2.1. Studies selection and the characteristics of the included studies
Using the above strategy, a total of 103 studies were identified. After further screening stepwise, 7 eligible cohort studies (Fig. 1) were finally included [8-14]. A total of 1166 CRC patients, of which 773 (66.3%) with ‘high’ KPNA2. The basic characteristics of the included studies are shown in Table 1.
2.2. KPNA2 expression and OS of CRC Patients
The analysis of the relationship between KPNA2 expression and OS in CRC patients showed that high expression of KPNA2 was significantly correlated with the shortening of OS in CRC patients (Fig. 2A, HR = 2.31, 95% CI 1.46-3.68), but there existed heterogeneity (I²= 55.4%, P=0.047); The results of sensitivity analysis showed that the overall trend did not change when each study was deleted item by item (Fig. 2B); Begg’ s funnel analysis of literature publication bias showed that no publication bias was found (Fig. 2C, PR > |z| = 1.000).
2.3. KPNA2 expression and clinicopathological features
Six studies reported the association between the degree of tumor invasion and KPNA2 expression. The results showed that CRC patients with higher expression of KPNA2 had a higher degree of tumor invasion (Fig. 3A, OR = 2.14, 95% CI = 1.55-2.94), and there was no significant heterogeneity between the results (I² = 0.0%, P = 0.645); Six studies demonstrated the association between KPNA2 expression and lymph node metastasis. The combined data of the included studies showed that the expression of KPNA2 in CRC patients was positively correlated with the degree of lymph node metastasis (Fig. 3B, OR = 2.20, 95% CI = 1.68-2.88), and the heterogeneity was not obvious (I² = 42.9%, P = 0.199); Four studies reported the association between KPNA2 expression and tumor distant metastasis. The analysis results showed that CRC patients with higher KPNA2 expression were more prone to distant metastasis (Fig. 3C, OR = 3.66, 95% CI = 1.81-7.40), and there was no significant heterogeneity among these studies (I² = 48.6%, P = 0.120).
Sensitivity analysis was performed after deleting each study one by one, and the overall trend did not change (Figure 4A-C). The results of Begg’ s funnel plot showed that there was no significant publication bias (Fig. 5A-C, Pr > | z | = 0.133; Pr > | z | = 1.000; Pr > | z | = 0.089).
The pooled data from four studies showed that the expression of KPNA2 was positively correlated with the tumor stage of CRC patients (OR = 1.90, 95% CI = 1.42 – 2.54), and no significant heterogeneity was observed between studies (Fig. 6A, I² = 0.0%, P = 0.555). The results of sensitivity analysis show that excluding any study does not change the overall trend, indicating that the results are statistically significant (Fig. 6B). Begg’ s funnel plot shows that there is no significant publication bias (Fig. 6C, PR > |z| = 0.734).
2.4. Subgroup analysis based on different cut-off values of KPNA2 expression
Since different cutoff values were used in each study to define the high expression of KPNA2, we further performed subgroup analysis according to different cutoff values to evaluate whether the correlation between KPNA2 expression and OS or tumor stage was statistically significant when different cutoff values were used. Subgroup analysis showed that when IRS ≥ 3 was used as the cut-off value, the higher the expression level of KPNA2, the shorter the OS, and this result was statistically significant (Fig. 7A, HR 2.69, 95% CI 1.38-5.24). When IRS ≥ 4 or ≥ 6 was used as the cut-off value, the higher the expression level of KPNA2, the deeper the depth of tumor invasion (Fig. 7B-C, OR 1.84, 95% CI 1.17-2.88; OR 3.23, 95% CI 1.64-6.25) and the more lymph node metastases (OR 2.58, 95% CI 1.78-3.72; OR 2.32, 95% CI 1.18-4.55) and the result was statistically significant. When IRS ≥ 6 was defined as the cut-off value, the higher the expression level of KPNA2, the greater the possibility of distant metastasis of CRC (Fig. 7D, OR 6.18, 95% CI 1.83-20.91). When IRS ≥ 4 was defined as the cut-off value, the higher the expression level of KPNA2, the lower the degree of tumor differentiation (Fig. 7E, OR 1.90, 95% CI 1.42-2.54), and the result was statistically significant.