Background:

Acute myeloid leukemia (AML) is the most common form of hematological malignant tumors that threatens human health. In the last decades, the rapid evolution in cytogenetics and molecular abnormalities made a breakthrough in the diagnosis and prognosis prediction of AML. However, there still has high heterogeneity among AML patients. Recently, researchers focused on long non-coding RNAs (lncRNAs), which once were addressed as products of "junk DNA", may play a key role in the initiation and progression of AML. Furthermore, a study from Ohio State's Comprehensive Cancer Center built a useful prognostic lncRNA score system for elder patients (>60 years) with cytogenetically normal AML by 48 lncRNAs [Garzon et al. PNAS 2014; 111:18679-84]. It can be expected that lncRNAs will promote the diagnosis and risk categories of AML in the near future. In this study, a risk scoring system was constructed upon 3 lncRNAs in de novo AML patients. We also sought to explore the functionality of these lncRNAs.

Methods:

By using Arraystar Human LncRNA Array V4.0, we obtained deregulated transcripts in AML, including 3,499 lncRNAs and 3,105 mRNAs (GSE103828). Expression patterns of all deregulated transcripts were extracted from 151 AML patients of The Cancer Genome Atlas (TCGA) RNA sequencing data. Through mathematical modeling, we identified 3 lncRNAs whose expression levels were independently associated with overall survival (OS). We then constructed a risk scoring system based on the 3 lncRNAs, age and 2008 WHO risk categories. Then the Receiver Operating Characteristic (ROC) was used to identify its test power and the best threshold score for 3-year survival status. Bone marrow samples from patients with AML and iron deficiency anemia (IDA) were collected. qRT-PCR was performed to verify the expression of lncRNAs in AML patients and IDA controls.

Results:

In the TCGA dataset, the area under ROC curve was 0.765, which indicated the risk scoring system has a good efficiency to predict 3-year survival status for AML patients, and the threshold score 1.639 was recommended to distinguish the high and low risk score groups (Figure A). Patients with higher risk score had a shorter OS (median 440.31 days vs. 886.16 days, p<0.001, Figure B) than those with lower risk score. To promote the practicality of the risk scoring system, we constructed Nomogram predictive modeling. The patient with 1.639 risk score was shown in the Nomogram (Figure C). We then performed qRT-PCR to verify the expression of lncRNAs and the availability of the risk scoring system. All of the 3 lncRNAs, RP11-222K16.2, LINC00899 and RP11-305O6.3, showed significantly lower expression in AML (n=46) compared to IDA controls (n=17, P<0.05), and the risk scoring system performed remarkable predicting effectiveness among AML patients (Figure D).

Conclusion:

We identified 3 deregulated lncRNAs in AML and constructed a risk scoring system based on the it, which provided distinct insights into the clinical and biological implications of lncRNAs expression in de novo AML patients. It may improve the risk stratification of newly-diagnosed AML patients. Further studies on the mechanisms of the lncRNAs are in process.

Figure.
Figure.
View largeDownload slide
Disclosures

No relevant conflicts of interest to declare.

Topics:
leukemia, myelocytic, acute, cancer, polymerase chain reaction, bone marrow specimen, dna, iron deficiency anemia, risk reduction, rna, messenger, sequence analysis, rna, area under the roc curve

Author notes

*

Asterisk with author names denotes non-ASH members.

© 2018 by The American Society of Hematology
2018
Sign in via your Institution