Immunotherapy vs. Chemotherapy in Subsequent Treatment of Malignant Pleural Mesothelioma: Which Is Better?

(1) Background: Malignant pleural mesothelioma (MPM) is a rare but aggressive tumor arising from the pleural surface. For relapsed MPM, there is no accepted standard of- are for subsequent treatment. Thus, we aimed to compare the efficacy of chemotherapy, targeting drugs, and immune-checkpoint inhibitors (ICIs) as subsequent therapy for relapsed MPM. (2) Methods: The study was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). We searched several acknowledged databases. Primary outcomes were defined as overall median progressive survival (mPFS) and median overall survival (mOS) in different treatment groups. Secondary outcomes were defined as objective response rate (ORR), the proportion of stable disease (SD), and progressive disease (PD). (3) Results: Ultimately, 43 articles were selected for the meta-analysis. According to the results of a pooled analysis of single-arm studies, ICIs showed a slight advantage in mOS, while chemotherapy showed a slight advantage in mPFS (mOS: 11.2 m vs. 10.39 m and mPFS: 4.42 m vs. 5.08 m for ICIs group and chemotherapy group, respectively). We identified only a few studies that directly compared the efficacy of ICIs with that of chemotherapy, and ICIs did not show significant benefits over chemotherapy based on mOS. (4) Conclusions: Based on current evidence, we considered that immunotherapy might not be superior to chemotherapy as a subsequent therapy for relapsed MPM. Although several studies investigated the efficacy of ICIs, targeting drugs, and chemotherapy in relapsed MPM, there was still no standard of care. Further randomized control trials with consistent criteria and outcomes are recommended to guide subsequent therapy in relapsed MPM and identify patients with certain characteristics that might benefit from such subsequent therapy.


Introduction
Malignant pleural mesothelioma (MPM) is a rare but aggressive tumor arising from the pleural surface, with one-year median overall survival (mOS) and about 2500 new cases per year in America [1][2][3]. The most common cause of the disease is asbestos exposure. Three histological sub-types encompass epithelioid, sarcomatoid mesothelioma, and biphasic mesothelioma. Because of its insidious onset, most patients are diagnosed with advanced disease and lose their chance for surgery, leading to a poor prognosis [4]. For unresectable MPM, a regimen of pemetrexed (Pem) and cisplatin (Cis) was approved as the standard of care in first-line treatment by the FDA in 2004 [5]. Currently, numerous studies are being conducted to explore the efficacy of novel agents and regimens for MPM first-line treatment. Fortunately, bevacizumab, nivolumab, and ipilimumab have improved patients' prognosis and are recommended as first-line treatment options [6,7].
However, there is no accepted standard-of-care for subsequent treatment; recommended options include pemetrexed, gemcitabine, vinorelbine, and some ICIs. Although

Data Extraction and Study Outcomes
We screened the title and abstract to identify eligible articles and then assessed the full text to select appropriate articles for qualitative and quantitative analysis.
We collected data from the literature as follows: first author, years of publication, study design, number of cases, previous treatment, current therapy patients received in the study, median follow-up time, patients' best response to current therapy, median progression-free survival (mPFS)/time to progression (mTTP), median overall survival (mOS), and toxicities, if reported. Patients' best response to current therapy included complete response (CR), partial response (PR), stable disease (SD), progression disease (PD), and death. Objective response rate (ORR) was defined as a proportion of CR and PR.
Primary outcomes were defined as overall mPFS and mOS in different treatment groups. Secondary outcomes were defined as a proportion of ORR, SD, and PD.

Risk of Bias for Articles in the Meta-Analysis
We assessed the risk of bias for eligible articles. For single-arm studies, the methodological index for non-randomized studies (MINORS) was applied. The Newcastle-Ottawa Quality Assessment Scale (NOS) was utilized for cohort studies, which includes eight items and has a total score of nine. As for RCTs, the Jadad Scale was implemented to assess any risk of bias. After reviewing the full text carefully, scores were given to each eligible article. Articles were considered as having a low risk of bias at scores of MINORS ≥ 13, NOS ≥ 7, or Jadad Scale ≥ 3.

Statistical Analysis
All procedures were conducted with STATA SE 16.0 (StataCorp, College Station, TX, USA) and RevMan 5.3 (Cochrane, London, UK). The pooled results were reported as overall rate with 95% confidence interval (CI) for single-arm studies and mean difference (MD) with 95% CI for cohort studies and RCTs. A random model was used when pooling all effect measures. The heterogeneity test was completed by I 2 test. I 2 ≤ 50% was thought to have acceptable heterogeneity. The results are presented as forest plots.

Article Selection
Initially, 2674 articles were searched in PubMed and Web of Science. 2217 articles remained after duplicates were removed. Excluding non-English articles, we screened 2113 abstracts and then screened 428 full texts. Based on the inclusion and exclusion criteria for this study, we assessed carefully for eligibility. Finally, 43 articles were selected for the meta-analysis . The flow diagram of article selection is shown in Figure 1.

Statistical Analysis
All procedures were conducted with STATA SE 16.0 (StataCorp, College Station, TX, USA) and RevMan 5.3 (Cochrane, London, UK). The pooled results were reported as overall rate with 95% confidence interval (CI) for single-arm studies and mean difference (MD) with 95% CI for cohort studies and RCTs. A random model was used when pooling all effect measures. The heterogeneity test was completed by I 2 test. I 2 ≤ 50% was thought to have acceptable heterogeneity. The results are presented as forest plots.

Article Selection
Initially, 2674 articles were searched in PubMed and Web of Science. 2217 articles remained after duplicates were removed. Excluding non-English articles, we screened 2113 abstracts and then screened 428 full texts. Based on the inclusion and exclusion criteria for this study, we assessed carefully for eligibility. Finally, 43 articles were selected for the meta-analysis . The flow diagram of article selection is shown in Figure 1.

Characteristics of Included Studies
All included studies are described in Tables 1 and 2. Most of the included studies were single-arm studies, while five [26,31,42,44,48] were RCTs, and one [45] was a cohort study. The single-arm studies mainly assessed the efficacy and toxicities of chemotherapy

Characteristics of Included Studies
All included studies are described in Tables 1 and 2. Most of the included studies were single-arm studies, while five [26,31,42,44,48] were RCTs, and one [45] was a cohort study. The single-arm studies mainly assessed the efficacy and toxicities of chemotherapy drugs (such as gemcitabine, vinorelbine, and irinotecan), targeting drugs (such as sorafenib, and dasatinib), and ICIs (such as tremelimumab, ipilimumab, and nivolumab). Among four RCTs, two compared ICIs and placebo, and one compared ICIs and chemotherapy drugs. The retrospective cohort study compared the efficacy of second-line immunotherapy and chemotherapy in real-world patients. , and gemcitabine; ICIs: immune checkpoint inhibitors. *: The methodological index for non-randomized studies (MINORS) was applied to assess single-arm studies. **: Jadad Scale was applied to assess RCTs. ***: The Newcastle-Ottawa Quality Assessment Scale (NOS) was applied to assess cohort studies.

Risk of Bias
The risk-of-bias assessment is detailed in Table 1. Only one single-arm study was considered high-risk, for it did not describe its sample size calculation, and the follow-up period was not long enough.
We identified only a few studies that directly compared the efficacy of ICIs with that of chemotherapy or placebo (Table 3). We found that targeted therapy showed superior mOS than placebo (MD: 5.58, 95%CI: 4.31-6.85, I 2 = 0%, Figure 4B), while ICIs did not show significant benefits over chemotherapy based on mOS ( Figure 4A).

Discussion
Most patients with MPM are diagnosed with advanced disease due to its insidious onset and receive chemotherapy with or without immunotherapy or targeted therapy. For patients with early-stage MPM, a multimodality treatment is the gold-standard therapy, which includes surgery and chemotherapy, with or without radiotherapy. Hyperthermic intrathoracic chemotherapy might also be an effective procedure to improve surgical radicality, resulting in a better OS [50]. However, most patients may experience disease progression and need to receive subsequent treatments.
In this meta-analysis, we pooled and compared the efficacy of different subsequent treatments for relapsed MPM, including chemotherapy, ICIs, and targeting drugs. Particular, we put an emphasis on the efficacy of ICIs and chemotherapy based on available data and found that ICIs might not be superior to chemotherapy as subsequent therapy in relapsed MPM.
The standard-of-care for MPM in first-line treatments has been modified based on clinical trials. Regimens recommended by NCCN include pemetrexed plus cisplatin with or without bevacizumab and nivolumab plus ipilimumab. However, regimens in subsequent lines remain controversial. In the past decades, physicians have conducted clinical trials to assess and compare different chemotherapy drugs, including gemcitabine, vinorelbine, oxaliplatin, cyclophosphamide, and etoposide. While ICIs and targeting drugs have recently shown significant efficacy in other malignancies, some investigators have also tried to explore the efficacy of certain agents for relapsed MPM, including pembrolizumab, nivolumab, tremelimumab, ipilimumab, avelumab, and belinostat. Unfortunately, few studies have shown inspiring results, and there are few studies comparing new regimens with commonly recommended chemotherapy.
This meta-analysis demonstrated that ICIs might not show superior effects over chemotherapy as subsequent treatment for relapsed MPM. According to the results of our pooled analysis of single-arm studies, ICIs showed a slight advantage in mOS, while chemotherapy showed a slight advantage in mPFS (mOS: 11.2 m vs. 10.39 m and mPFS: 4.42 m vs. 5.08 m for ICIs group and chemotherapy group, respectively). Moreover, patients receiving chemotherapy showed lower PD rates. Nevertheless, the study designs of the pooled single-arm studies were not the same, and confounding factors were hard to adjust. Thus, RCTs and cohort studies were needed to directly compare their efficacy.
RCTs or cohort studies are shown in Table 3, with only two studies comparing chemotherapy and ICIs. The PROMISE-meso trial compared pembrolizumab with gemcitabine/vinorelbine and demonstrated that pembrolizumab was not superior to chemotherapy in PFS and OS [42]. It also found no relationship between the efficacy of ICIs and the extent of PD-L1 expression. In the retrospective cohort study, chemotherapy included gemcitabine ± vinorelbine, while ICIs included pembrolizumab and nivolumab ± ipilimumab [45]. It found that second-line ICIs showed significantly improved OS. Based on the results of the two studies, the forest plot demonstrated that ICIs did not show significant benefits over chemotherapy in mOS ( Figure 4A). Several factors might explain this. Based on the results of basic research, ICIs function through inflammatory microenvironments, but tumor types of genomic losses, microsatellite instability, and low tumor mutation burden might contradict this [51]. In this way, the efficacy of ICIs might be reduced, and their benefits compared with chemotherapy might be weakened. In clinical practice, patients who became refractory to first-line chemotherapy were normally considered insensitive to subsequent chemotherapy. However, few studies have reported the median duration of response to previous chemotherapy, which might obscure the efficacy of second-line chemotherapy and narrow the difference between chemotherapy and ICIs. Moreover, patients in the cohort study were older than those in the RCT. In real-world settings, patients' performance status, response to prior chemotherapy, expression of PD-L1, and economic situations might be considered when choosing between ICIs or chemotherapy. These factors might indeed influence outcomes. Hence, further studies should focus on these factors to identify the potential groups of patients that might benefit from subsequent treatments.
Regardless, any kind of therapy other than placebo may be beneficial for mOS and mPFS in second-line treatment for relapsing MPM ( Figure 4B-E).
To our knowledge, this is the first meta-analysis to directly compare the efficacy of ICIs and chemotherapy as subsequent treatment in relapsed MPM based on survival data. We integrated the most up-to-date evidence and demonstrated that ICIs might not be superior to chemotherapy in subsequent therapy.
Nevertheless, there are several limitations. First of all, most enrolled studies were single-arm studies. Only one RCT and one cohort study compared subsequent ICIs and chemotherapy. Secondly, outcomes of those studies were not the same, and potential bias might influence the pooled analysis. Thus, more RCTs and cohort studies with high-level evidence and consistent outcome definitions are urgently needed to validate our results.
To conclude, this study demonstrated that ICIs might not be superior to chemotherapy as subsequent therapy in relapsed MPM. Although several studies investigated the efficacy of ICIs, targeting drugs, and chemotherapy in relapsed MPM, there remains no standard of care. Nonetheless, just as ICIs and antiangiogenics drugs have been recommended for first-line treatment, novel treatments may attenuate negative outcomes from therapy. Thus, we recommend that more RCTs with consistent criteria and outcomes be conducted to guide subsequent therapy in relapsed MPM and identify patients with certain characteristics that might benefit from such subsequent therapy.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: All data generated or analyzed during this study are included in this published article.

Conflicts of Interest:
The authors declare no conflict of interest.