Neoadjuvant Chemoimmunotherapy for NSCLC

This meta-analysis investigates the association of neoadjuvant chemoimmunotherapy with patient benefits vs neoadjuvant chemotherapy in patients with non–small cell lung cancer (NSCLC).

to determine eligibility for inclusion, once again by 2 independent reviewers (M.S. and C.P., L.G., K.N., K.K., M.H.S., L-.R.D., A.A., or A.S.).In all cases, disagreements were resolved by a third reviewer (M.D.).In addition, abstracts from the 2023 American Society of Clinical Oncology conference, European Society for Medical Oncology conference, European Lung Cancer Congress, World Conference on Lung Cancer, and American Association for Cancer Research conference were screened for updates on published RCTs and for results on new RCTs.The most updated online data were included, and online material from conference websites was also used.

Quality Assessment
Risk of bias was assessed by 2 independent reviewers (M.S. and K.N.) using the revised Cochrane risk of bias tool 16,17 for RCTs and the Joanna Briggs Institute checklist for single-arm studies.Disagreements were resolved by a third reviewer (M.D.).

Data Extraction
Relevant data were extracted independently by 2 researchers (M.S. and C.P., L.G., K.N., K.K., M.H.S., L-.R.D., A.A., or A.S.) from included articles using a prespecified form.Discrepancies were discussed with a third reviewer (M.D.) and were resolved through consensus.Extracted data included surgical, pathological, and efficacy outcomes, as well as adverse events (eMethods in Supplement 1).For RCTs, hazard ratios (HRs) for EFS and OS and their 95% CIs were also extracted from each included study.When available, HRs by subgroup stratified by sex, age, histology, PD-L1 level, stage, and type of platinum agent received (carboplatin vs cisplatin) were also extracted.All single arms were extracted if they met inclusion criteria.Only studies that reported the outcome of interest were included in the relevant analysis.

Statistical Analysis
Outcomes of interest included surgical (surgical resection rate and R0 resection), pathological (major pathological response [MPR] and complete pathological response [pCR]), and efficacy (EFS and OS) outcomes, as well as adverse events (SRAEs and TRAEs) (eMethods in Supplement 1).We performed analyses using metafor 18 and meta 19 packages in R statistical software version 3.4.0(R Project for Statistical Computing).
Outcome data from single-arm studies were pooled independently for neoadjuvant chemoimmunotherapy and chemotherapy by an inverse variance random-effects metaanalysis using logit transformation.Pooled proportion estimates were compared between chemotherapy and chemoimmunotherapy by univariate metaregression. 20For singlearm studies, the incidence of death was calculated using the number of death events and the median follow-up time, with a comparison done by subgroup analysis.For RCT data, we performed a restricted maximum likelihood meta-analysis of risk ratios (RRs) and HRs for time-dependent data.Subgroup analyses were exploratory.For each analysis, we included the 95% CIs, I 2 statistic, τ 2 statistic, and χ 2 .All tests were 2-sided, and a P value < .05 was considered significant unless otherwise indicated.
We also compared these clinical variables across all trials, including RCTs and nonrandomized single-arm trials. 8,10,11, For ngle-arm studies, 8,10,11, there was also a significantly higher MPR and pCR for chemoimmunotherapy than chemotherapy (eFigures 27-28 in Supplement 1). Therewas no significant difference in the pooled prevalence of patients who had a complete response or progressive disease, but there was an increase in the proportion of patients with a partial response or stable disease for chemoimmunotherapy vs chemotherapy (eFigures 29-32 in Supplement 1).
For single-arm studies, 8,10,11, no trend was observed for surgical resection (eFigure 33 in Supplement 1). Therewas a significant increase in the prevalence of R0 resections (eFigure 34 in Supplement 1) and lobectomies (eFigure 35 in Supplement 1) and a decrease in pneumonectomies for chemoimmunotherapy vs chemotherapy (eFigure 36 in the Supplement).No differences were found for the incidence of death between for chemoimmunotherapy vs chemotherapy (eFigure 37 in the Supplement).

Discussion
To our knowledge, this is the most comprehensive metaanalysis to date comparing neoadjuvant chemoimmunotherapy and chemotherapy using RCT data.Overall, we found that neoadjuvant chemoimmunotherapy was associated with improved OS, EFS, MPR, and pCR compared with neoadjuvant chemotherapy; in addition, chemoimmunotherapy was associated with improved resectability and an increased rate of R0 resections, with a similar rate of adverse events.
Our meta-analysis showed an improvement in EFS with neoadjuvant chemoimmunotherapy over neoadjuvant chemotherapy across age (≥65 and <65 years), sex (male and female), and histology (squamous and nonsquamous cancer) groups.There was an improvement for patients with stage II and stage III disease.Chemoimmunotherapy was associated with a benefit in EFS for all 3 categories for PD-L1 level (<1%, 1%-49%, and ≥50%), although the OS benefit was restricted to the subgroup with a PD-L1 level of 1% or greater based on the current maturity of OS data.This highlights that the restriction in the approval of neoadjuvant chemoimmunotherapy exclusively for patients with a PD-L1 level of 1% or greater by the European Medicines Agency was based on a subgroup analysis from a single trial (CheckMate 816) 8 and that the available evidence now suggests that this patient population may have an EFS benefit with neoadjuvant chemoimmunotherapy.It will be important to assess whether this translates into an OS benefit as the number of studies reporting OS and the follow-up time for these studies increase.
A major concern of the neoadjuvant approach is progression on therapy preventing surgery. 71 neoadjuvant chemoimmunotherapy was associated with a reduced risk of not undergoing surgery compared with chemotherapy alone owing to a reduced risk of progression precluding surgery; however, there was an increased risk in adverse events precluding surgery.Across all RCTs, 8,10,11,21,23,41,[61][62][63][64][65][66][67][68][69][70] 7.0% to 22.3% of patients were not resected in chemoimmunotherapy arms.Patient refusal was the reason precluding surgery in 1.0% to 8.9% of patients receiving chemoimmunotherapy and progression on therapy for 0% to 7.4% of these patients.While it is possible that patients who progressed on therapy could have benefited from up-front surgery, these patients were likely to develop early metastasis even if resected up front.There remains a clear gap in our understanding of the relative benefits associated with neoadjuvant vs adjuvant strategies for this patient population, although significant challenges exist around the feasibility of trials aimed at resolving this open question.Certainly, neoadjuvant chemoimmunotherapy may also be associated with downstaging of disease prior to surgery and a reduced extent of resection required for curative surgery, while an adjuvant strategy does not offer such an opportunity. 71Conversely, data in this study on resectability suggest that concerns that neoadjuvant chemoimmunotherapy may be associated with poor surgical outcomes compared with neoadjuvant chemotherapy may not be relevant.
It is worth noting that results from the General Thoracic Surgery Database indicate that 30-and 90-day mortality after neoadjuvant chemotherapy are not different from outcomes of patients undergoing up-front surgery after risk adjustment. 72deally, a meta-analysis of surgical outcomes from adjuvant studies compared with those from neoadjuvant studies may help more definitively resolve this important question if a pure neoadjuvant vs adjuvant trial fails to emerge.

Limitations
This study has several limitations.Meta-analyses of nonrandomized clinical trials are subject to a high level of bias owing to the inherent nature of indirect comparisons, which assume that trial designs and patient populations of individual arms are similar enough to be compared.Our subanalysis of nonrandomized studies may thus also be subject to confounding.However, our meta-analysis also included large RCTs comparing neoadjuvant chemoimmunotherapy with chemotherapy, which represents an important strength.Other limitations of our study include the variability in definitions for pathological and efficacy end points and differences in follow-up time.Additionally, differences in inclusion criteria may affect outcomes of individual studies, and we are not able to rule out the possibility of multiplicity and a type I error based on the large number of end points tested.Furthermore, studies varied in the number of neoadjuvant cycles, type of immunotherapy drugs (pembrolizumab, nivolumab, or durvalumab) used in combination with chemotherapy, and dose and nature of the adjuvant treatment, which could be associated with EFS and OS.Importantly, the appropriate adjuvant treatment regimen for patients treated with neoadjuvant chemoimmunotherapy remains controversial, and there is currently limited evidence to suggest the superiority of a perioperative approach compared with a pure neoadjuvant approach.However, all RCTs presented in this study had the commonality of a neoadjuvant chemoimmunotherapy approach.In terms of number of cycles, the neoSCORE trial 40 compared 2 vs 3 cycles of neoadjuvant chemoimmunotherapy and found that at 12 months, the OS rate was 92.3% in the 2-cycle arm and 86.2% in the 3-cycle arm.These findings potentially suggest that the number of neoadjuvant cycles is associated with outcomes.

Conclusions
This meta-analysis found that neoadjuvant chemoimmunotherapy was superior to neoadjuvant chemotherapy given that it was associated with improved efficacy (EFS and OS) and pathological (MPR and pCR) outcomes and increased surgical resection rate and R0 resections, without an increase in the rate of SRAEs or TRAEs.In addition, neoadjuvant chemoimmunotherapy was superior across sex, age, histology, and PD-L1 levels in EFS.This finding has increased importance given the recent restriction by the EMA of neoadjuvant chemoimmunotherapy to patients with PD-L1 levels of 1% or greater.Future studies should continue to assess the benefit associated with neoadjuvant chemoimmunotherapy by subgroup as OS matures.In addition, future studies may be able to assess whether the specific type of chemotherapy or immunotherapy is associated with outcomes for patients treated with neoadjuvant chemoimmunotherapy.