Meta-analysis of TEAEs associated with PD-1 and PD-L1 inhibitors

On 25 April, Yucai Wang from Mayo Clinic, Rochester, MN, USA, and colleagues, published in the Journal of the American Medical Association (JAMA) results from a systematic review and meta-analysis¹ of adverse events (AEs) associated with the clinical use of programmed cell death ligand 1 (PD-L1) and PD-1 inhibitors.

PD-1 and PD-L1 inhibitors has been widely used to treat various types of cancers, with nivolumab and pembrolizumab (anti-PD-1), as well as atezolizumab, avelumab, and durvalumab (anti-PD-L1) having been approved by the Food and Drug Administration (FDA) for different indications². Given the widespread use of such inhibitors against cancer, the authors sought to evaluate their safety profile among different pathologies from published clinical trials.

Study design

• Literature search was conducted from 01 October 2017 to 15 December 2018
• Search engines used:
  • PubMed
  • Web of Science
  • Scopus
  • Embase
• Search engine terms:
  • Nivolumab
  • Pembrolizumab
  • Atezolizumab
  • Avelumab
  • Durvalumab
  • PD-1 inhibitor
  • PD-L1 inhibitor
• Bayesian multilevel regression models were used for data analysis
• The incidence probability of AEs was adjusted by study characteristics like:
  • Dosing schedule
  • Therapeutic regimen
  • Cancer type
  • AE type
• The meta-analysis focused on AEs that were either reported in at least 10% of the studies or were possibly immune-related AEs (irAEs)

Key findings

• Literature search identified 5179 relevant publications:
  • After screening: 125 studies were finally included in the meta-analysis involving N = 20,128 patients in total
• PD-1 and PD-L1 inhibitors used in these studies included:
  • Nivolumab, n = 46 publications
  • Pembrolizumab, n = 49 publications
  • Atezolizumab, n = 15 publications
  • Avelumab, n = 9 publications
  • Durvalumab, n = 6 publications
• Patients who developed at least one AE of any grade: 66% (n = 12,277/18,610) among 106 studies
• Patients who developed at least one Grade ≥ 3 AE: 14% (n = 2627/18,715) among 110 studies
• The meta-analysis identified 75 AEs that were most commonly observed
• Overall mean incidence of any grade AEs: 1.66% (95% CI, 1.47−86)
• Mean incidence of Grade ≥ 3 AEs: 0.11% (95% CI, 0.08−14)
• Most common AEs across studies:

Any grade AE	Incidence	95% CI
Fatigue	18.26%	16.49−20.11
Pruritus	10.61%	9.46−11.83
Diarrhea	9.47%	8.43−10.58
Grade ≥ 3 AE	Incidence	95% CI
Fatigue	0.89%	0.69−1.14
Anemia	0.78%	0.59−1.02
Aspartate aminotransferase (AST) increase	0.75%	0.56−0.99

• Most common endocrine irAEs across studies:

Any grade endocrine irAEs	Incidence	95% CI
Hypothyroidism	6.07%	5.35−6.85
Hyperthyroidism	2.82%	2.40−3.29
Hyperglycemia	1.20%	0.91−1.55
Thyroiditis	0.75%	0.52−1.04
Adrenal insufficiency	0.69%	0.50−0.93
Grade ≥ 3 endocrine irAEs	Incidence	95% CI
Hyperglycemia	0.24%	0.13−0.38
Adrenal insufficiency	0.18%	0.10−0.30
Type I diabetes	0.18%	0.10−0.30
Hypohysitis	0.16%	0.09−0.27
Hypothyroidism	0.08%	0.04−0.13

• Most common other irAEs across studies:

Any grade other irAEs	Incidence	95% CI
Diarrhea	9.47%	8.43−10.58
AST increase	3.39%	2.94−3.89
Vitiligo	3.26%	2.80−3.79
Alanine aminotransferase (ALT) increase	3.14%	2.71−3.62
Pneumonitis	2.79%	2.39−3.23
Colitis	1.24%	0.99−1.54
Grade ≥ 3 other irAEs	Incidence	95% CI
AST increase	0.75%	0.56−0.99
ALT increase	0.70%	0.52−0.93
Pneumonitis	0.67%	0.50−0.89
Diarrhea	0.59%	0.45−0.77
Colitis	0.47%	0.34−0.65

• AEs that were more likely to be severe (high risk ratio [RR]: incidence of AE with Grade ≥ 3 divided by any grade AE incidence) included:
  • Hepatitis: RR = 50.59%
  • Lipase increase: RR = 42.01%
  • Gamma-glutamyltransferase increase: RR = 41.96%
  • Type I diabetes: RR = 41.86%
  • Colitis: RR = 37.90%
• Treatment-related deaths were reported in 89.6% of the evaluated studies (112/125)
• Overall incidence of treatment-related deaths: 0.45%
• Most common causes of treatment-related deaths (n = 82) were:
  • Pneumonitis: 28.0%
  • Pneumonia: 6.1%
  • Sepsis: 8.5%
  • Respiratory failure: 6.1%
  • Cardiovascular failure: 3.7%
• In total, respiratory causes accounted for the most treatment-related deaths (48%)
• When PD-1 and PD-L1 inhibitors were compared:
  • Nivolumab (3 mg/kg every two weeks) led to higher mean incidence of any grade (OR = 1.28; 95% CI, 0.97−79) and Grade ≥ 3 (OR = 1.30; 95% CI, 0.89−2.00) AEs compared to pembrolizumab (10 mg/kg every two weeks) and when compared to PD-L1 inhibitors (OR = 1.81; 95% CI, 1.04−3.01)
  • PD-1 inhibitors led to a higher overall mean incidence of Grade ≥ 3 AEs when compared to PD-L1 inhibitors (OR = 1.58; 95% CI, 1.00−2.54)

Conclusions

This meta-analysis has identified the most common AEs occurring with PD-1 and PD-L1 inhibitors when used for the treatment of various types of cancers in clinical trials. This systematic analysis will aid in the future management of such treatments and provides a useful tool for clinicians.