Trial selection
Overall, 2,315 citations were identified by the researchers, and 66 potentially eligible articles were retrieved in full text. We excluded 55 reports but included 4 additional studies from other sources, resulting in 11 publications describing the efficacy of IMT between 2004 and 2018. However, we found significant negative results associated with immunopotentiators. Therefore, we concluded that immunopotentiators could not benefit HGGs and excluded 2 studies (Figure 1A). Nonetheless, the use of TMZ also had a large impact on the results. We excluded 3 studies whose SOC regimen did not contain TMZ (Figure 1B). Thus, 6 studies were used for the meta-analysis (Supplementary Figure 1).
Main characteristics of the studies
First, we studied those papers using AdvHSV-tk + GCV/DC vaccines/the TGF-β2 inhibitor trabedersen/Cpg-ODN/recombinant nonpathogenic polio–rhinovirus chimera (PVSRIPO). We classified these studies into groups according to their use of viral therapy (AdvHSV-tk + GCV/PVSRIPO), DC therapy, or immunopotentiators (trabedersen/Cpg-ODN). A total of 524 participants were provided with IMT, and 747 participants were provided with SOC (a total of 1,271 participants).
A total of 914 participants from viral therapy studies (393 patients in the experimental arm and 521 patients in the control arm) were included. Three of the studies containing TMZ in the SOC regimen also reported PFS. The details can be found in Table 1. There were three studies on DC therapy. All studies used TMZ in the SOC regimen. Ninety participants underwent DC therapy (43 patients in the experimental arm and 47 patients in the control arm) (Table 2). An immunopotentiator (IP) was used in two studies that applied TMZ in the SOC regimen (88 patients in the experimental arm and 179 patients in the control arm) (Table 3).
After preliminary qualitative synthesis, we found that HGGs did not benefit from immunopotentiators (trabedersen/Cpg-ODN). Moreover, the use of TMZ in the standard of care regimen also had a large impact on evaluating efficacy. TMZ has been included in the SOC since 2005[19]. We thought it was more practical to evaluate SOC regimens that contained TMZ.
Thus, in the final meta-analysis, we included 481 participants who were confirmed to have HGG by clinical, radiological or MRI evidence. A total of 174 participants were provided with IMT (including VT and DC vaccine), while 307 participants were provided with SOC (TMZ and/or radiotherapy).
Primary endpoints
OS
First, we employed a random-effect model to assess the efficacy of IMT versus SOC according to the HR of OS. A total of 1,271 participants in 11 studies were included in this meta-analysis. The results showed that IMT decreased the risk of death by 26% compared with the SOC (HR = 0.74, 95% CI 0.56-0.99; p = 0.0458). However, substantial heterogeneity was found (tau2=0.1315; I2= 65.5%). Patients were divided into subgroups according to immunotherapy type (Figure 2A, B). We found that the source of heterogeneity was the use of TMZ and IPs.
We employed a fixed-effect model to assess the efficacy of immunotherapies other than IPs versus TMZ-containing SOC by determining the HR of OS. A total of 481 participants in 6 studies were included in this meta-analysis. The results showed that IMT decreased the risk of death by 45% compared with the SOC (HR = 0.55, 95% CI 0.42-0.72; p < 0.0001). Substantial heterogeneity was not found (tau2=0.0544; I2= 29.7%).
Publication bias was explored with an inverted funnel plot, which showed slight asymmetry around the 95% CI cutoff, and Egger’s test showed no significant bias (p = 0.089) (Figure 2C).
Sensitivity analysis was performed to assess which study influenced our results. We did not find any study that had a potential impact on the results, and the results did not change by removing any one study (Figure 2D).
PFS
With respect to PFS, we used a fixed-effect model to assess the efficacy of IMT versus the SOC according to the HR. We pooled six trials, for a total of 316 participants (113 patients were treated with IMT, and 203 patients were treated with SOC). The results showed that IMT decreased the risk of recurrence by 41% compared with the SOC (HR = 0.67, 95% CI 0.53-0.84; p = 0.0005). Significant heterogeneity was not found (tau2 = 0.0784, p = 0.018; I2=38.7%) (Figure 3B).
Publication bias was explored with an inverted funnel plot, which showed slight asymmetry around the 95% CI cutoff, but Egger’s test showed no significant bias (p = 0.1857) (Figure 3A).
Sensitivity analysis was performed to assess which study influenced our current results. We did not find any study that had a potential impact on the results, and the results did not change by removing any one study (Figure 3C).
Secondary endpoints
Subgroup analysis
In addition to exploring the source of heterogeneity, we also tried to identify the factors influencing the efficacy of IMT versus SOC. (Table 4)
As mentioned above, dividing the studies into subgroups according to immunotherapy type manifested a statistical difference between viral therapy, DC vaccines and immunopotentiators (IPs) in the meta-analysis (p < 0.05). This result indicated that IPs did not benefit HGGs. Dividing the studies into subgroups according to SOC type showed that the use of TMZ may have some synergy.
We excluded studies whose SOC regimen did not contain TMZ and IP trials and pooled 6 trials into the meta-analysis. The results showed that randomized controlled trials had a smaller effect size than historical controlled trials. (p = 0.015; Figure 4B). Moreover, the results also suggested that the area from which the patients were recruited may be a main factor for the efficacy of IMT in HGG patients in China, and these patients may benefit more from IMT than patients in other areas of the world. (p = 0.05; Figure 4B).
In addition, subgroup analyses according to clinical stage and clinical trial’s type showed that these factors were unlikely to be the main factors influencing the efficacy of IMT (Figure 4C, 4D).
Galbraith radial plot
We drew a Galbraith radial plot and found that no study fell outside the 95% confidence interval cutoff. This hinted at homogeneity within the studies regarding viral therapy and DC therapy. (Supplementary Figure 2A)
Cumulative meta-analysis
We ordered these studies on the basis of the year of publication. The statistical results of the cumulative meta-analysis showed that the confidence interval decreased and the results became more convincing as the publication year increased.