Overexpression of lncRNA PANDAR predicts adverse prognosis in acute myeloid leukemia

Abstract

Background and purpose

Abundant studies have shown that lncRNA PANDAR plays an oncogenic role in human solid tumors. Although abnormal expression of PANDAR has been well investigated in solid tumors, it was rarely studied in hematologic diseases. Hence, the aim of this study was to determine the PANDAR expression level and its clinical significance in patients with acute myeloid leukemia (AML).

Materials and methods

For detecting the expression level of PANDAR in 119 AML patients and 26 controls, real-time quantitative PCR was used in this study. The prognostic values were evaluated by using Kaplan–Meier analysis, Cox regression analyses, and logistic regression analysis.

Results

PANDAR was significantly overexpressed in AML and might be a promising biomarker which could distinguish AML from normal samples (P<0.001). Patients with high expression of PANDAR (PANDAR^high) were older and showed higher bone marrow blasts than patients in PANDAR^low group (P=0.029 and 0.032, respectively). Significant differences between these groups were also detected regarding risk group and karyotype finding (P=0.009 and 0.041, respectively). Importantly, PANDAR^high patients presented a significant lower complete remission rate compared to PANDAR^low patients (P<0.001). Furthermore, Kaplan–Meier analysis showed that PANDAR^high patients had shorter overall survival compared to PANDAR^low patients observing the whole AML cohort, and also in the non-M3 group of patients (P<0.001 and P=0.005, respectively). Multivariate analysis of Cox and logistic regression analysis confirmed that high PANDAR expression was an independent unfavorable risk factor for overall survival and complete remission in both observed patient groups.

Conclusion

These results revealed that PANDAR was overexpressed in AML, and that higher PANDAR expression was associated with poor clinical outcome. Our study therefore suggests that PANDAR expression is a promising biomarker for prognostic prediction for AML.

Keywords:

• long noncoding RNA
• PANDAR expression
• acute myeloid leukemia
• complete remission
• overall survival

Introduction

Acute myeloid leukemia (AML) is a cytogenetically and molecularly heterogeneous disease which is marked by uncontrolled clonal expansion of blast cells.¹ Although the new treatment strategies based on molecular biology of AML have been adopted in recent years, the prognosis of the disease remains poor.^2–4 It has become apparent that karyotype abnormalities have important value for AML diagnosis classification, prognostic evaluation, and guiding individual treatment.^5,6 Cytogenetic aberrations together with several gene mutations including NPM1, CEBPA, TP53, TET2, DNMT3A, and FLT3-ITD have a strong impact on clinical outcome of AML patients.⁷ In addition to genetic abnormalities, the aberrant expression of some genes, such as overexpression of ERG, BAALC, and EVI1, also has been proven to affect prognosis for AML patients.⁷ These important findings open up a new field for discovering novel promising biomarkers for AML patients, especially for those who are at risk of poor outcome, so that these patients can be treated with optimized treatment strategies.

Long noncoding RNAs (lncRNAs) are regarded as a kind of noncoding RNA, which are longer than 200 nucleotides. Recently, many studies have reported that lncRNAs play vital roles in gene expression regulation through association with key transcription factors and microRNAs.^8,9 lncRNAs could act as an important component in every step of cell biology, which includes the adjustments of transcription initiation and transcription and posttranscriptional level.¹⁰ Recently, increasing number of research papers revealed that lncRNAs were relevant to many human diseases, especially to human cancers, and many studies began to explore the molecular mechanisms of lncRNA function in the pathogenesis of these disease or cancers.¹¹ With the deepening of the research, it is becoming increasingly apparent that most of the susceptibility to cancer is not caused by the variation of coding sequences of DNA but by the noncoding regulatory sequences, especially by lncRNAs.¹⁰

lncRNA PANDAR, which is located at 6p21.2, plays a vital role in regulation of apoptosis by inhibiting the expression of proapoptotic genes through interaction with the transcription factor NF-YA.¹² To date, the abnormal expression of PANDAR has been reported in various solid cancers, such as hepatocellular carcinoma, gastric cancer, and breast cancer.¹³ However, there are few reports about the expression of PANDAR in blood cancer. Therefore, we focused on exploring the PANDAR expression level and its connection with clinical implication in AML patients.

Materials and methods

Patients and treatment

A total of 119 de novo AML patients and 26 healthy donors were included in the present research, which was approved by the Ethics Committee and Institutional Review Board of the Affiliated People’s Hospital of Jiangsu University. Bone marrow (BM) was collected from all the participants after they signed the informed consents. BM mononuclear cells (BMMNCs) were extracted from BM specimen using Lymphocyte Separation Medium (TBD Sciences, Tianjin, People’s Republic of China). Treatment protocols for AML were described previously.¹⁴

Cytogenetics and mutation analysis

By conventional R-banding method, karyotype was analyzed at the time of initial diagnosis. Risk classification based on the karyotype findings has been done as previously reported.¹⁵ Mutations in C-KIT, NPM1, DNMT3A, N/K-RAS, and U2AF1 were detected by high-resolution melting analysis,^16–20 whereas FLT3-ITD and CEBPA mutations were detected by direct DNA sequencing.^21,22

RNA isolation and reverse transcription

Total RNA was extracted by using Trizol reagent (Invitrogen, Carlsbad, CA, USA). The specific procedure of reverse transcription was conducted as previously reported.²³

Real-time quantitative PCR

The primers for PANDAR are as follows: forward: 5′-CTCCATCATGCCAA GTTCTGC-3′ and reverse: 5′-GAAGGCAGGCAAGACTCGAA-3′. PANDAR expression was detected by real-time quantitative PCR using AceQ qPCR SYBR Green Master Mix (Vazyme Biotech Co., Pis-cataway, NJ, USA). The reaction condition of real-time quantitative PCR was conducted as reported earlier.^24,25 Relative PANDAR expression levels were calculated by 2^−ΔΔCT method.

Statistical analysis

SPSS software version 20.0 (IBM Corporation, Armonk, NY, USA) was used to carry out the statistical analysis. Meanwhile, receiver operating characteristic (ROC) curve and area under the ROC were applied to assess the value of PANDAR expression. Besides, Pearson’s chi-squared analysis was conducted to detect the difference of categorical variables between PANDAR^high group and PANDAR^low group. Through Kaplan–Meier method and Cox regression analysis, the effect of PANDAR expression on prognosis was analyzed. Logistic regression analysis was used to identify the independent risk factors of complete remission (CR). In all tests, P<0.05 was defined as statistically significant.

Results

PANDAR expression in AML

The expression level of PANDAR in controls ranged from 0.000 to 2.926 (median 0.294). PANDAR transcript level in AML patients ranged from 0.005 to 306.109 (median 1.862). Through nonparametric test, PANDAR was found to be significantly upregulated in AML (P<0.001, Figure 1). Besides this, significant upregulation of PANDAR was also found in non-M3-AML and cytogenetically normal AML subgroup of patients (Figure 1).

Figure 1 Expression of PANDAR in controls, whole-cohort AML patients, non-M3 AML patients, and CN-AML patients.

Notes: The distributions of the PANDAR expression in controls, whole-cohort AML patients, non-M3 AML patients, and CN-AML patients are presented with scatter plots. The median level of PANDAR expression in each group is shown with horizontal line.

Abbreviations: AML, acute myeloid leukemia; CN-AML, cytogenetically normal AML.

Distinguishing capacity of PANDAR expression

The ROC curve analysis was applied to evaluate whether PANDAR expression could be used as a biomarker for the diagnosis of AML. The results showed that area under the curve value was 0.800 (95% CI: 0.716–0.883), which suggested the PANDAR expression level might be a potential biomarker in discriminating AML from controls (P<0.001, Figure 2A). In addition, when the cutoff value was 0.840, the sensitivity and specificity of diagnosis of AML were 65.5% and 80.8%. For non-M3-AML and CN-AML patients, significant differences also existed (Figure 2B and C, respectively).

Figure 2 Discriminative capacity of PANDAR expression by ROC curve analysis.

Notes: (A) For whole-cohort AML. (B) For non-M3 AML. (C) For CN-AML.

Abbreviation: AML, acute myeloid leukemia; AUC, area under the curve; CN-AML, cytogenetically normal AML; ROC, receiver operating characteristic.

The connection between PANDAR expression level and clinical characteristics in AML

By the set cutoff value based on the basis of ROC curve, the whole cohort of AML patients was divided into two groups. Clinical features and laboratory parameters representation between PANDAR^high and PANDAR^low groups is separately shown in Table 1. No significant differences were observed in sex, white blood cells (WBCs), hemoglobin, and platelets between two groups (P>0.05). However, patients with PAN-DAR high expression were older than patients in the PANDAR low-expressed group (P=0.029). Patients in PANDAR^high group showed higher BM blasts than patients in PANDAR^low group (P=0.032). Moreover, significant differences between these two groups were also detected regarding risk group and karyotype finding (P=0.009 and 0.041, respectively). Patients in PANDAR^high group had higher frequency of poor karyotypes (15%, 12/78) than patients in PANDAR^low group (2%, 1/41). There was no correlation between PANDAR expression and the common gene mutations (Table 1, P>0.05).

Table 1 Comparison of clinical manifestations and laboratory features between AML patients with low and high PANDAR expression

Patient’s parameters	High (n=78)	Low (n=41)	P-value
Sex, male/female	52/26	27/14	1.000
Median age, years (range)	57 (15–86)	51 (17–80)	0.029
Median WBC, ×10^9/L (range)	13.2 (7–185.4)	5.7 (3–528)	0.214
Median hemoglobin, g/L (range)	78 (32–138)	76 (34–126)	0.569
Median platelets, ×10^9/L (range)	40 (5–415)	34 (4–264)	0.160
BM blasts, % (range)	49.8 (5.0–94.5)	30 (1.0–97.5)	0.032
Risk classification			0.009
Favorable	18 (23%)	18 (44%)
Intermediate	42 (54%)	22 (54%)
Poor	12 (15%)	1 (2%)
No data	6 (8%)	0 (0%)
Karyotype			0.041
Normal	34 (44%)	16 (39%)
t(8;21)	4 (5%)	3 (8%)
t(15;17)	14 (18%)	14 (34%)
t(16;16)	0 (0%)	1 (2%)
Complex	11 (14%)	1 (2%)
Others	9 (12%)	6 (15%)
No data	6 (7%)	0 (0%)
Gene mutation
CEBPA (+/−)	7/56	5/31	0.735
NPM1 (+/−)	8/55	1/35	0.149
FLT3-ITD (+/−)	9/54	3/33	0.528
C-KIT (+/−)	2/61	1/35	1.000
N/K-RAS (+/−)	3/60	2/34	1.000
IDH1/2 (+/−)	5/58	0/36	0.155
DNMT3A (+/−)	6/57	1/35	0.417
U2AF1 (+/−)	3/60	0/36	0.552
CR (–/+)	53/21	14/27	<0.001

Abbreviations: AML, acute myeloid leukemia; BM, bone marrow; CR, complete remission; WBC, white blood cell.

Effect of PANDAR expression on chemotherapy response in AML

In order to explore the impact of PANDAR expression in clinical prognosis with AML patients, we analyzed 115 AML patients with available follow-up data. Compared with PANDAR^low group, patients in PANDAR^high group had a lower CR rate (P<0.001, Table 1). We then analyzed the expression level of PANDAR in AML patients who achieved CR and those without CR, and showed it in scatter plots (P<0.001, Figure 3). Additionally, clinical characteristics of patients with CR and non-CR were further compared. Significant differences were found in PANDAR expression, age, WBCs, BM blast, risk group, and karyotype finding (P<0.05, Table 2). Logistic regression analysis including the most predictive factors was further performed which revealed that PANDAR expression was an independent risk factor that affected CR in whole-cohort AML and non-M3 AML patients (P=0.010 and 0.005, respectively, Tables 3 and 4).

Table 2 Comparison of clinical manifestations and laboratory features between CR and non-CR in AML patients receiving induction therapy

Patient’s parameters	CR (n=48)	Non-CR (n=67)	P-value
PANDAR expression	0.639 (0.005–190.798)	3.500 (0.051–306.109)	<0.001
Sex, male/female	30/18	46/21	0.551
Median age, years (range)	46.5 (18–81)	62 (17–86)	<0.001
Median WBC, ×10^9/L (range)	4.95 (0.3–528)	28.8 (0.7–185.4)	0.001
Median hemoglobin, g/L (range)	77.5 (34–126)	81 (32–138)	0.748
Median platelets, ×10^9/L (range)	32 (4–153)	42 (5–415)	0.073
BM blasts, % (range)	27 (1.0–97.5)	56 (5.0–94.5)	0.003
Risk classification			<0.001
Favorable	25 (52%)	8 (12%)
Intermediate	20 (42%)	43 (64%)
Poor	3 (6%)	10 (15%)
No data	0 (0%)	6 (9%)
Karyotype			<0.001
Normal	16 (34%)	33 (49%)
t(8;21)	4 (8%)	3 (4%)
t(15;17)	21 (44%)	4 (6%)
t(16;16)	0 (0%)	1 (2%)
Complex	3 (6%)	9 (13%)
Others	4 (8%)	11 (17%)
No data	0 (0%)	6 (9%)
Gene mutation
CEBPA (+/−)	5/37	7/48	1.000
NPM1 (+/−)	3/39	6/49	0.728
FLT3-ITD (+/−)	4/38	8/47	0.545
c-KIT (+/−)	2/40	1/54	0.577
N/K-RAS (+/−)	0/42	5/50	0.067
IDH1/2 (+/−)	0/42	5/50	0.067
DNMT3A (+/−)	3/39	4/51	1.000
U2AF1 (+/−)	0/42	3/52	0.256

Abbreviations: AML, acute myeloid leukemia; BM, bone marrow; CR, complete remission; WBC, white blood cell.

Table 3 Univariate and multivariate analyses of variables for CR and OS in whole-cohort AML patients

Variables	CR				OS
	Univariate analysis		Multivariate analysis		Univariate analysis		Multivariate analysis
	OR (95% CI)	P-value	OR (95% CI)	P-value	HR (95% CI)	P-value	HR (95% CI)	P-value
WBC	0.261 (0.109–0.623)	0.002	0.442 (0.160–1.222)	0.116	2.361 (1.525–3.655)	<0.001	1.565 (0.959–2.555)	0.073
Age	0.138 (0.054–0.353)	<0.001	0.165 (0.058–0.467)	0.001	2.546 (1.639–3.956)	<0.001	1.615 (0.985–2.650)	0.058
PANDAR expression	0.205 (0.091–0.466)	<0.001	0.248 (0.109–0.742)	0.010	2.367 (1.453–3.856)	0.001	1.835 (1.051–3.205)	0.033
Risk classification	0.219 (0.107–0.451)	<0.001	0.307 (0.142–0.664)	0.003	2.091 (1.611–2.714)	<0.001	1.593 (1.152–2.201)	0.005
FLT3-ITD mutation	0.618 (0.173–2.211)	0.460	–	–	1.061 (0.526–2.140)	0.868	–	–
NPM1 mutation	0.628 (0.148–2.674)	0.529	–	–	1.871 (0.894–3.916)	0.096	1.536 (0.707–3.338)	0.279
CEBPA mutation	0.927 (0.272–3.155)	0.903	–	–	1.196 (0.593–2.415)	0.617	–	–
c-KIT mutation	2.700 (0.237–30.824)	0.424	–	–	0.684 (0.167–2.796)	0.597	–	–
N/K-RAS mutation	Undetermined	0.999	–	–	4.240 (1.625–11.064)	0.003	4.556 (1.712–12.126)	0.002
IDH1/2 mutation	Undetermined	0.999	–	–	4.859 (1.879–12.564)	0.001	4.340 (1.411–13.350)	0.010
DNMT3A mutation	0.981 (0.207–4.639)	0.980	–	–	1.123 (0.485–2.597)	0.787	–	–
U2AF1 mutation	Undetermined	0.999	–	–	5.353 (1.608–17.819)	0.006	1.249 (0.270–5.772)	0.776

Notes: Variables including WBC (≥30×10⁹ vs <30×10⁹/L), age (≤60 vs >60 years), PANDAR expression (low vs high), risk classification (favorable vs intermediate vs poor), and gene mutations (mutant vs wild type). Multivariate analysis includes variables with P<0.200 in univariate analysis.

Abbreviations: AML, acute myeloid leukemia; CR, complete remission; HR, hazard ratio; OR, odds ratio; OS, overall survival; WBC, white blood cell.

Table 4 Univariate and multivariate analyses of variables for CR and OS in non-M3 AML patients

Variables	CR				OS
	Univariate analysis		Multivariate analysis		Univariate analysis		Multivariate analysis
	OR (95% CI)	P-value	OR (95% CI)	P-value	HR (95% CI)	P-value	HR (95% CI)	P-value
WBC	0.368 (0.138–1.001)	0.050	0.382 (0.126–1.164)	0.090	1.710 (1.083–2.699)	0.021	1.350 (0.811–2.250)	0.249
Age	0.190 (0.064–0.569)	0.003	0.221 (0.069–0.707)	0.011	1.780 (1.125–2.816)	0.014	1.348 (0.801–2.270)	0.261
PANDAR expression	0.223 (0.083–0.596)	0.003	0.210 (0.070–0.624)	0.005	2.063 (1.206–3.530)	0.008	1.948 (1.059–3.583)	0.032
Risk classification	0.435 (0.188–1.002)	0.051	0.633 (0.254–1.578)	0.327	1.692 (1.223–2.340)	0.001	1.549 (1.069–2.246)	0.021
FLT3-ITD mutation	0.467 (0.092–2.386)	0.360	–	–	1.138 (0.556–2.331)	0.723	–	–
NPM1 mutation	1.023 (0.234–4.478)	0.976	–	–	1.462 (0.695–3.075)	0.317	–	–
CEBPA mutation	1.571 (0.444–5.559)	0.483	–	–	0.905 (0.446–1.836)	0.781	–	–
c-KIT mutation	2.083 (0.125–34.750)	0.609	–	–	0.414 (0.057–2.992)	0.382	–	–
N/K-RAS mutation	Undetermined	0.999	–	–	4.023 (1.520–10.649)	0.005	5.354 (1.962–14.610)	0.001
IDH1/2 mutation	Undetermined	0.999	–	–	4.529 (1.736–11.811)	0.002	6.957 (2.201–21.991)	0.001
DNMT3A mutation	1.602 (0.330–7.781)	0.559	–	–	0.816 (0.349–1.907)	0.639	–	–
U2AF1 mutation	Undetermined	0.999	–	–	5.375 (1.596–18.102)	0.007	1.613 (0.298–8.725)	0.579

Notes: Variables including WBC (≥30×10⁹ vs <30×10⁹/L), age (≤60 vs >60 years), PANDAR expression (low vs high), risk classification (favorable vs intermediate vs poor), and gene mutations (mutant vs wild type). Multivariate analysis includes variables with P<0.200 in univariate analysis.

Abbreviations: AML, acute myeloid leukemia; CR, complete remission; HR, hazard ratio; OR, odds ratio; OS, overall survival; WBC, white blood cell.

Figure 3 Expression of PANDAR in CR and non-CR AML patients receiving induction therapy.

Notes: The distributions of the PANDAR expression in CR and non-CR groups are illustrated with scatter plots. The median level of PANDAR expression in each group is shown with horizontal line.

Abbreviations: AML, acute myeloid leukemia; CR, complete remission.

The relationship between PANDAR expression and prognosis in AML patients

The survival analysis indicated that in the whole-cohort AML patients with high PANDAR expression had a shorter overall survival (OS) time than those who were in PANDAR low-expressed group (P<0.001, Figure 4A). In non-M3 AML, patients with PANDAR high expression also had a shorter OS compared with those with PANDAR low expression (P=0.005, Figure 4B). Regretfully, patients with PANDAR high expression did not presented a significant shorter OS than patients with PANDAR low expression among CN-AML (P=0.238, Figure 4C). Multivariate analysis which included variables in univariate analysis with P<0.2 (WBC [≥30×10⁹/L vs <30×10⁹/L], age [≤60 vs >60 years], risk group [favorable vs intermediate vs poor], PANDAR expression [high vs low], gene mutations [mutant vs wild type]). Multivariate analysis further showed that PANDAR expression was a significant independent risk factor in affecting OS among whole-cohort AML patients and non-M3 AML patients (P=0.033 and 0.032, respectively, Tables 3 and 4).

Figure 4 Prognostic value of PANDAR expression in AML.

Notes: (A) For whole-cohort AML patients. (B) For non-M3 patients. (C) For CN-AML patients. Overall survival was analyzed between PANDAR^high and PANDAR^low groups and performed by Kaplan–Meier method.

Abbreviations: AML, acute myeloid leukemia; CN-AML, cytogenetically normal AML; cum, cumulative.

Discussion

Lately, more and more researchers are devoted to exploring noncoding RNA and AML.²⁶ Many studies have proved that lncRNAs indeed played an important regulatory role in human cancers, and it was closely related with the occurrence and the development of various tumors.^14,27 Also, increasing number of research papers have shown that the abnormal expression of PANDAR was connected with the tumorigenesis of various solid tumors.^28–32 In the first report published by Hung et al, it was indicated that PANDAR inhibited the expression of proapoptotic genes by interacting with the transcription factor NF-YA.¹² Thereafter, Li et al²⁸ found that PANDAR was upregulated in thyroid cancer. Further investigating the regulatory mechanism of PANDAR, Li et al²⁸ also found that knockdown of PANDAR could promote apoptosis of thyroid cells by reducing the expression of Bcl2 and activating Bax. In addition, Sang et al³⁰ also reported that PANDAR, which was obviously upregulated in breast cancer tissues and cell lines, could affect the cell cycle by regulating its downstream target p16^INK4A. In summary, PANDAR played a significant role in various cancers, including in cancer initiation and progression, and it could serve as an oncogene in these cancers.

In the studies examining the expression level of PANDAR, many reports showed that PANDAR was associated with the prognosis of cancers. For instance, Li et al found that PANDAR was upregulated in thyroid cancer tissue and cell lines, and it could be a promising therapeutic target and important biomarker for thyroid cancer.²⁸ Similarly, an article reported that the expression level of PANDAR in hepatocellular carcinoma was crucially associated with the size of tumor nodule, vascular invasion, and TNM stage.²⁹ Moreover, overexpression of PANDAR was relevant to the poorer survival and shorter recurrence duration for the disease in hepatocellular carcinoma patients, and it could be recognized as a potential tumor biomarker and therapeutic target.²⁹ However, the effect of PANDAR expression on prognosis in blood malignancies remains poorly defined. Findings from our study demonstrated that high expression of PANDAR indicated a poor prognosis in AML patients. PANDAR expression level influenced CR rate, with the PANDAR^high group having lower CR rate in comparison to the PANDAR^low group. Logistic regression analysis showed that PANDAR expression was an independent prognostic factor for CR. More importantly, Kaplan–Meier survival analyses clearly showed that patients with higher expression of PANDAR had a shorter OS than those patients with lower expression. Univariate and multivariate Cox regression analyses revealed the increased PANDAR expression was an independent unfavorable risk factor in AML patients.

Our study was the first to report that PANDAR was upregulated in AML and was also the first to demonstrate the prognostic value of PANDAR in AML.

Conclusion

Expression of PANDAR was frequently upregulated in AML, and high expression of PANDAR as an independent unfavorable risk factor for CR and OS in whole-cohort and non-M3 AML patients. Therefore, our findings indicated that PANDAR was a potential biomarker for AML and it might effectively predict the outcome of AML patients.

Acknowledgments

This work was supported by National Natural Science Foundation of China (81270630), Medical Innovation Team of Jiangsu Province (CXTDB2017002), 333 Project of Jiangsu Province (BRA2016131), Six Talent Peaks Project in Jiangsu Province (2015-WSN-115), Zhenjiang Clinical Research Center of Hematology (SS2018009), China Postdoctoral Science Foundation funded project (2016M601748), Youth Medical Talents Project of “Ke Jiao Qiang Wei” project of Jiangsu province (QNRC2016450, QNRC2016449), and Key Medical Talent Program of Zhenjiang City.

Disclosure

The authors report no conflicts of interest in this work.