HCC patients with mutations in LRP1B have a worse prognosis
The mutation rate of LRP1B gene is high in TCGA patients with HCC, ranking at the 13th place, reaching about 8% (Figure 1A), and the overall survival of LRP1B mutant HCC patients is significantly lower than that of LRP1B wild-type HCC patients (Figure 1B). Then we analyzed the differentially expressed genes of LRP1B mutant and wild-type HCC patients, concluding that 187 genes demonstrate specifc expression manner in LRP1B mutant HCC patients. There were 134 up-regulated genes and 53 down-regulated genes shown as Figure 1C and Figure 1D.
The risk model constructed by 11 genes can better predict the prognosis of HCC patients
Univariate Cox regression analysis was performed with 187 differentially expressed gene as continuous variables. At the same time, the Hazard ratio (HR) of each gene was calculated. With P-value < 0.01 as the threshold, 68 genes were finally selected out. Protective genes with HR value less than 1 were favorable for prognosis, while risk genes with HR value greater than 1 were unfavorable for prognosis. It was turned out that 3 of the 68 genes were protective genes, and the remaining 65 genes were risk genes. The forest map of the top 20 genes with the smallest P-value among these 68 genes is shown in Figure 2A. The optimal number of genes was 11 (Figure 2B, with the smallest lambda value), 11 genes were CELSR3, KLRB1, CENPA, CDCA8, PKIB, ADAMTS5, FTCD, CDX2, SFN, MYT1L and ZP3, respectively.
The risk score model was established for predicting survival: Riskscore=(0.019437033*CELSR3)+(-0.138416034*KLRB1)+(0.070137596*CENPA)+(0.093717620*CDCA8)+(0.007794412*PKIB)+(0.166654021*ADAMTS5)+(-0.041895786*FTCD)+(0.054670700*CDX2)+(0.006845825*SFN)+(0.026858059*MYT1L)+(0.047343209*ZP3). We calculated the risk score for each patient and divided the TCGA data set and the ICGC validation set into the high-risk group and the low-risk group according to the median of the risk score.
Survival analysis showed that in TCGA data set and ICGC validation set, the high-risk HCC samples had the worse overall survival (Figure 2C). In addition, the AUC of 1-year, 3-year and 5-year survival time of TCGA dataset was 0.8119, 0.7622 and 0.7001 respectively (Figure 2D). The AUC of 1-year, 3-year and 5-year survival in ICGC validation set were 0.7182, 0.7297 and 0.7545 respectively, which indicated that the risk model could predict the prognosis of HCC patients effectively in both datasets. At the same time, we found that the expression of 11 genes was significantly different between high-risk and low-risk groups in TCGA and ICGC dataset (Figure 2E). In general, the risk score calculated from the risk model constructed by Celsr3, Klrb1, CENPA, CDCA8, PKIB, Adamts5, FTCD, CDX2, SFN, MYT1L and ZP3 could predict the prognosis of patients with HCC.
Risk score is an independent prognostic marker of HCC
We included age, sex, stage, LRP1B status, HBV index, and risk score for the next investigation. The results are shown in Figure 3A. It was found that risk score and age were significantly associated with overall survival, and the samples with high risk score had a higher risk of death and were unfavorable for prognosis (HR=3.24, 95% CI: 2.26-4.6, P < 0.001).
In order to further explore the prognostic value of risk score in HCC patients with different clinicopathological factors (including age and stage), we regrouped patients by age and stage, performing a Kaplan-Meier survival analysis. It could be demonstrated that the overall survival rate of the high-risk group is clearly lower than that of the low-risk group of the samples in different ages and stages (Figure 3B-3C). These results confirm that the risk score can be used as an independent indicator to predict the prognosis for HCC patients.
Nomogram model can better predict the prognosis of HCC patients
We use four independent prognostic factors: age, gender, radiotherapy status as well as risk score to construct the nomogram model (Figure 4A). For each patient, draw three lines upwards to determine the points obtained from each factor in the Nomogram. The "Total Points" axis is determined by the sum of these points, of which draw a line to generate the probability of HCC patients surviving 1, 3, and 5 years. The one-year and two-year corrected curves in the calibration chart are relatively close to the ideal curves (Figure 4B-4D).
Immune status of HCC patients in high and low-risk groups
We used the CIBERSORT method combined with LM22 feature matrix to estimate the difference of immune infiltration between 22 immune cells in high-risk and low-risk groups of patients with HCC. Figure 5A summarized the results of immune cell infiltration in 352 HCC patients. There are significant differences in the infiltration ratios of 7 types of immune cells, such as Macrophages (M0), between high and low-risk groups (Figure 5B).
Since the expression of immune checkpoints has become a biomarker of immunotherapy for HCC patients, we analyzed the correlation between patient risk score and key immune checkpoints (CTLA4, PDL1, LAG3, TIGIT, IDO1, TDO2). It could be seen that the risk score is closely associated with the 6 checkpoints (Figure 6A). At the same time, 2 of the 6 immune checkpoints (PDL1, TDO2) have the significant differences in the high and low-risk groups of HCC patients (Figure 6B-6C).
LRP1B gene can be used a HCC prognostic marker
Since the LRP1B gene demonstrate specific expression among different prognostic stages, next we testified whether LRP1B gene could be developed as a clinical marker to predict prognostic stages for HCC (Table 1). The results clearly support that the accuracy rate using LRP1B gene as a HCC prognostic marker reaches the standard.