The effect of anticancer treatment on cancer patients with COVID‐19: A systematic review and meta‐analysis

Abstract Background The relationship between cancer and COVID‐19 has been revealed during the pandemic. Some anticancer treatments have been reported to have negative influences on COVID‐19‐infected patients while other studies did not support this hypothesis. Methods A literature search was conducted in WOS, PubMed, Embase, Cochrane Library, CNKI and VIP between Dec 1, 2019 and Sept 23, 2020 for studies on anticancer treatments in patients with COVID‐19. Cohort studies involving over 20 patients with cancer were included. The characteristics of the patients and studies, treatment types, mortality, and other additional outcomes were extracted and pooled for synthesis. RRs and forest plots were adopted to present the results. The literature quality and publication bias were assessed using NOS and Egger's test, respectively. Results We analyzed the data from 29 studies, with 5121 cancer patients with COVID‐19 meeting the inclusion criteria. There were no significant differences in mortality between patients receiving anticancer treatment and those not (RR 1.17, 95%CI: 0.96–1.43, I2=66%, p = 0.12). Importantly, in patients with hematological malignancies, chemotherapy could markedly increase the mortality (RR 2.68, 95% CI: 1.90–3.78, I2=0%, p < 0.00001). In patients with solid tumors, no significant differences in mortality were observed (RR 1.16, 95% CI: 0.57–2.36, I2=72%, p = 0.67). In addition, our analysis revealed that anticancer therapies had no effects on the ICU admission rate (RR 0.87, 95% CI: 0.70–1.09, I2=25%, p = 0.23), the severe rate (RR 1.04, 95% CI: 0.95–1.13, I2=31%, p = 0.42), or respiratory support rate (RR 0.92, 95% CI: 0.70–1.21, I2=32%, p = 0.55) in COVID‐19‐infected patients with cancer. Notably, patients receiving surgery had a higher rate of respiratory support than those without any antitumor treatment (RR 1.87, 95%CI: 1.02–3.46, I2=0%, p = 0.04). Conclusions No significant difference was seen in any anticancer treatments in the solid tumor subgroup. Chemotherapy, however, will lead to higher mortality in patients with hematological malignancies. Multicenter, prospective studies are needed to re‐evaluate the results.


| Study protocol
We planned, conducted, and reported the systematic review and meta-analysis in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) 2015 Statement (Supplement 1). 12 The whole protocol has been registered in the PROSPERO database (CRD42020200736).

| Literature search
Given that many early studies were conducted by Chinese researchers, both English and Chinese databases were searched to minimize language bias. The searched English databases included Web of Science (WOS), PubMed, Embase, and Cochrane Library, while the Chinese databases included the China National Knowledge Infrastructure (CNKI) and the China Science and Technology Journal Database (VIP). One researcher (HQ L) with meta-analysis experience drafted the search strategy, which was revised and approved by other researchers. The following medical subject headings (MeSH) and non-MeSH keywords were arranged in the search sentence: (COVID-19 OR SARS-CoV-2 OR 2019-nCoV OR coronavirus) AND (tumor OR carcinoma OR cancer OR hematolog* OR haematolog* OR leukemia OR lymphoma OR myeloma) ( Table 1). The published dates of studies were limited to Dec-01, 2020 to Sept-23, 2020, with no restriction on language. The lists of references were screened to identify any missed studies. The literature from different sources was then imported into Endnote (version X9.0) for duplicate exclusion.

| Study selection and definition
In this systematic review, any research articles meeting the following criteria were included for the further data extraction and synthesis: (a) studies reporting the effects of any antitumor treatments on mortality, ICU admission rate, rate of respiratory support or severe/critical rate in patients with cancer diagnosed with COVID-19; (b) patients ≥18 years old, and (c) the relative risk (RR) can be extracted or relevant statistics are provided for calculation.
Studies meeting the following criteria were excluded: (a) review, news, editorial, comment, guideline, clinical experience, basic research, study protocol or case report; (b) cancer patients <20 or cannot be separated from non-cancer patients; (c) patients were diagnosed with other viral pneumonia, such as SARS or MERS and (d) data derived from the same group of patients.
Two independent reviewers (HQ L and D Y) carried out the literature screening with blindness to each other. The titles and abstracts were screened in the first two rounds for efficiency. Then full articles were obtained for subsequent selection according to the criteria. Disagreements were resolved via consultation with a senior reviewer (C C).
The diagnosis of COVID-19 should be based on RT-PCR or antibody tests. Due to the changing standard for diagnosis, the shortage of testing kits in some regions, and the unsatisfactory accuracy of laboratory tests, 13,14 a CT finding or a consensus based on symptoms by ≥2 skillful physicians was also acceptable. No restriction was cast on cancer types, but cancer needed to be concurrent with COVID-19, and a cancer history was obviously unacceptable. Any type of antitumor treatment should be administered within 3 months before the diagnosis of COVID-19. The end-points should be measured in hospitals or medical institutions. Respiratory support was defined as mechanical ventilation, facial mask or any other mechanical technique improving the respiratory function. The definition of the severe/critical rate should conform with the Diagnosis and Treatment Protocol for Novel Coronavirus Pneumonia released by the National Health Commission, 15 with no limitation on its version.

| Data extraction and quality assessment
Two authors (HQ L and D Y) extracted the data from the included studies independently and then cross-checked their results. Disagreements were resolved via consensus or consultation with a senior reviewer (C C). The following data were collected in a worksheet: first author, published date, country, study design, number of patients, number of females, median age, comorbidities, detection of COVID-19, cancer type, interpretation type, and outcome. The relative risks (RRs) were obtained from the papers or calculated based on original statistics. The Newcastle-Ottawa Quality Assessment Scale for Cohort Studies was adopted in the quality assessment 16 (Supplement 2). Eight questions in the scale were arranged into three groups: patient selection, comparability and outcome reliability. Two reviewers (XY C and ZH S) assessed the risk of bias independently with blindness to each other. Disagreements were settled by a third reviewer (YT Z).

| Data synthesis and statistical analysis
The data synthesis was performed on RevMan (version 5.3) and the publication bias was calculated with Stata (version 15.1). Relative risks and 95% confidence intervals (CIs) were calculated to compare the mortality rate and other additional outcomes between patients receiving antitumor treatments or not. A p-value <0.05 was deemed statistically significant. The inconsistency index (I 2 statistic) and Cochran's Q test were adopted in the assessment of heterogeneity. The 50% I 2 was defined as a cut-off for low and high heterogeneity. With low heterogeneity, a fixed-effects model was used to estimate the average effect and its precision. If the heterogeneity was high, a random model was adopted. Subgroup analyses were then performed on specific antitumor treatments and different cancer types (solid tumor or hematological malignancy). The minimum number of articles for data synthesis was two in each group. The funnel-plot asymmetry designed by Egger et al. 17 was employed to estimate the publication bias.

| Search results
A total of 5015 records were identified in our initial search. Of these, 1009 papers were duplicates and thus excluded. After review of titles and abstracts, 3744 papers that did not fulfill our criteria for full-text review were removed, leaving 262 papers for further evaluation. Another 233 papers were excluded because they were case reports/series, basic studies, editorials, comments, guidelines, articles that were not relevant to cancer/COVID-19, articles with no control group, articles with fewer than 20 patients, or overlapping data sources. Eventually, 29 studies were included in our systematic review and meta-analysis ( Figure 1).

| Study characteristics
A total of 5121 patients with cancer in 29 studies were included in this meta-analysis. [6][7][8][9][10][11] The characteristics of the studies included in this meta-analysis are presented in Table 2. Of the remaining 29 studies, eight were conducted in China, six in the United States, four in the United Kingdom, three in Spain, two in France, two in Italy, and four was performed in multiple countries. Patients with COVID-19 were mainly confirmed by real-time reverse transcriptase-polymerase chain reaction (RT-PCR). The sample sizes of the included studies ranged between 25 and 928, and the NOS scores varied from 5 to 7 (Supplement 3).

| DISCUSSION
This systematic review and meta-analysis, in which a total of 5121 patients with cancer with COVID-19 from 29 studies were included, is the largest study discussing the question to our knowledge. Our work did not suggest that the antitumor treatments would lead to poorer prognosis in patients with solid tumors diagnosed with COVID-19. In contrast, patients with hematological malignancies are at higher risk of death if they receive chemotherapy in three months before the COVID-19 diagnosis.
Since the first report by Liang et al, 5 the treatment of cancer patients with COVID-19 has been a hot topic. Cytotoxic chemotherapy, which can decrease the leukocyte count and lead to immunosuppressive status, has been reported to result in a high infection rate and poor prognosis. 38,41 The SARS-CoV-2 is more likely to trigger cytokine storm (CS) than other pulmonary infections. A CS will subsequently increase the incidence of the acute respiratory distress syndrome (ARDS), which has been observed in approximately 15% of cases. 42 According to the study of Wan et al, 43 IL-6 was elevated significantly in the serum of severe cases, while CD4 + T cells, CD8 + T cells and natural killer cells were lower than those in mild cases.

F I G U R E 4
Forest plot for the association between antitumor treatments and the mortality rate in solid tumor patients with COVID-19 using random-effects model This may be explained by the reciprocal circle between the CS and the immunosuppressive status caused by chemotherapy and the cancer itself. In addition, chemotherapy for hematological malignancies will lead to a much higher rate of neutropenia and lymphocytopenia, which is considered a risk factor for mortality in patients with COVID-19 in many studies. 44 The elevated RR of the severe/critical rate in chemotherapy proves to support the theory. The adverse impact of cytotoxic chemotherapy on prognosis was also revealed in other viral infections. 45,46 Moreover, cytotoxic agents vary in their mechanisms and some agents were found to have anti-CS effects, 47 which may account for the high heterogeneity of chemotherapy. Targeted agents, which are highly selective to on co-molecular targets, are generally thought to cause fewer side effects. 48 The results of targeted therapy are similar to those of chemotherapy.
Patients receiving recent surgeries were reported to have a higher risk of viral infection and severe events, 8 partially due to their frequent visits to hospitals and postoperative negative nitrogen balance. However, our results did not support this hypothesis. The higher rate of respiratory support in surgery patients may be explained by the routine use of postsurgical life support. In addition, the patients included in our meta-analysis had distinct admission dates, which ranged from January to late May. Notably, their clinical management strategies have changed during this F I G U R E 5 Forest plot for the association between antitumor treatments and the mortality rate in hematological malignancies patients with COVID-19 using random-effects model period. 49,50 Additionally, many elective operations were postponed or canceled while the remaining operations received special attention and care.
Radiotherapy has been confirmed to decrease lymphocytes and may lead to lymphopenia in some cases. 51 Interestingly, our results showed that patients receiving radiotherapy tended to have a better prognosis than those not receiving radiotherapy, but a significant difference was not reached. Several scholars have supported the hypothesis that low-dose radiation may mitigate the CS via pre-consumption of immune reserves and a reduction in virus loading. 52,53 Hence, further investigations are warranted.
Immunotherapy represents another effective anticancer therapy with remarkable clinical benefits.There exist three major approaches to T cell-based cancer immunotherapy, which are immune checkpoint blockade (ICI), adoptive cell transfer therapy, and active vaccination. 54 Our results showed that immunotherapy had the highest risk among all anticancer treatments. The potential mechanism could be the activation of T cells by ICIs and a subsequent uncontrolled aberrant inflammatory response. 55 Some researchers have now been working on a risk assessment scoring system to decide which patients with cancer could receive immunotherapy. 56 To conclude, the prescription of immunotherapy should be used with extraordinary caution.
Although this meta-analysis was carried out strictly conforming with the PRISMA, there were some limitations. The reliability of the results was to some extent weakened due to the lack of sufficient data. Some studies involved were single-center and small-sample studies, indicating the possibility of admission bias and sampling error. The ICU admission rate and the rate of respiratory support should be interpreted with caution, as they were highly related to the physicians' experience. Due to the small sample size, chemotherapy had to be handled as a whole and subgroup analysis based on their individual pharmacological mechanism was difficult to perform. Furthermore, the effects of age, cancer type, and comorbidities were hard to evaluate. To conclude, the results of this systematic review should be interpreted with caution. However, the studies included were still the core of the evidence to date. A more persuasive study may re-evaluate our conclusions.
This study was designed to provide physicians with more information about the safety of anticancer treatments in the COVID-19 era. Bundles of studies have reported that the delay or cancelation of planned treatments during the pandemic might have a negative influence on patient prognosis. [57][58][59] Although a 2-month delay of treatment for stage I/ II cancers was reported to be acceptable, 60 the effect of delay in high-stage cancers remains unclear, especially in patients older than 75. 61 The clinical strategy for cancer management should be made based on the local medical capacity, the neighboring epidemic condition and the specific patient's condition. Telemedicine has been advocated by many experts in the follow-up of non-urgent cancer patients. 62,63 E-visits, remote care management, and remote patient monitoring aids can be implemented using the social networks. For those at high risk of complications if their treatments are postponed, a systematic evaluation of the patient's conditions including RT-PCR on nasopharyngeal swabs and thoracic CT is necessary. 64 For those with oncologic emergencies, large lung masses, head and neck cancers and chemotherapy, or radiotherapy for high-stage cancers, 57 the active anticancer treatment should be received without any delay.
In conclusion, our results suggest that the chemotherapy for patients with hematological malignancies should be administrated with great caution. There was no stable evidence to confirm the adverse effect of any antitumor therapies in patients with solid tumors with COVID-19. Some adverse tendencies have appeared in chemotherapy, surgery and immunotherapy, but none have reached a significant difference. Multicenter and prospective studies are needed to re-evaluate our conclusions.