Causal effects of COVID‐19 on cancer risk: A Mendelian randomization study

In contemporary literature, little attention has been paid to the association between coronavirus disease‐2019 (COVID‐19) and cancer risk. We performed the Mendelian randomization (MR) to investigate the causal associations between the three types of COVID‐19 exposures (critically ill COVID‐19, hospitalized COVID‐19, and respiratory syndrome coronavirus 2 (SARS‐CoV‐2) infection) and 33 different types of cancers of the European population. The results of the inverse‐variance‐weighted model indicated that genetic liabilities to critically ill COVID‐19 had suggestive causal associations with the increased risk for HER2‐positive breast cancer (odds ratio [OR] = 1.0924; p‐value = 0.0116), esophageal cancer (OR = 1.0004; p‐value = 0.0226), colorectal cancer (OR = 1.0010; p‐value = 0.0242), stomach cancer (OR = 1.2394; p‐value = 0.0331), and colon cancer (OR = 1.0006; p‐value = 0.0453). The genetic liabilities to hospitalized COVID‐19 had suggestive causal associations with the increased risk for HER2‐positive breast cancer (OR = 1.1096; p‐value = 0.0458), esophageal cancer (OR = 1.0005; p‐value = 0.0440) as well as stomach cancer (OR = 1.3043; p‐value = 0.0476). The genetic liabilities to SARS‐CoV‐2 infection had suggestive causal associations with the increased risk for stomach cancer (OR = 2.8563; p‐value = 0.0019) but with the decreasing risk for head and neck cancer (OR = 0.9986, p‐value = 0.0426). The causal associations of the above combinations were robust through the test of heterogeneity and pleiotropy. Together, our study indicated that COVID‐19 had causal effects on cancer risk.


| INTRODUCTION
Coronavirus disease 2019 (COVID-19) is a pandemic caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); it has caused more than 660 million confirmed cases and 6.8 million deaths as of February 2023 (Available online: https://covid19.who.int/). 1,2 It is noteworthy that COVID-19 causes not only acute symptoms from asymptomatic or mild respiratory diseases to multi-system diseases but also its potential long-term effects of COVID-19 have gradually attracted attention, such as fatigue, shortness of breath, and loss of smell. [3][4][5][6][7] Therefore, determining the long-term effects of COVID-19 on health will provide significant guiding value for public health promotion policies.
Cancer is a severe public health problem and the world's second most significant cause of death. 8 The diagnosis and treatment of cancer patients met a severe test during the prevalence of COVID-19. For cancer patients, the systemic immunosuppressive states caused by the malignant tumor and anticancer treatment weak their ability to resist the infection of SARS-CoV-2. 9,10 Furthermore, SARS-CoV-2 infection could affect the immune status, disease progression, and routine treatment of cancer patients. 11 Cancer patients with COVID-19 have significantly increased odds of COVID-19-related mortality and all-cause in-hospital mortality than noncancer patients. 12,13 At the same time, some studies have counted the number of newly diagnosed cancers and disease stages during the COVID-19 pandemic. [14][15][16] The latest cancer statistics indicated that the level of cancer screening, the number of diagnoses, and the degree of disease in various countries during the new crown pneumonia epidemic are significantly different from those during the non-epidemic period. 14 There was a significant decline in newly diagnosed cancer patients and an increase in advanced cancer cases due to COVID-19 restrictions, reduction or delay in cancer screening tests, and clinical visits. 15,16 However, the generalization ability of the above data is intensely questioned, and it could not reflect the real impact of COVID-19 on cancer risk. Therefore, the causal associations of COVID-19 with cancer risk remain to be further studied, which is crucial to the long-term health management of millions of people infected with COVID-19.
Mendelian randomization (MR) is an epidemiological method to assess the potential causal association between exposure and outcome. 17 MR can minimize the conventional confounding and reverse causation because genetic variation is randomly distributed during meiosis, independent of environment, disease onset, and progression. 18 Several studies used MR to reveal the causal associations between COVID-19 and chronic diseases, such as type 2 diabetes. 19 MR studies also revealed many causal links for cancers, such as the gut microbiota to colorectal cancer and rheumatoid arthritis to cancers. 20,21 As to the causal links between cancer and COVID-19, only two studies revealed the effect of genetic predisposition to different types of cancers on COVID-19 susceptibility and severity, and only lung adenocarcinoma showed a causal association with COVID-19 risk. 22,23 The present study aimed to explore the potential causal associations between three COVID-19 exposures (critically ill COVID-19, hospitalized COVID-19, and SARS-Cov-2) and cancer risk using MR analysis.    Table 1 presents the summary of the GWAS summary databases of 33 types of cancers.
The clump was performed to rule out linkage disequilibrium (LD) with r 2 < 0.001 and physical distance within 10 000 kb to ensure the independence of IVs. We further harmonized the effect allele for exposure and outcome data sets and removed palindromic SNPs. We 3. MR-Egger could provide a causal effect through the slope coefficient from Egger regression and also detect small study bias. 29 T A B L E 1 Summary of the cancer data sets.

| Sensitivity analysis
A series of sensitivity analyses were used to assess the robustness of the statistically significant causal associations. The Cochran'Q statistic was applied to evaluate the heterogeneity. 30 The MR-Egger intercept analysis evaluated the horizontal pleiotropy, which means IVs affect both exposure and outcome through a pathway not mediated by causal effect. 29 The Leave-one-Out analysis could evaluate whether a single SNP could drive significant results, and it was conducted by removing the SNPs one by one. The MR Pleiotropy Residual Sum and Outlier (MR-PRESSO) could detect the influence of outliers.

| Statistically
All the analyses were performed with the "TwoSampleMR" R package

| Causal effects of critically ill COVID-19 on cancers
We found the genetic liabilities to critically ill COVID-19 were suggestive causally associated with the increased risk of HER2-  (Figure 3 and Additional file 1:

| Causal effects of hospitalized COVID-19 on cancers
The genetic liabilities to hospitalized COVID- 19 (Figure 4 and Additional file 1:

| Causal effects of SARS-CoV-2 infection on cancers
The

| Sensitivity analysis
The results of sensitivity analyses verified the robustness of the above causal associations (Additional file 1:  Figures S1-S3).
The symmetry of funnel plots showed no obvious heterogeneity (Additional file 2: Figures S4-S6). Furthermore, the Leave-one-out analyses showed that no single SNP drove the causal effect (Additional file 2: Figures S7-S9).

| DISCUSSION
The COVID-19 epidemic has created many health problems worldwide. Current observational studies indicated that the characteristic of new cancer cases in the COVID-19 pandemic was less in number LI ET AL.   should be carried out to explore its impact mechanism to achieve early detection and treatment and to improve the prognosis of related cancer patients.

| CONCLUSIONS
The present study suggested that the genetic liability to critically ill COVID-19, hospitalized COVID-19, and SARS-CoV-2 infection may increase the risk for gastrointestinal cancers and HER2-positive breast cancer. In contrast, the genetic liability to SARS-CoV-2 infection may decrease the risk for head and neck cell carcinoma.