Nivolumab Plus Ipilimumab vs Nivolumab Alone in Advanced Cancers Other Than Melanoma

Key Points Question Does combination therapy of standard-dose nivolumab and ipilimumab improve clinical outcomes and justify additional toxicity compared with nivolumab monotherapy in advanced cancers other than metastatic melanoma? Findings In this meta-analysis of 8 clinical trials including 1727 patients, the addition of ipilimumab to standard-dose nivolumab was not associated with a clinically meaningful improvement in overall survival or progression-free survival over standard-dose nivolumab alone. The combination was associated with substantially higher treatment-related high-grade toxicities without commensurate clinical benefit. Meaning Combination therapy with nivolumab and ipilimumab may be unnecessary in many patient populations, and nivolumab or other anti–programmed death 1–directed therapy alone may deliver equivalent clinical outcomes with lower toxicity (clinical and financial) in many advanced cancers other than melanoma.


Search Strategy
We searched PubMed, EBSCO Information Services, Embase, and The Cochrane Library databases for relevant English-language papers and abstracts that had been published by October 31, 2022.The following search terms were used: "nivolumab plus ipilimumab versus nivolumab."We also performed a manual search to find applicable studies in the references and related citations.

Eligibility and Inclusion Criteria
We included studies that fulfilled the following criteria: (a) population: patients with metastatic/advanced solid malignancies, excluding metastatic/advanced melanoma; (b) intervention: the following dose regimens ("nivo3 ipi1") were allowed: nivo 3 mg/kg every 2 weeks plus ipi 1mg/kg every 6 weeks until progression; nivo 3 mg/kg every 3 weeks plus ipi 1 mg/kg every 3 weeks for four doses followed by nivo 3 mg/kg every 2 weeks maintenance (c) control: the following doses were allowed: nivo monotherapy at a dose of 3 mg/kg every 2 weeks; flat dosing of nivo 240 mg every 2 weeks; (d) prospective studies-phase I, I/II, II or III clinical trials.
Studies of advanced/ metastatic melanoma were excluded.We also excluded studies or arms within studies that had combination dosing of nivo+ipi at doses other than as defined as nivo 3 ipi 1 as above.For example, excluded doses include nivo 1 mg/kg every 2 weeks plus ipi at a dose of 3 mg/kg every 6 weeks.Nivo1 ipi3 is a dosing regimen that is known to be even more toxic than nivo3 ipi1 and is approved in only one malignancy outside of melanoma thus far (hepatocellular carcinoma), so was excluded from the analysis for homogeneity.

Outcome Measures
The primary outcome measures were overall survival (OS), progression free survival (PFS), grade 3 or 4 adverse events (AEs), and treatment related discontinuations.Detailed information about grade 3-4 adverse events were also extracted and compared including hepatotoxicity, gastrointestinal toxicity, pneumonitis, endocrine dysfunction, dermatitis, and fatigue.

Data Extraction and Synthesis
Two reviewers independently screened the titles and abstracts of the retrieved citations.Discrepancies were resolved by discussions.A standardized extraction form was prepared using Microsoft Excel.The extracted data included first author, year of publication, study design, patient population, trial phase, study title, treatments, patients in each treatment arm, total patients in the study, median patient age by treatment arm, eastern cooperative oncology group (ECOG) performance status of patients by arm, median OS, PFS with 95% CI, treatment related discontinuations and detailed information on grade 3-4 adverse events.Summary statistics of Kaplan-Meier curves (OS and PFS) were extracted separately, also using Microsoft Excel.
'difference in restricted mean survival time' (dRMST)-based meta-analysis; the robustness of log-rank even with small departures from proportional hazards; the included studies not having major departures; the high 'event rate' in each included study given the patient population of advanced malignancies, put together, led to the determination that estimation of HR (and ln(HR), SE) for each individual study with K-M summary statistics followed by computing the pooled HR with inverse variance weighting (using the RevMan software) would be the more robust method overall for this meta-analysis.Extreme care was taken in the estimation of number of events and number censored for each specified time interval (along with the published number at risk), ensuring that the K-M curves generated from these numbers matched the published K-M curves, as explained in Tierney et al. Trials, 2007 (ref 4).The snips of calculation spreadsheets for both OS and PFS HR and the K-M curves generated have been provided for each included study (as a separate supplement file), for transparency and to enable comparison to the published K-M curves.
The estimation of pooled OR for dichotomous data (treatment-related high-grade adverse events and treatment-related discontinuations) using the Mantel-Haenszel method was also performed using the RevMan software.

Assessment of heterogeneity:
The I-square (I2) test was used to assess impact of study heterogeneity.None of the analyses had an I 2 > 50% (i.e., no severe heterogeneity) and the main outcome measures (OS, PFS, grade 3-4 AE, treatmentrelated discontinuations) had an I 2 ≤ 2%; therefore, the fixed effect model was chosen for all analyses.

Search Results and Study Characteristics
The initial search identified 168 publications.After exclusion of duplicates and unrelated publications, 57 publications remained.Of these, 36 were discarded after reading the titles and abstracts.After assessing the full text, 13 reports were further excluded and 8 studies were included for data analysis.(Process outlined in the PRISMA schema in Figure S1).While 3 of the studies included treatment arms dosing both nivo 3 ipi 1 as well as nivo 1 ipi 3 (Sharma, 2019;Janjigian, 2018;Antonia, 2016;Omuro, 2017), we only included data from the nivo 3 ipi 1 arms as described above.The control group of nivo monotherapy was either 3 mg/kg or 240 mg every 2 weeks.