Prevalence, determinants, and outcomes of SARS‐COV‐2 infection among cancer patients. A population‐based study in northern Italy

Abstract Background It is well established that cancer patients infected with SARS‐CoV‐2 are at particularly elevated risk of adverse outcomes, but the comparison of SARS‐CoV‐2 infection risk between cancer patients and cancer‐free individuals has been poorly investigated on a population‐basis. Methods A population‐based study was thus conducted in Friuli Venezia Giulia region, northeastern Italy, to estimate prevalence and determinants of SARS‐CoV‐2 infection among cancer patients, as compared to cancer‐free individuals, and to evaluate adverse outcomes of SARS‐CoV‐2 infection. The study included 263,042 individuals tested for SARS‐CoV‐2 in February–December 2020 with cancer history retrieved through the regional cancer registry. Odds ratios (ORs) of SARS‐CoV‐2 positivity, with corresponding 95% confidence intervals (CIs), were calculated using multivariable logistic regression models, adjusted for sex and age. Hazard ratios (HRs) adjusted for sex and age for intensive care unit (ICU) admission and all‐cause death were estimated using Cox models. Results Among 26,394 cancer patients tested for SARS‐CoV‐2, the prevalence of infection was 11.7% versus 16.2% among 236,648 cancer‐free individuals, with a corresponding OR = 0.59 (95% CI: 0.57–0.62). The prevalence was much higher (29% in both groups) during the second pandemic wave (October–December 2020). Among cancer patients, age ≥80 years and cancer diagnosis ≥13 months before SARS‐CoV‐2 testing were the major risk factors of infection. Among 3098 infected cancer patients, the fatality rate was 17.4% versus 15.8% among 23,296 negative ones (HR = 1.63, 95% CI: 1.49–1.78), and versus 5.0% among 38,268 infected cancer‐free individuals (HR = 1.23, 95% CI: 1.12–1.36). No significant differences emerged when considering ICU admission risk. Conclusion Albeit cancer patients reported reduced SARS‐CoV‐2 infection risk, those infected showed higher mortality than uninfected ones and infected cancer‐free population. Study findings claim for continuing to protect cancer patients from SARS‐CoV‐2, without reducing the level of oncologic care.


| INTRODUCTION
Biologic (e.g., immune depression related to anticancer treatments) and non-biologic factors (e.g., most people with cancer needs to interface with health institutions) have been hypothesized to expose cancer patients at higher risk of severe acute respiratory syndrome 2 coronavirus (SARS-CoV-2) infection and, as a consequence, at higher risk of COVID-19 illness than correspondent uninfected cancer patients. [1][2][3][4] The prevalence of SARS-CoV-2 infection among cancer patients showed large variations in estimates, largely dependent on study design, place, and time of investigation. SARS-CoV-2 infection was documented among 0.7% of 59,989 Italian cancer patients undergoing active antitumor treatment according to a retrospective nation-wide study. 5 Among 4789 cancer patients tested for SARS-CoV-2 in northern Italy (i.e., the first area affected by  in Europe), 723 (15.1%) turned out to have acquired SARS-CoV-2 infection, 1 whereas, a study from Reggio Emilia province found that out of 27,386 cancer patients, 1527 were tested for SARS-CoV-2, and 447 (1.6%, 29.5% of those tested) resulted positive. 6 Other investigations have indirectly assessed the prevalence of SARS-CoV-2 infection in cancer patients. For instance, 6%-7% of 5700 COVID-19 patients hospitalized in New York turned out to be cancer patients, a proportion higher than expected according to cancer prevalence in the general population. 7 Several investigations have highlighted that cancer patients infected with SARS-CoV-2 are at particularly elevated risk of adverse outcomes. 1,3,6,[8][9][10][11] A pooled analysis of 52 studies published as of July 2020 projected a high probability of death for infected cancer patients, with a case fatality rate of 25.6%. 8 Findings from an international cohort study identified factors associated with an increased 30day all-cause mortality, including old age, male sex, smoking, number of comorbidities, and active cancer. 9 Conversely, few large investigations evaluated the risk of COVID-19 and adverse outcomes among cancer patients as compared to cancer-free individuals in the general population, reporting heterogeneous results. 3,6 The aims of this population-based study were to estimate prevalence of, and factors associated with, SARS-CoV-2 positivity among people with cancer who underwent molecular swab test in northeastern Italy, making comparisons with the cancer-free population. Furthermore, the study aimed to assess adverse outcomes of infection, that is, admission in intensive care unit (ICU) and death, among SARS-CoV-2-positive cancer patients, both as compared to negative ones and to positive cancerfree individuals.

| METHODS
A retrospective population-based study was carried out on all resident individuals of the Friuli Venezia Giulia (FVG) region (northeastern Italy, 1,210,000 inhabitants) who underwent molecular SARS-CoV-2 testing between February and December 2020.
We took advantage from real-world data derived from the regional health information system warehouse, which covers the whole resident population and includes the totality of health-related databases available--for administrative purposes--in both public and affiliated private facilities. De-identified data can be cross-linked through an encrypted code, which is changed every 6 months in order to guarantee patient anonymity. The study was approved by the Bioethics Committee of the Veneto Regional Authority (protocol No. 245343/2020).
For the aims of this investigation, the following data sources were used: microbiological laboratory records; hospital discharge database; Johns Hopkins Adjusted Clinical Group (ACG) system ® database; cancer registry; diabetes registry; and municipal registries.
The study population consisted of 263,042 individuals who underwent a SARS-CoV-2 nasopharyngeal swab test analyzed by means of real-time reverse transcriptionpolymerase chain reaction (RT-PCR), between 26 February and 31 December 2020. Tested people were classified as positive or negative to SARS-CoV-2 infection based on test results. For individuals with multiple tests, those with all negative results were considered as SARS-CoV-2 negative at the date of the first test, while those with at least one positive result were considered as SARS-CoV-2 positive at the date of the first positive test.
To assess the cancer history, the above data were linked with the regional cancer registry that records incident cancer-free population. Study findings claim for continuing to protect cancer patients from SARS-CoV-2, without reducing the level of oncologic care.

K E Y W O R D S
cancer patients, Italy, mortality, risk factors, SARS-COV-2 cancer cases since 1995 and was updated up to December 2017. The cancer registry data includes all malignant neoplasms (International Classification of Diseases 10th version, ICD-10, codes: C00-C96) and benign, in situ, and uncertain behavior neoplasms of bladder (ICD-10: D30.3, D09.0, D41.4). For the years 2018-2020, information on cancer history was obtained from hospital discharges and the pathological laboratory records. For patients with multiple cancers, the most recent cancer diagnosis was considered, before or up to 30 days from SARS-CoV-2 testing (i.e., concurrent diagnosis). The hospital discharge database (updated on 31 December 2021) was used to assess admissions to ICUs, while information on vital status was obtained from municipal registries (updated to 6 February 2021). The Johns Hopkins ACG system ® database (based on the following data sources: hospital discharge, pharmaceutical prescriptions, access to emergency departments, and prescription charge exemptions), was used to assess the presence of comorbidities up to the year 2019. 12 The presence of chronic conditions was evaluated including the following Medical Conditions (MCs) items: congestive heart failure, diabetes, hypertension, ischemic heart disease, obstructive pulmonary disease; the presence of cardiovascular and respiratory diseases was derived from the corresponding Major Expanded Diagnosis Clusters (MEDCs). The presence of diabetes mellitus was assessed through the regional diabetes registry updated to 2019.
Among all individuals tested for SARS-CoV-2, the odds ratios (ORs) of a positive result, with the corresponding 95% confidence intervals (CIs), were calculated for cancer patients as compared to cancer-free individuals (reference category), overall and in strata of selected characteristics, using multivariable logistic regression models, adjusted for sex and age class (<40, 40-59, 60-69, 70-79, and ≥80 years). 13 Within the group of cancer patients tested for SARS-CoV-2, risk factors for a positive result were also evaluated, by calculating the ORs, and corresponding 95% CI, adjusted for sex and age class.
The risks of admission to ICU and the risk of death for any cause among SARS-CoV-2-positive cancer patients were evaluated both in comparison with SARS-CoV-2negative cancer patients and with SARS-CoV-2-positive cancer-free individuals, overall and in strata of patient characteristics. Hazard ratios (HRs) of the event of interest, and the corresponding 95% CI, were estimated using proportional Cox model adjusted for sex and age class. 14 Time at risk of ICU admission was computed from the date of SARS-CoV-2 test up to the date of first ICU admission, death, or 31 December 2020, whichever came first. Time at risk of death was computed from the date of SARS-CoV-2 test up to the date of death, or 6 February 2021, whichever came first.

| RESULTS
As of December 2020, 263,042 residents in the study area (19.5% of the whole population) underwent at least one molecular swab test for SARS-CoV-2. A history of cancer diagnosed before--or at time of--testing was documented in 26,394 of them (10.0%). Table 1 illustrates the distribution of tested individuals according to cancer history, SARS-CoV-2 test result, and selected characteristics. People tested for SARS-CoV-2 with a history of cancer had a lower prevalence of SARS-CoV-2 infection (11.7%) than cancer-free individuals (16.2%), with a corresponding 41% risk reduction (sex and age adjusted OR = 0.59, 95% CI: 0.57-0.62). Statistically significant reduced risks of testing positive to SARS-CoV-2 for cancer patients were documented in all the examined subgroups. However, during the second wave of the pandemic in Italy (i.e., October-December 2020) the risk reduction was less marked (OR = 0.74, 95% CI: 0.71-0.78).
Among 26,394 cancer patients, women were at a slightly higher risk of testing positive than men (OR = 1.09, 95% CI: 1.01-1.17), patients aged 80 years or older were at significantly increased risk than younger ones (OR = 1.40, 95% CI: 1.25-1.57) ( Table 1). Thirteen percent of cancer patients with two or more chronic conditions were infected with SARS-CoV-2, as compared to 11% among cancer patients without chronic conditions (OR = 1.12, 95% CI: 0.99-1.26). Diabetes mellitus was significantly associated with increased risk (OR = 1.17, 95% CI: 1.06-1.29). The detection of SARS-CoV-2 infection significantly increased with increasing time elapsed between cancer diagnosis and testing, from 5.9% among those tested within 12 months from cancer diagnosis to 15.7% among those tested more than 60 months after cancer (OR = 2.88, 95% CI: 2.58-3.20). The proportion of people with cancer infected with SARS-CoV-2 greatly increased from 3.6% of those tested during the first wave (i.e., February-May 2020) of the pandemic to 0.5% in June-September, and 29.5% during the second wave (i.e., October-December) (OR = 11.22, 95% CI: 9.91-12.72), mirroring the pattern found in cancer-free individuals.
Given such remarkable difference between the two pandemic waves in the FVG region and the large prevalence of infection during the second wave, a subgroup analysis was carried out with focus on the 9331 cancer patients tested in October-December 2020 ( Table 2). SARS-CoV-2 infection was documented in 29.1% of 8597 patients with solid tumors and in 34.5% of 734 patients with hematological malignancies. Wide intra-group variations were noted, from 21.3% (bladder cancer) to 38.7% (women with endometrial cancer) in individuals with solid tumors; and from 29.6% (Hodgkin lymphoma) to 46.5% (multiple myeloma) in those with hematological T A B L E 1 Distribution of 263,042 tested individuals for SARS-CoV-2 according to cancer history a and result of SARS-CoV-2 test (February-December 2020) b malignancies. As compared with 119,326 cancer-free individuals, patients with solid tumors were at lower risk of SARS-CoV-2 positivity (OR = 0.72, 95% CI: 0.68-0.76), whereas, no difference emerged for those with hematological malignancies (OR = 1.00). For most cancer sites, statistically significant risk reductions emerged. Only people with multiple myeloma turned out to be at increased risk of infection (OR = 1.55, 95% CI: 1.01-2.37). Within the group of cancer patients, those with hematological malignancies were at significantly higher risk of testing positive than those with solid tumors (OR = 1.32, 95% CI: 1.12-1.55) ( Table 2 and Figure 1A). In comparison with colorectal cancer (i.e., the most frequent type in both sexes combined), patients with bladder cancer had significantly reduced risk (OR = 0.62, 95% CI: 0.48-0.81), whereas women with endometrial cancer were at a significantly increased risk (OR = 1.50, 95% CI: 1.12-2.02) ( Figure 1B). Among hematological cancer patients, those with multiple myeloma showed a 1.82-fold more elevated risk of infection (95% CI: 1.12-2.95) than non-Hodgkin lymphoma patients ( Figure 1C). As compared to patients with a recent diagnosis (i.e., <13 months), all those tested 13 or more months after cancer diagnosis were at significantly increased risk of infection (OR = 1.86, 95% CI: 1.56-2.21, for 13-24 months, OR = 1.89, 95% CI: 1.62-2.18, for 25-60 months, OR = 2.14, 95% CI: 1.89-2.42, for >60 months) ( Figure 1D).
The clinical impact of SARS-CoV-2 infection among the whole group of 26,394 cancer patients was assessed considering ICU admission and all-cause death among those positive as compared to those negative to SARS-CoV-2 (Table 3) Among individuals positive to SARS-CoV-2, the fatality rate was 17.4% among 3098 with cancer history and 5.0% among 38,268 cancer-free ones, with a corresponding sex and age adjusted 1.2-fold higher risk of death (95% CI: 1.12-1.36) ( Table 4). Conversely, no statistically significant F I G U R E 1 OR a of SARS-CoV-2 positive test during the second pandemic wave (October-December 2020) among 9331 cancer patients b . Notes: a Odds Ratios (OR) and corresponding 95% confidence intervals, estimated using logistic regression models adjusted for age and sex. b According to type of neoplasm (A-C) and time from cancer diagnosis (D). Friuli Venezia Giulia Region, Italy, 1 October-31 December 2020. * Reference Category

| DISCUSSION
The findings from this population-based study showed that, between February and December 2020 in northeastern Italy, 26,394 cancer patients who were tested for SARS-CoV-2 by RT-PCR had about 40% lower risk of resulting positive, as compared to 236,648 cancer-free individuals in the general population who were tested in the same geographical area. On the other side, the study found that cancer patients who were positive to SARS-CoV-2 reported higher risks of death, as compared to both negative cancer patients (HR = 1.6) and, to a lesser extent, to cancer-free individuals who were positive to SARS-CoV-2 (HR = 1.2). Among cancer patients, female sex, age older than 80 years, presence of diabetes mellitus, a longer cancer history, and, above all, the second wave of the pandemic in Italy (i.e., October-December 2020, during which 29.5% of tested cancer patients turned out to be positive) were factors associated with SARS-CoV-2 infection. Findings from a large population-based study conducted in the United States, 3 including more than 2.5 million cancer patients, of whom 1200 (0.05%) with COVID-19 reported a higher risk of COVID-19 among cancer patients, in particular among those with a recent cancer diagnosis (OR = 7.1). The authors suggested that patients with a recent cancer were more exposed to SARS-CoV-2 infection due to high levels of health care contacts. In this study, conversely, we found that a less recent cancer diagnosis was associated with greater risk of testing positive to SARS-CoV-2. Another population-based study conducted in Northern Italy during the first pandemic wave 6 reported an elevated probability of being tested for SARS-CoV-2, especially in those with a cancer diagnosis made in the previous 2 years. Of note, differently from our study, the two above mentioned studies included not only individuals tested for SARS-CoV-2, but they considered all those not tested as uninfected. Such heterogeneous results are probably due to non-biologic factors, in particular to different approaches for access to the local health care system. In line with the results of other investigations, biologic factors, conversely, are likely to explain the increased risk of infection documented among patients with hematological neoplasms during the second pandemic wave, as compared to those with solid tumors. 3,15 Patients with hematological malignancies such as leukemias, myelodysplastic syndromes, myeloproliferative neoplasms, lymphomas, and multiple myeloma have usually longlasting immunodeficiency due to the malignancy itself, anticancer treatments, or as a consequence of procedures such as hematopoietic stem-cell transplantation. This makes them particularly susceptible to bacterial and viral infections, such as SARS-CoV-2 pneumonia. Consistently with the literature, 16 our data indicated that patients with hematological neoplasms are a population at high-risk of poor COVID-19 outcomes, even in comparison with patients with solid tumors.
Our findings regarding adverse clinical outcomes, particularly death, are in agreement with those of Wang et al., 3 showing higher mortality among patients with both cancer and COVID-19 as compared to both patients with cancer without COVID-19 and patients with COVID-19 without cancer. Similar findings emerged also in other investigations. 1,6,8 The particularly high mortality within 30 days from a positive test was also in agreement with other studies. 9,11 Among cancer patients, when comparing those positive versus those negative to SARS-CoV-2, we found that higher mortality risks emerged for males, older ones, those with chronic diseases and diabetes, in line with the findings of other studies conducted among cancer patients 2,9,11,[17][18][19][20] but also in line with risk factors associated with COVID-19 mortality in the general population. 19,21 Conversely, when comparing SARS-CoV-2-positive individuals with cancer history to those without cancer, our findings highlighted that cancer patients were at higher risk of death when they were younger, without cardiovascular diseases or chronic conditions, and with a more recent cancer diagnosis. No differences in mortality emerged according to sex, presence or absence of respiratory diseases or diabetes, or the period of testing. These results are in line with those reported by Mangone et al. 6 Reasons for being tested for SARS-CoV-2 need to be considered with regard to study limitations. These included chiefly the presence of COVID-19 illness symptoms, the close contacts with SARS-CoV-2-positive individuals, and screening for SARS-CoV-2 infection before hospital admission or during hospital stay. It is possible that tested cancer patients included a larger proportion of people who were under active cancer treatment for whom molecular swab test was mandatory (i.e., those admitted to hospitals for chemotherapy or surgical procedures). Due to absence or the very low number of observed adverse outcomes in some sub-groups of patients (e.g., selected cancer types), our study had not sufficient statistical power to evaluate the risk of event--in particular ICU admission--in all subgroups.
A major strength of this study is the whole coverage of the resident population in the study area, thanks to the availability of a centralized health system data warehouse. This allowed to include all the RT-PCR test for SARS-CoV-2 performed in the Friuli Venezia Giulia region during the study period in public facilities. Molecular swab tests performed in private health facilities and rapid antigen tests were not available, but they were unusual in 2020 in the study area. Another study strength was the use of data from a population-based cancer registry with a long history and high quality standards in terms of completeness and accuracy of collected data (accredited to the International Association of Research or Cancer and the Italian Association of Cancer Registries). 22 We are aware that data derived from health system databases can suffer of a reduced quality and completeness of information due to their administrative nature (i.e., they are not collected for clinical or research purposes). For instance, the information on the presence and the number of chronic conditions were derived from the ACG system ® database and not from patient's anamnesis, thus we cannot exclude underestimation and misreporting of comorbidities. In addition, some information is not available in the regional health system data warehouse (e.g., smoking habits, body mass index, and type and phase of cancer treatment).
In conclusion, this investigation showed that, albeit the lower risk of testing positive to SARS-CoV-2 of cancer patients in Italy, those who were infected suffered of a higher risk of death.
Particular attention should be taken in order to further protect oncologic patients against COVID-19, while continuing to assure them an adequate access to medical care. In order to reduce the risk of SARS-CoV-2 infection among cancer patients and to guarantee the best continuity of care, especially among the elderly, practical recommendations have been made by Italian oncologists. 23 These measures aim to: (1) favor social distancing and (2) reduce immunodeficiency and iatrogenic treatmentrelated events that increase the risk of infections. Among the others, the recommendations included: allowing presence of a single caregiver for a limited time; promoting telephone/telematics triage of disease-free cancer patients and patients on oral agents treatment; choosing oral targeted therapy over intravenous agents when multiple treatment options are available; employing regimens with a longer interval or prolonging cycle length for diseaseactive patients receiving intravenous agents; and avoiding preoperative chemotherapy with high risk of neutropenia for resectable cancer patients.