HPV status is associated with altered PIWI-interacting RNA expression pattern in head and neck cancer

Abstract

Objectives: As HPV-induced cases of oral malignancy increase, it is important to understand the molecular differences between HPV positive and negative head and neck squamous cell carcinoma (HNSCC). PIWI-interacting RNAs (piRNAs) are a class of small non-coding RNAs aberrantly expressed in cancer. We analyzed global piRNA expression patterns to define the HNSCC piRNA transcriptome and assess whether HPV infection status associates with changes in piRNA levels.

Materials and methods: A total of 498 HNSCC small RNA sequencing libraries were acquired from the Cancer Genomics Hub (cgHUB) Data Repository and a custom sequence analysis pipeline was developed to deduce piRNA expression from raw sequencing data. Expression matrices were aligned to clinicopathological features in order to analyze piRNA expression patterns across different HNSCC groups. The association of a piRNA signature with HPV-positive patient survival was evaluated using a Cox proportional hazard model.

Results: Analysis of piRNA levels between HNSCC and non-malignant tissues revealed distinct expression patterns, with 87 piRNAs exclusively expressed in tumor samples. HPV infection status affected the expression of 41 of these piRNAs. Eleven (26.8%) piRNAs were significantly downregulated in HPV16/18 tumors compared to other HPV types. Remarkably, expression of a combination of five-piRNAs in HPV-positive HNSCC tumors was associated with worse overall survival.

Conclusion: The expression of specific piRNAs is deregulated in HNSCC, and changes with both HPV status and type. Importantly, a five-piRNA signature is able to delineate a subset of HPV-positive HNSCC patients with poor outcome, highlighting the potential utility of piRNAs in patient management.

Keywords: Head and Neck cancer; Human papillomavirus; Kiwi-interacting RNAs; Non-coding RNA; Survival; piRNA.

Figure 1. Genome-wide expression of piRNAs in head and neck non-malignant and tumor tissue

The expression of 305 piRNAs in non-malignant (blue) and tumors (red) are presented in a circos plot, which relates piRNA genes to the chromosomal location.

Figure 2. HPV status and piRNA expression patterns

Comparison of piRNA expression patterns between HPV negative (average of 370 HPV− HNSCC tumor piRNA transcriptomes, purple bars) and HPV positive cases (average of 83 HPV+ HNSCC tumor piRNA transcriptomes, yellow bars).

Figure 3. Expression boxplots of the five survival-associated piRNAs (RPKM) in HPV+ tumors

On each boxplot, the average is denoted by a black diamond, the line within each boxplot indicates the median value, the border of the box indicates the 25th and 75th percentiles, and the dashed line represents the range.

Figure 4. A five piRNA signature delineates the outcome of patients with HPV+ HNSCC

The expression values of the piRNAs in the model were transformed into a score representing the probability of death by disease by multiplying the expression values of each piRNA by their respective Cox proportional hazard coefficient (Table 2), and then summing their values. The Kaplan-Meier curve shows patients with a high score (top 30^th percentile) had worse survival than those with a low score (bottom 30^th percentile).