Antitumor Activity and Treatment-Related Toxicity Associated With Nivolumab Plus Ipilimumab in Advanced Malignancies: A Systematic Review and Meta-Analysis

Combining immune checkpoint inhibitors has shown its efficacy compared to monotherapy in advanced malignancies. We conducted this meta-analysis to provide latest evidence on the objective response rate (ORR) and incidence of treatment-related high-grade adverse events (AEs) during nivolumab and ipilimumab combination treatment and further explore from different drug dose level. PubMed and the 2019 American Society of Clinical Oncology (ASCO) annual meeting abstracts were searched for qualified clinical trials up to June 2019. Of the 23 clinical trials (13 from publications and 11 from ASCO abstracts) included, 2,114 and 2,674 patients were eligible for efficacy and safety analysis, respectively. Pooled analysis suggested that the overall ORR was achieved in 34.5% [95% confidence interval (CI), 29.1–40.4%] of patients. There was no significant difference between nivolumab 3 mg/kg plus ipilimumab 1 mg/kg every 3 weeks (N3I1-Q3W) and nivolumab 1 mg/kg plus ipilimumab 3 mg/kg every 3 weeks (N1I3-Q3W) arms in ORR [30.8% vs 41%; odds ratio (OR), 0.72; 95% CI, 0.39–1.30; P = 0.275]. Grade 3–4 AEs related to combination therapy occurred in 39.9% (95% CI, 33.5–46.7%) of patients; the most commonly reported grade 3–4 treatment-related AEs were diarrhea (5.28%), colitis (3.96%) and increased alanine aminotransferase (3.51%). Incidence of grade 3–4 AEs were significant lower in N3I1-Q3W arm than in N1I3-Q3W arm (31.3% vs 55.9%; OR 0.52; 95% CI, 0.32–0.87; P = 0.012). Treatment-related death was rare and occurred in 2.0% (95% CI, 1.5–2.7%) of patients. Our comprehensive study provides more precise data on the incidence of treatment-related high-grade AEs and ORR among patients receiving nivolumab and ipilimumab combination regimens. Patients on the N3I1-Q3W arm had comparable ORR and significantly occurred less grade 3–4 AEs than patients on the N1I3-Q3W arm. Our finding is of great importance in assisting clinical trial design and clinical medication choice.


INTRODUcTION
Therapeutic strategies for advanced cancers have dramatically evolved over the past decade. As the traditional chemotherapy gradually couldn't achieve satisfied clinical outcomes in some clinical settings, immune checkpoint inhibitors (ICIs), which specifically target cytotoxic T lymphocyte antigen-4 (CTLA-4) and programmed death-1/ligand-1 (PD-1/PD-L1), have largely altered the treatment predicament in various advanced cancer types (Martins et al., 2019). Compared with monotherapies, combined use of anti-CTLA-4 and anti-PD-1/PD-L1 appears to exert durable response and longer survival benefit in a large proportion of advanced cancer patients (Hodi et al., 2016;Wolchok et al., 2017;Hellmann et al., 2018).
Among all the ICIs, ipilimumab and nivolumab are the most widely used ICI drugs till now, and these two drugs are the earliest and the most frequently used as combination regimens in clinical settings. Ipilimumab is a fully human IgG1 CTLA-4 ICI antibody which block the CTLA-4-B7 interaction and nivolumab is a fully human IgG4 PD-1 ICI antibody which can block the PD-1-PD-L1 interaction between T cells and tumor cells. Both of these two drugs can enhance the T-cell function through different ways in depleting tumor cells and thus might induce clinical response in cancer patients (O'Day et al., 2007;Buchbinder and Desai, 2016).
Accumulating clinical trials has been initiated to evaluate the clinical outcomes of nivolumab plus ipilimumab across various tumor types such as melanoma (Tawbi et al., 2018), lung cancer (Hellmann et al., 2018), renal cell carcinoma (Motzer et al., 2018) and colorectal cancer (Overman et al., 2018). Nevertheless, merely focusing on the response rate and survival benefit brought by the combination use seems insufficient, the treatment-related adverse events (AEs) or immune-related AEs also occur during the ICI treatment. Some AEs were slight and unrecognizable, while other AEs such as grade 3 or more AEs were severe and might lead to treatment discontinuation, hospitalization, and even death (Martins et al., 2019). The frequency and spectrum of high-grade AEs during nivolumab plus ipilimumab combination treatment, however, have not been well investigated. A recent meta-analysis showed that the immunotherapy combination could produce more clinical benefits while with increased highgrade AEs (Wei et al., 2019). Subsequent question was raised that how we clinicians can formulate an optimal combination regimen in reducing the incidence of treatment-related highgrade AEs while not compromising its efficacy at the same time.
Herein, by reviewing the latest evidence in cancer immunotherapy progress, we conducted this meta-analysis trying to exhibit the frequency and spectrum of high-grade/fatal AEs and the objective response rate (ORR) related to nivolumab and ipilimumab combination therapy. We also sought to further explore the outcomes from different drug dose level.

Search Strategy
We systematically searched the PubMed database to identify the clinical trials that investigated the combined nivolumab and ipilimumab use in cancer patients and report the related results without language restrictions. Besides, the 2019 American Society of Clinical Oncology (ASCO) annual meeting abstracts were also retrieved as potential sources. For PubMed search, the following keywords were used: "Ipilimumab, " "Yervoy, " "MDX-010, " "BMS-734016, " "nivolumab, " "Opdivo, " "BMS-936558, " "MDX1106. " PubMed search was up to June 1, 2019. We only searched the nivolumab and ipilimumab because they are the most frequently used combined ICIs in clinical trials.

Study Selection
We applied the Population, Intervention, Comparator, Outcome, and Study design (PICOS) approach to identify eligible studies. Clinical trials (S) that investigated nivolumab and ipilimumab combination use (I, C) in advanced cancer patients (P), and provided information on ORR and high-grade AEs (O) were selected. We included all the prospective clinical trials that meet the following items: (1) investigating the combined use of nivolumab and ipilimumab in patients with advanced solid tumors; (2) with results that reported the ORR/the incidence of treatment-related grade 3-4 AEs/the number of treatment-related death; (3) the 2019 ASCO annual meeting abstracts were included if they meet the above two criteria. We excluded trials that: (1) involved combination regimens with other treatment modalities (e.g. nivolumab plus ipilimumab plus radiotherapy); (2) investigated the neo-adjuvant nivolumab combined with ipilimumab in cancer patients; (3) were quality of life analysis or cost-effective assessment of the trials; (4) the results didn't report the specific number or rate of objective response and AEs data. Besides, case reports, editorials, letters and correspondences were excluded. Review and systematic review were screened for potential omitted qualified trials despite they were excluded from our study. In the event of duplicated trials, we selected the most recent trials into our study. Discrepancies regarding the inclusion and exclusion criteria were resolved by consensus (Figure 1).

Data extraction
The data were extracted by 1 reviewer (HX) primarily and were reviewed by another reviewer (PT) following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Efficacy and safety data were separately extracted from trial results. The number of events (objective response, treatment-related grade 3 or more AEs) were recorded. Numbers of objective response were calculated as numbers of partial response + numbers of complete response. In addition, the frequency and spectrum of treatment-related grade 3-4 AEs and fatal AEs (i.e. one specific AE) were also recorded from publications (owing to the limited information on ASCO abstracts, they were not included in this analysis). Besides, information on first author name, ASCO abstracts number, study year, NCT number, phase, cancer type, doses and frequency of nivolumab plus ipilimumab combination, median follow-up duration were also recorded ( Table 1). For those trials that had multiple arms, we only included the nivolumab plus ipilimumab combination arms and extracted data from each arm. Data were extracted by two reviewers independently and discrepancies were resolved by discussion.

Statistical Analysis
For efficacy analysis, the number of patients available for efficacy assessment and the number of patients with objective response were recorded from each arm. For safety analysis, the number of patients available for safety assessment and the number of patients with grade 3-4 AEs or fatal AEs were also recorded from each arm. The observed ORR and incidence of treatment-related grade 3-4 or fatal AEs is reported by arm with 95% confidence intervals (CI). Fixed effects models or random effects models were selected according to the heterogeneity. Heterogeneity was assessed according to the I 2 value. The log-odds transformation and restricted maximum likelihood estimation were applied in all models. Besides, the 0.5 adjustment were applied to handle proportions equal to 0 or 1. Meta regression included four variables (sources [publications vs ASCO abstracts], sample size [≥ 100 vs <100], cancer type and different drug dose. Odds ratio [OR] and its corresponding 95% CI were calculated as exponentiate the results from the meta-regression models. Statistical significance was considered as two-side P <0.05. All analyses were conducted using the "meta-for" and "meta" package from R 3.6.0 (R project).
Of all the trials included, 4, 2, 13, 4 and 1 studies were phase1, phase 1/2, phase 2, phase 3 and phase 3b/4 clinical trial, respectively. For each trial we only included cohorts with nivolumab plus ipilimumab arm, which resulted in 2,114 and 2,674 patients were eligible for efficacy and safety analysis, respectively. The most common cancer types were melanoma (six clinical trials, nine cohorts), lung cancer (five clinical trials, seven cohorts) and renal cell carcinoma (three clinical trials, five cohorts). The most commonly selected dose combination was nivolumab 3 mg/kg plus ipilimumab 1 mg/kg every 3 weeks (N3I1-Q3W, 12 cohorts) and nivolumab 1 mg/kg plus ipilimumab 3 mg/kg every 3 weeks (N1I3-Q3W, 11 cohorts). The median follow-up duration ranged from 3.1 months to 27.2 months. The baseline characteristics of trials included in this study can be seen in Table 1.

DIScUSSION
This meta-analysis investigated the efficacy and safety related to nivolumab and ipilimumab combination therapy in advanced cancer patients. The results showed that roughly 1/3 patients received combined nivolumab and ipilimumab therapy would achieve ORR; meanwhile, nearly 40% of the patients would occur grade 3-4 treatment-related AEs; treatment-related death     AEs, adverse events; ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, γ-glutamyl transpeptidase. ";" indicates multi-arms in one study by row; Bold values indicate the incidence of a specific adverse event exceeds 1%.

Renal and Urinary Disorders
Frontiers in Pharmacology | www.frontiersin.org November 2019 | Volume 10 | Article 1300 was rare (2%). Moreover, we found that patients on the N3I1-Q3W arm had comparable ORR and significantly experience less grade 3-4 AEs than patients on the N1I3-Q3W arm, suggesting that the N3I1-Q3W regimen might be a better choice when we decided to administrate the combination therapies. By combining the latest clinical trial progress, we were able to draw the spectrum of severe and fatal treatment-related AEs associated with nivolumab plus ipilimumab regimen. Although several previous meta-analyses (Wang et al., 2018;Zhang et al., 2018;Wei et al., 2019) focused on the efficacy and safety of combination ICIs, our study is the first to investigate the estimated ORR and incidence of high-grade treatment-related AEs following the administration of nivolumab plus ipilimumab in solid tumors; moreover, our study is the first that we know of  to compare the efficacy and safety from different drug dose level in combined nivolumab and ipilimumab therapy. ICIs, including anti-CTL A-4, anti-PD-1 and anti-PD-L1 antibodies, are undoubtedly the most important progress in cancer treatment over the past decade. The indications for these drugs are continuing expanding across many clinical advanced settings, transforming many of the previous standard treatment modalities and bringing new dawn to traditionally "incurable" patients. Clinical evidence has shown the fact that nivolumab combined with ipilimumab could bring more durable responses compared with either agent alone in melanoma or lung cancer patients (Larkin et al., 2015;Antonia et al., 2016b;Hodi et al., 2016;Wolchok et al., 2017). A meta-analysis also concluded that combination ICIs could bring more ORR, progression-free survival (PFS) and overall survival (OS) benefits compared to control arms (Wei et al., 2019). The reason why we select ORR as the main indicator for efficacy of nivolumab and ipilimumab combination rather than PFS or OS is that most of the included studies didn't meet the OS end-point and the definition of PFS is not consistent across various tumor types. The best ORR can be achieved 59% in melanoma in the trial conducted by Hodi et al. (2016). Our study also shows that 47% of melanoma patients receiving nivolumab and ipilimumab combination therapy can achieve complete or partial response. Then comes renal cell carcinoma, in which ORR could be achieved in around 43% of patients. Lung cancer patients only had 31% objective response benefit. Our results might be helpful in patients' selection when the combination ICIs being an option.
Despite combined ICIs therapy showed its efficacy compared to ICI monotherapy in malignancies, however, the treatmentrelated AEs or immune-related AEs increased accordingly. In a comprehensive network meta-analysis performed by Xu et al., they provided a safety ranking of ICIs in cancer treatment (Xu et al., 2018). Their results demonstrated the pooled incidence of all grade AEs in ICIs combination was 57.7%, while in nivolumab was 14.4% and in ipilimumab was 25.2%. From their study we can know that combined ICIs could increase the AEs incidence, despite this 57.7% associated with ICIs combination might be inappropriate because they only included two trials. Another limitation is that they failed to show treatment-related grade 3-4 AEs associated with ICI combinations. By pooling 30 cohorts comprising 2,664 patients we were able to provide the relatively reliable incidence of grade 3-4 AEs (roughly 40%) related to combination use of nivolumab and ipilimumab, in comparison of 46% in nivolumab and 51% in ipilimumab from Xu's study (Xu et al., 2018). From this point, the combination ICIs therapy might be acceptable and it wouldn't increase the incidence of high-grade AEs compared with monotherapy. In addition, we exhibited the toxicity spectrum of grade 3-5 AEs associated with ICI combination. In a study conducted by Zhao et al. (2018), they demonstrated the most common treatmentrelated serious AEs were pneumonitis (8.2%), interstitial lung disease (5.6%) and colitis (3.6%) related to nivolumab therapy. While our study demonstrated the most commonly reported grade 3-4 treatment-related AEs were diarrhea (5.28%), colitis (3.96%) and increased alanine aminotransferase (3.51%) in the combination therapy. As for the fatal AEs related to ICI therapy, one meta-analysis found its incidence was 1.23% associated with ICIs combination therapy (Wang et al., 2018), and in our study this index was 2.0%. Regarding the spectrum of the fatal AEs related to ICIs combination, both of us showed the cardiac disorders and pneumonitis were the major cause of treatment-related death, though they rarely happened (< 1%).
In view of the drug doses during ICIs use, previous pooled analyses showed that ipilimumab 10 mg/kg every three weeks had a higher risk of grade 3-4 AEs than 3 mg/kg every three weeks (OR, 3.08; 95% CI, 1.52-6.32) (Xu et al., 2018), and no significant differences were found regarding fatal irAEs across different doses of ipilimumab (3 mg/kg vs 10 mg/kg for ipilimumab monotherapy; 1 mg/kg vs 3 mg/kg for combination ipilimumab therapy) (Wang et al., 2018). These results demonstrated incidence of high-grade AEs ipilimumab might be dose-dependent (Weber et al., 2012;Feng et al., 2013;Eggermont et al., 2016). This might explain our results that incidence of high-grade AEs was significant higher in N1I3-Q3W arm than in N3I1-Q3W arm. N3I1-Q3W is an ideal dose combination which didn't eliminate the efficacy of combination therapy but rather decrease the incidence of grade 3-4 AEs.
Limitations of this study should be stated as well. We performed this meta-analysis from the study level; thus, we were unable to analyze the patient level variables such as patients' sex and previous drug consuming that might affect the outcomes of our results. In addition, a significant proportion of the trials were from ASCO annual meeting abstracts with relatively short follow-up, which might lead to underestimation of their rates and overestimation of drug safety (Saini et al., 2014). Thirdly, published studies only reported the treatmentrelated AEs with an incidence above ≥1% or ≥5%, and there were only two studies posting their results in clinicaltrials.gov (Wolchok et al., 2017;Motzer et al., 2018); even though we've collected the data from the supplementary materials, some treatment-related grade 3-4 AEs might also be omitted in this study. Fourthly, we only analyzed two typical ICIs combination (nivolumab and ipilimumab) in this study, yet the efficacy or safety profile of other ICI combination [e.g. tremelimumab plus durvalumab (Antonia et al., 2016a;Calabro et al., 2018) and pembrolizumab plus ipilimumab (Long et al., 2017)] still remain unknow.

cONcLUSIONS
In this comprehensive meta-analysis of 23 clinical trials, we provided the efficacy and complete toxicity profile and spectrum of treatment-related grade 3-4 AEs of combining nivolumab and ipilimumab in advanced cancer patients. We found that patients treated with N3I1-Q3W regimen had comparable ORR and experienced significantly less grade 3-4 adverse events than those who treated with N1I3-Q3W regimen. Our finding is of great importance in assisting clinical trial design and clinical medication choice.

DATA AvAILABILITy STATeMeNT
The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.