Jennifer Strouse ¹ Karmela Kimi Chan ² Rachel Baccile ³ Gong He ⁴ Diana K.N. Louden ⁵ Mihai Giurcanu ⁶ Arohi Singh ⁷ John Rieth ⁸ Noha Abdel-Wahab ^10,9 Tamiko Katsumoto ¹¹ Namrata Singh ¹² Sherin Rouhani ¹³ Pankti Reid ^14*

• ¹ Division of Immunology, The University of Iowa, Iowa City, Iowa, United States
• ² Division of Rheumatology, Hospital for Special Surgery, New York, New York, United States
• ³ Center for health and The Social Sciences, University of Chicago Medicine, Chicago, United States
• ⁴ Division of Hematology and Oncology, University of Michigan, Ann Arbor, Michigan, United States
• ⁵ University Libraries, University of Washington, Seattle, Washington, United States
• ⁶ Department of Public Health Sciences, The University of Chicago, Chicago, Illinois, United States
• ⁷ University of Chicago Medicine, Chicago, United States
• ⁸ Division of Hematology, Oncology, and Blood & Marrow Transplantation, The University of Iowa, Iowa City, Iowa, United States
• ⁹ Section of Rheumatology & Clinical Immunology, Department of General Internal Medicine, and Department of Melanoma Medical Oncology, University of Texas MD Anderson Cancer Center, Houston, Texas, United States
• ¹⁰ Department of Rheumatology & Rehabilitation, Assiut University Hospitals, Assiut, Egypt
• ¹¹ Division of Immunology and Rheumatology, Department of Medicine, Stanford University, Stanford, California, United States
• ¹² Division of Rheumatology, University of Washington, Seattle, Washington, United States
• ¹³ Massachusetts General Hospital, Massachusetts General Hospital Cancer Center, Boston, Massachusetts, United States
• ¹⁴ Division of Rheumatology, Department of Medicine, Committee on Clinical Pharmacology and Pharmacogenomics, University of Chicago Medicine, Chicago, United States

Background The impact of steroid-sparing immunosuppressive agents (SSIAs) for immune-related adverse events (irAEs) on tumor outcome is not well-known. This systematic review evaluates tumor outcomes for corticosteroid (CS) monotherapy versus CS with SSIA (CS-SSIA) for irAE treatment with a focus on melanoma. Methods Search was conducted through 1/5/23 using PubMed, Embase, Cochrane CENTRAL, and Web of Science. We included case series, retrospective/prospective observational studies and interventional clinical trials. Individual-level data was analyzed using KM curves and Cox regression for overall survival (OS) and progression free survival (PFS). Time to SSIA was treated as a time-varying exposure using landmark analysis (landmark timepoint=3 months after irAE) to account for immortal time bias. For group-level data, meta-analysis compared the use of SSIA to No SSIA for irAEs. Results Of twenty-two publications with individual-level data, 147 patients with any cancer (57 CS, 90 CS-SSIA) and 65 with melanoma (18 CS, 47 CS-SSIA) underwent landmark analysis. Twenty-two publications underwent group-level evaluation and four were included in the meta-analysis. CS-SSIA versus CS showed higher risk of all-cause mortality and progression (HR 2.75, 95%CI: 1.44-5.27, p<0.01 and HR 1.75, 95%CI: 1.07-2.85, p=0.03, respectively). Melanoma showed worse OS and PFS for CS-SSIA versus CS (HR 5.68, 95%CI: 1.31-24.67, p=0.02 and HR 2.68, 95%CI: 1.12-6.40, p=0.03, respectively). In the meta-analysis of group-level data (n=2558), we found worse OS and PFS for CS-SSIA versus No SSIA (HR 1.58, 95%CI: 1.25; 2.01, p<0.01 and 1.70, 95%CI: 1.25-2.33, p<0.01). Tumor necrosis factor-alpha inhibitors (TNFi) were the most common SSIA. In the melanoma cohort, TNFi had worse OS and PFS versus CS (HR 6.46, 95%CI: 1.43-29.19, p = 0.02 and HR 7.49, 95%CI: 2.29-24.48, p<0.01, respectively). TNFi versus Other SSIAs showed a trend toward worse OS and worse PFS (HR 6.96, 95%CI: 0.90-53.65, p=0.06 and HR 21.5, 95%CI: 2.63-175.8, p<0.01, respectively). Meta-analysis showed a concern for TNFi compared to Other SSIA (HR 1.56, 95%CI: 1.17-2.09, p<0.01 respectively). Conclusions While our results raise concern about the effects of CS-SSIA and TNFi for irAE therapy on tumor outcomes, prospective randomized controlled trials are needed to definitively assess the effect of SSIAs on tumor outcomes.