Determinants of SARS-CoV-2 outcomes in patients with cancer vs controls without cancer: a multivariable meta-analysis with genomic imputation

Background: SARS-CoV-2 is known to impact patients with cancer adversely. Previous meta-analyses have lacked clarity on the recency of cancer diagnosis, anti-cancer treatment durations, and SARS-CoV-2 specific variants of concern (VOC). This study aimed to compare SARS-CoV-2 multivariable-adjusted clinical outcomes between patients with cancer and those without cancer, identifying key risk factors spanning pre- and post-Omicron periods. Methods: In this systematic review and meta-analysis, we identified from Medline, Embase, Cochrane Central, and the WHO COVID-19 Research Database prospective and retrospective case–control studies and cohort studies published from 1st January 2019 to 22nd November 2024. We included case–control and cohort studies comparing at least 10 patients with active cancer (diagnosed or treated within three years prior to SARS-CoV-2 infection) to controls without cancer using multivariable analyses. Exclusion criteria included lack of clarity about active/inactive status of cancer, lack of a control group without cancer, lack of multivariate analysis comparing outcomes of interest in patients with active cancer vs patients without cancer, case reports or case series, and SARS-CoV-2 diagnosis not confirmed via laboratory testing. Outcomes measured were SARS-CoV-2 infection severity (WHO ordinal scale) and mortality differences by tumour type, treatment, and VOC (using sequencing data from NCBI Genbank and GISAID). A random-effects meta-analysis model was applied. The systematic review was PRISMA compliant and was registered with PROSPERO, CRD420234454524. Findings: Of 35,501 studies initially identified, 30 met eligibility criteria and were included in the meta-analysis, comprising 281,270 patients with cancer and 18,876,411 controls. Using the Agency for Healthcare Research and Quality (AHRQ) risk of bias standards, 21 studies were rated good, one study rated was fair, and eight studies were rated poor. We found higher mortality odds ratios (OR) in patients with cancer infected with SARS-CoV-2: 1·40 (95% CI: 1·12–1·73, I² = 98·1%) for solid tumours and 2·10 (95% CI: 1·43–3·07, I² = 97·3%) for haematological malignancies, with the difference in mortality between these groups not reaching statistical significance (Q (1) = 3·32; p = 0·0068). Amongst the solid cancers, thoracic and colorectal were linked to increased odds of mortality (ORs: 2·63 [95% CI: 1·65–4·20, I² = 98·7%], and 1·65 [95% CI: 1·26–2·15, I² = 92·7%], respectively). Metastatic cancers (OR: 3·59; 95% CI: 1·07–12·04, I² = 99·5%) were also linked to greater odds of mortality compared to localised cancers (OR: 1·76; 95% CI: 1·32–2·34, I² = 96·6%; p = 0·26). No cancer types showed a reduced risk vs controls. Mortality varied significantly among VOCs; Alpha (OR: 4·59; 95% CI: 2·66–7·92, I²: N/A) and Omicron (OR: 2·74; 95% CI: 1·84–4·09, I² = 90·2%) were more associated with death than the ancestral Wu-1 (OR: 1·43; 95% CI: 1·14–1·80, I² = 98·2%) and Delta (OR: 1·94; 95% CI: 1·65–2·29, I²:N/A) variants (X² (4) = 20·4; p = 0·0004). Interpretation: This comprehensive meta-analysis indicates that patients with active cancer with SARS-CoV-2 have a higher risk of mortality and hospitalisation than those without cancer. The risk of death was comparable between active solid and haematological tumours. SARS-CoV-2 severity and mortality risks were higher with thoracic, colorectal, or any metastatic cancers. Additionally, differences were noted in mortality risks across VOCs, diverging from VOC-associated mortality patterns in the general population. However, the strict three-year cutoff used to define active cancer excludes studies that used broader cancer criteria (i.e., any history of cancer), which may limit generalisability. Further limitations include varied definitions of disease severity, retrospective data collection, incomplete vaccination or lineage data, and significant between-study heterogeneity, potentially influencing these findings. Funding: Cancer Research UK; UK Research and Innovation.