforms are expressed at different ratios and differentially contribute to sustain hERG1 currents in the tissues where hERG1 is functionally expressed. For example, whereas hERG1B is expressed at low levels in the human heart,[9] it represents the main hERG1 isoform in tumor cells, such as neuroblastomas and leukemias.[10] This fact makes hERG1B a promising tumor-specific target.[6] To be exploited for diagnostic and therapeutic purposes, the differential expression of hERG1A and hERG1B transcripts in primary tumors must be well defined. Whereas a high expression of hERG1B has been reported in adult primary AML cases,[5] no data on the differential expression of hERG1A and hERG1B in ALL have been reported so far.

In the present study, we analyzed the expression and prognostic impact of the two hERG1 encoding genes in two cohorts of pediatric ALL patients, BCP-ALL and T-ALL. In particular, we examined the expression of hERG1A and hERG1B mRNA by SYBR Green real-time quantitative PCR (Rt-qPCR) in 100 BCP-ALL (n=94 children and n=6 infants below 1 year of age) and 111 T-ALL patients. Expression values were compared with those obtained in pooled CD19+ B and CD3+ T cells, respectively. All the patients studied were enrolled in the AIEOP LAL 2000-R2006 therapy protocol, whose details are reported in.[11] The clinico-pathological characteristics of the patients, along with the expression of the two hERG1 transcripts, are shown in Table 1.

Table 1

Distribution of clinical and biological features and results of the univariate analysis in children with BCP-ALL and T-ALL

	hERG1A median (1st–3rd quartile)				P	hERG1B median (1st–3rd quartile)				P
BCP-ALL
Sex	F (43)	M (51)				F (43)	M (51)
	0.03 (0.01–0.04)	0.02 (0.01–0.08)			0.359	8.91 (3.90–24.49)	5.4 (2.39–12.40)			0.1554
Age	1–5 years (57)	6–9 years (22)	10–17 years (15)			1–5 years (57)	6–9 years (22)	10–17 years (15)
	0.03 (0.01–0.06)	0.03 (0.01–0.07)	0.01 (0.01–0.06)		0.2155	6.95 (3.01–18.00)	6.68 (2.48–16.63)	4.91 (1.41–15.88)		0.9248
IF	Pre pre B (6)	Call (55)	Pre-B (31)	B-lin (2)		Pre pre B (6)	Call (55)	Pre-B (31)	B-lin (2)
	0.01 (0–0.01)	0.03 (0.01–0.07)	0.03 (0.01–0.04)	0.01,0.02	0.0326	7.07 (1.55–15.88)	8.1 (4.20–22.78)	4.71 (2.10–9.43)	5.81,6.43	0.2968
WBC	<50 000 (67)	>50 000 (27)				<50 000 (67)	>50 000 (27)
	0.03 (0.01–0.07)	0.02 (0.01–0.06)			0.7052	6.97 (3.88–17.32)	5.22 (1.42–15.88)			0.2643
PDN response	PPR (14)	PGR (79)				PPR (14)	PGR (79)
	0.05 (0.03–0.14)	0.03 (0.01–0.05)			0.0492	2.25 (1.50–9.78)	6.72 (3.44–17.32)			0.3493
CNS	Positive (1)	Negative (93)				Positive (1)	Negative (93)
	0.01	0.03 (0.01–0.07)				0.82	6.69 (2.57–16.72)
MRD	SR (30)	MR (40)	HR (11)			SR (30)	MR (40)	HR (11)
	0.04 (0.03–0.11)	0.01 (0.01–0.03)	0.01 (0.01–0.03)		0.2127	6.97 (2.48–16.63)	6.43 (4.06–15.88)	3.31 (1.02–16.80)		0.3855
Risk group	SR (27)	MR (43)	HR (24)			SR (27)	MR (43)	HR (24)
	0.04 (0.03–0.11)	0.02 (0.01–0.04)	0.01 (0.01–0.07)		0.2613	8.05 (3.90–25.43)	6.56 (4.40–16.94)	2.25 (1.26–10.13)		0.2103
Translocation	12;21 (25)	MLL (3)	No (66)			12;21 (25)	MLL (3)	No (66)
	0.02 (0.01–0.07)	0.00, 0.01, 0.01	0.03 (0.01–0.07)		0.0593	6.3 (2.48–25.43)	0.82, 1.42, 5.22	8.1 (2.73–15.88)		0.1858
T-ALL
Sex	F (18)	M (93)				F (18)	M (93)
	0.66 (0.11–1.42)	0.76 (0.20–4.29)			0.2250	2.91 (1.90–14.00)	6.35 (2.35–20.59)			0.2921
Age	1–5 years (35)	6–9 years (31)	10–17 years (45)			1–5 years (35)	6–9 years (31)	10–17 years (45)
	0.49 (0.20–2.86)	0.71 (0.09–3.70)	1.35 (0.31–4.86)		0.4145	5.11 (2.54.15.94)	6.57 (1.53—23.66)	4.85 (1.94–20.18)		0.9664
IF	Early T (53)	Thym (41)	T mature (15)			Early T (53)	Thym (41)	T mature (15)
	1.1 (0.25–4.25)	0.6 (0.16–1.19)	1.41 (0.76–15.00)		0.1186	5.11 (1.94–17.06)	4.85 (2.18–19.37)	3.3 (2.67–31.31)		0.5783
WBC	<50 000 (42)	>50  000 (68)				<50  000 (42)	>50  000 (68)
	0.23 (0.09–0.74)	1.85 (0.49–8.00)			<0.001	12.98 (2.70–23.66)	3.41 (1.92–12.13)			0.0312
PDN response	PPR (37)	PGR (73)				PPR (37)	PGR (73)
	0.49 (0.13–1.50)	2.13 (0.74–6.60)			0.0045	3.3 (1.90–12.27)	6.83 (2.20–23.66)			0.5758
CNS	Positive (6)	Negative (101)				Positive (6)	Negative (101)
	6.31 (1.35–9.97)	0.71 (0.17–3.70)			0.0951	16.29 (2.92–31.31)	5.71 (2.35–19.37)			0.4684
MRD	SR (14)	MR (56)	HR (27)			SR (14)	MR (56)	HR (27)
	1.11 (13.00–1.99)	0.65 (0.21–4.54)	0.93 (0.16–2.86)		0.9691	7.61 (3.14–22.95)	4.86 (1.65–19.32)	3.73 (1.9–16.25)		0.3565
Risk group	SR (13)	MR (48)	HR (50)			SR (13)	MR (48)	HR (50)
	1.03 (0.13–1.42)	0.49 (0.17–4.06)	1.13 (0.26–4.29)		0.4728	10.11 (3.43–22.95)	5.11 (1.92–19.32)	4.72 (2.42–18)		0.3572

Abbreviations: BCP-ALL, B-cell precusor acute lymphoblastic leukemia; CNS, central nervous system; HR, high risk; MR, medium risk; MRD, minimal residual disease; ND, no data; PDN, prednisone; PGR, prednisone—good responder; PPR, prednisone—poor responder; SR, standard risk; T-ALL, T acute lymphoblastic leukemia; WBC, white blood count.

Note: When the number of observations is less than four, the actual values of hERG1A and hERG1B are given.

Levels of hERG1A and hERG1B mRNA expression in pediatric BCP-ALL (n=94) and T-ALL (n=111) patients in different subgroups measured by SYBR Green RQ-PCR. Levels of the hERG1A and hERG1B transcripts were normalized to levels of the corresponding transcript in normal CD19+ cells or CD3+ cells. Statistical analyses were performed with R. Values reported in each column are referred to the median value (1st–3rd quartile). The healthy donor specimens used as calibrator in RQ-PCR experiments are a pool of five healthy buffy coats. CD19+ cells or CD3+ cells were sorted after pooling to be used as calibrators in BCP- or T-ALL analyses, respectively. RQ-PCR experiments were always performed in triplicate. BCP-ALL: total median hERG1A: 0.03 (0.01–0.08); total median hERG1B: 5.11 (1.42–18). T-ALL: total median hERG1A: 0.76 (0.17–4.08); total median hERG1B: 5.11 (2.00–19.98).

BCP-ALL: <85% blasts median hERG1A: 0.04 (0.01–0.08) vs >85% blasts median hERG1A: 0.02 (0.01–0.06), P=0.38; <85% blasts median hERG1B: 6.12 (1.64–16.40) vs >85% blasts median hERG1B: 5.88 (1.52–17.6), P=0.27.

In BCP-ALL children (Table 1, upper panel) the hERG1A isoform was downregulated (median value=0.03; 0.01–0.07), whereas hERG1B was upregulated (median value=6.68; 2.48–16.63), compared with normal B cells. Although generally hypo-expressed, hERG1A was higher in CALL and pre-B immunophenotype subgroups (P=0.0326) and in prednisone poor responder (PPR) patients compared with prednisone good responder (PGR) patients (P=0.0492). A marginally statistically significant higher expression was evidenced in BCP-ALL patients with no chromosomal translocations, compared with patients with either the 12;21 or the 4;11 translocation (P=0.0593).

In the infant subgroup of BCP-ALL, both hERG1A and hERG1B transcripts were hypo-expressed (median values: 0.03 and 0.24, respectively; Supplementary Table 1S).

Similarly to BCP-ALL patients, T-ALL patients (Table 1, lower panel) showed an overexpression of the hERG1B transcript and a downregulation of hERG1A, although the gap between the two isoforms was less evident (median values: 5.11; 2.00–19.98 and 0.76; 0.17–4.08, respectively). A higher expression of hERG1A was detected in T-ALL patients with a WBC count ⩾50.000 compared with patients with a WBC count <50.000 (median value: 1.85 vs 0.23; P=0.001) and in PPR compared with PGR patients (median value: 2.13 vs 0.49; P=0.005). An indication of overexpression was evidenced also in patients with involvement of the central nervous system (CNS) (P=0.096). The hERG1B transcript was in general overexpressed, in particular in patients with WBC count <50.000 (P=0.031) and, although not significantly, in standard-risk (SR) patients.

Finally, we evaluated the impact of the differential expression of hERG1A and hERG1B on relapse in the two cohorts of BCP-ALL and T-ALL patients. The optimal cutoff value was determined on the basis of the receiver operator characteristic analysis. In BCP-ALL, a cutoff value with proper sensitivity and specificity was found only for the hERG1A transcript. The cumulative relapse rate at 5 years was 32.3% in patients with hERG1A <0.03 and 13.4% in patients with hERG1A ⩾0.03 (P=0.04) (Figure 1a). After adjusting for risk groups in a multivariate Cox model, this association was not statistically significant (HR of relapse in patients with hERG1A <0.03 versus ⩾0.03:1.88; CI 0.77–4.60; P=0.166).

Figure 1

Cumulative incidence of relapse in BCP-ALL and T-ALL patients according to the expression of hERG1A and hERG1B. Cumulative incidence of relapse was estimated by adjusting for competing risks (death) and compared using the Gray test. Statistical analyses were performed with R. (a) Cumulative incidence of relapse in BCP-ALL patients according to the expression of hERG1A. (b) Cumulative incidence of relapse in T-ALL patients according to the expression of hERG1A. (c) Cumulative incidence of relapse in T-ALL patients relative to the expression of hERG1B, according to different cutoffs (6.8 and 1.3) and multivariate analysis.

In T-ALL patients, discriminant values of expression were obtained for both hERG1A and hERG1B. The cumulative relapse rate at 5 years was 37% in patients with hERG1A ⩾0.74 and 22% in patients with hERG1A <0.74 (P=0.020) (Figure 1b). On multivariate analysis, hERG1A lost its statistically significant association with relapse (HR=1.61 95%CI 0.73–3.54 P=0.2404). Patients with hERG1B ⩾6.8 relapsed with higher frequency compared with patients <6.8 (5 years' cumulative incidence of relapse: 38 vs 22%, P=0.17). The Cox model after adjusting for classical prognostic factors (immunophenotype, risk group and WBC) identified hERG1B as an independent factor of higher risk of relapse (HR 2.6; CI 1.26–5.30, P=0.009) (Figure 1c, left panel). Notably, a lower hERG1B cutoff value of 1.3 identified a group of patients (with hERG1B<1.3) with no relapse (Figure 1c, right panel, P=0.03). In a limited set of four patients classified as early T-cell precursor leukemia (ETP-ALL), usually associated with a very high risk of relapse,[12] hERG1B expression was always higher than the cutoff value of 1.3 (Supplementary Table 2S).

In the present study we investigated, for the first time, the differential expression of the two main isoforms of hERG1 potassium channels, hERG1A and hERG1B, in ALL pediatric patients. hERG1 channels exert a relevant role in tumor biology,[6] and their use as prognostic markers in human malignancies is emerging.[5, 13] In leukemias, hERG1 channels regulate either cell migration and transendothelial migration (in AML), or chemoresistance (in ALL). We previously reported that AML blasts from adult patients express both hERG1A and hERG1B transcripts[5] and that several leukemia cell lines preferentially express the hERG1B isoform.[10] However, no data regarding the prognostic relevance of the differential expression of the two hERG1 isoforms in leukemia patients have been reported so far. We show here that, in ALL blasts, either B or T lineage, the hERG1 transcript that is overexpressed compared with normal CD19+B or CD3+T cells is exclusively hERG1B. The genetic mechanisms underlying such overexpression could be related to the GpC islands and consensus sites for transcription factors,[14] which differentiate the hERG1B from the hERG1A promoter. Moreover, the hERG1B-encoded protein, hERG1B, has peculiar biophysical features, which makes hERG1B-sustained currents optimal to allow cell cycle progression in tumor cells.[10]

Notably, the two hERG1 transcripts may have diagnostic and therapeutic relevance in ALL. In particular, the hERG1A transcript, which is generally hypo-expressed in both BCP-ALL and T-ALL, could identify groups of patients with a higher rate of relapse either when deeply downregulated (in BCP-ALL) or when slightly upregulated (in T-ALL).

The most relevant data provided here clearly show that, in the T-ALL cohort, the overexpression of hERG1B has a negative impact on outcome. Indeed, hERG1B expression displays a hazard ratio comparable to that of other factors used for patients' stratification in pediatric T-ALL. The high expression of hERG1B in ETP-ALL, although obtained in very few patients, further reinforces the negative impact of hERG1B on T-ALL outcome. In T-ALL no independent prognostic molecular marker, except Notch1 mutation profile,[15] has clinical relevance, and patients' stratification relies on MRD status and the T cell phenotype.[16] Hence, the expression of hERG1B isoform could be exploited for future stratification of pediatric T-ALL.

Finally, the hERG1B overexpression may have a therapeutic relevance, independently of the ALL immunophenotype, either B or T. In fact, we provided evidence that hERG1 blockers can overcome chemoresistance both in vitro and in leukemia mouse models.[7] However, several of the many hERG1 blocking drugs that are available on the market can cause severe cardiotoxicity. Hence, the preferential targeting of the hERG1B isoform could be an approach for overcoming such hindrances.[6] Indeed, we have recently provided evidence that a novel pyrimido-indole compound has a clear antileukemic effect as it preferentially blocks hERG1B-sustained currents, with no adverse cardiac effect (Gasparoli L, unpublished results). This, or similar drugs, could hence be proposed for a patient's tailored therapeutic approach especially in nonresponsive pediatric T-ALL, such as ETP-ALL, with a high hERG1B expression.