Page 274 - EJMO-9-3

Eurasian Journal of
Medicine and Oncology WGCNA and LASSO for osteoporosis biomarkers

in the identification of a total of 1,020 DEGs. In addition, a observed in the OP group, whereas LOC286177 and
volcano plot (Figure 2B) and a heatmap (Figure 2C) were PRPF39 shosed no significant differential expression
then generated to visualize the distribution of DEGs. (Figure 7A-G). Therefore, subsequent analyses focused on
the five validated genes, including NUCB1, PEX19, MTA1,
3.2. Construction of weighted gene co-expression DRAP1, and PCDHGA1.
networks
To investigate the association among OP-associated 3.5. ROC curve analysis
DEGs, we constructed a weighted gene co-expression Next, we performed ROC curve analysis to evaluate
network using the WGCNA package. The optimal soft- the diagnostic performance of the five hub genes. In the
threshold power was determined to be 7 (Figure 3A), GSE35958 dataset, all five genes such as NUCB1, PEX19,
and a total of 10 distinct co-expression modules were MTA1, DRAP1, and PCDHGA1 demonstrated excellent
identified (Figure 3B-D). Among these, the black module diagnostic accuracy, each achieving an AUC of 1.000
exhibited the strongest correlation with OP (correlation (Figure 8A). In the GSE35956 dataset, NUCB1, MTA1, and
coefficient = 0.99, p<0.0001), establishing it as the key DRAP1 also achieved AUCs of 1.000, while PEX19 and
module. Subsequent intersection analysis between the PCDHGA1 showed AUCs of 0.960 and 0.920, respectively
OP-associated DEGs and the genes within the WGCNA (Figure 8B). In both datasets, the AUC values of all five
module revealed 721 core candidate genes (Figure 3E). genes were >0.7, suggesting strong diagnostic potential.
3.3. Functional enrichment analysis 3.6. GSEA
Subsequently, we performed GO and KEGG enrichment To explore the potential pathways of these five hub genes,
analysed on the 721 key genes. In the BP assessment, they we performed single-gene GSEA. The results showed
were mainly associated with functions such as regulation that the high-expression group of NUCB1 was highly
of response to endoplasmic reticulum (ER) stress and enriched in the galactose metabolism, starch and sucrose
small GTPase-mediated signal transduction. In the CC metabolism pathways (Figure 9A). The high-expression
assessment, they were mainly enriched in the ER-Golgi group of PEX19 was highly enriched in the adherens
intermediate compartment and the RNA polymerase II junction and metabolism of xenobiotics by cytochrome
transcription regulator complex. In the MM assessment, P450 pathways (Figure 9B). Both MTA1 and DRAP1 high
they were mainly associated with DNA-binding expression groups were highly enriched in phenylalanine
transcription factor binding and chromatin DNA binding metabolism and SNARE interactions in vesicular transport
(Figure 4A). KEGG pathway analysis revealed significant pathways (Figure 9C and D). The high-expression group
enrichment in pathways including human T-cell leukemia of PCDHGA1 was highly enriched in glycosaminoglycan
virus 1 infection, human immunodeficiency virus 1 degradation and vasopressin-regulated water reabsorption
infection, and focal adhesion (Figure 4B-D). pathways (Figure 9E).
3.4. Machine learning screening and expression 3.7. Immune infiltration analysis
validation of pivotal genes
To analyze the differences in immune levels between
To further screen the pivotal genes, we analyzed 721 key osteoporotic and normal conditions, we analyzed immune
genes using the LASSO regression algorithm, identifying cell infiltration in the control and OP groups using the
seven pivotal genes, namely LOC286177, nucleobindin GSE35958 dataset. The results showed that T cell subsets
1 (NUCB1), peroxisomal biogenesis factor 19 (PEX19), and macrophage subsets were the major subpopulations
metastasis associated 1 (MTA1), DRA aassociated protein 1 of immune cells in both groups (Figure 10A). Notably, the
(DRAP1), protocadherin gamma A1 (PCDHGA1), and pre- expression levels of CD4⁺ T cells, regulatory T cells, M0,
mRNA processing factor 39 (PRPF39) (Figure 5A and B). M1, and M2 macrophages, and several other immune cell
Next, we validated these seven hub genes in both types were significantly altered in OP samples compared to
the GSE35958 dataset and an independent OP dataset, normal samples (Figure 10B).
GSE35956. In the GSE35958 dataset, the expression 3.8. Potential targeted drug prediction and
levels of LOC286177, NUCB1, PEX19, MTA1, DRAP1, identification of core components
and PCDHGA1 were significantly elevated in the OP
group compared with normal controls, while PRPF39 To explore potential therapeutic drugs for OP, we predicted
expression was significantly decreased (Figure 6A-G). In potential Chinese or Western therapeutic drugs for these
the GSE35956 dataset, consistent elevation of NUCB1, five genes by screening them with p<0.05 as a criterion
PEX19, MTA1, DRAP1, and PCDHGA1 was also through Coremine Medical database. The results showed

Volume 9 Issue 3 (2025) 266 doi: 10.36922/EJMO025240252