Safety of systemic anti-cancer treatment in oncology patients with non-severe COVID-19: a prospective cohort study

Background The viral pandemic coronavirus disease 2019 (COVID-19) has disrupted cancer patient management around the world. Most reported data relate to incidence, risk factors, and outcome of severe COVID-19. The safety of systemic anti-cancer therapy in oncology patients with non-severe COVID-19 is unknown. Methods ONCOSARS-1 was a prospective, single-center, academic observational study. Adult patients with solid tumors treated in the oncology day unit with systemic anti-cancer therapy during the initial phase of the COVID-19 pandemic in Belgium were included. All patients (n=363) underwent severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) serological testing after the peak of the pandemic in Belgium. Additionally, 141 of these patients also had a SARS-CoV-2 RT-PCR test during the pandemic. The main objective was to determine the safety of systemic cancer treatment, measured by the rate of adverse events according to the Common Terminology Criteria for Adverse Events, in SARS-CoV-2-positive patients compared with SARS-CoV-2-negative patients. Results Twenty-two (6%) of the 363 eligible patients were positive for SARS-CoV-2 by RT-PCR and/or serology. Of these, three required transient oxygen supplementation, but none required admission to the intensive care unit. Hematotoxicity was the only adverse event more frequently observed in SARS-CoV-2 positive patients than in SARS-CoV-2-negative patients: 73% vs 35% (P<0.001). This association remained signicant (odds ratio (OR) 4.1, P=0.009) even after adjusting for performance status and type of systemic treatment. Hematological adverse events led to more treatment delays for the SARS-CoV-2-positive group: 55% vs 20% (P<0.001). Median duration of treatment interruption was similar between the two groups: 14 and 11 days, respectively. Febrile neutropenia, infections unrelated to COVID-19, and bleeding events occurred at a low rate in the SARS-CoV-2-positive patients. safe in oncology patients during the COVID-19 pandemic. SARS-CoV-2-positive


Introduction
The viral pandemic coronavirus disease 2019 , caused by the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), can present broadly, from asymptomatic infection through symptoms of mild general upper respiratory infection, life-threatening acute respiratory distress syndrome, and death [1].
Acute infection is diagnosed by reverse transcription polymerase chain reaction (RT-PCR) allowing the detection of viral RNA from a nasopharyngeal swab. Unfortunately, logistic, nancial, and sometimes legal issues preclude the vast majority of caregivers around the world, including us, to test patients before each hospital admission, treatment, or diagnostic procedure [2]. Early in the pandemic, speci c infection prevention measures and recommendations were implemented in hospitals. These aimed to identify infected patients before they entered the oncology department, protect patients and staff, and maintain treatment quality [3,4].
The majority of scienti c data reported so far regarding COVID-19 infection in cancer patients relate to the severe form of the disease and focus mainly on incidence, risk factors and outcomes [5][6][7]. Several meta-analyses suggest that cancer patients are at higher risk of developing the disease and having a severe form, but also present higher mortality rates [8][9][10].
However, around 80% of COVID-19-positive patients have only mild symptoms of the disease, and an unde ned number are asymptomatic transmitters [1]. Although of utmost importance, the safety of systemic anti-cancer treatments in these patients with a non-severe form of COVID-19 is unknown. Indeed, in the presence of mild symptoms, COVID-19 infection can remain undetected due to the absence or unavailability of repeatable rapid screening methods. Access to SARS-CoV-2 RT-PCR testing was limited in Belgium during the initial pandemic due to logistics; its availability focused mainly on patients with severe COVID-19 that required hospitalization. The Belgian authorities authorized serological testing to detect post-infection antibodies in the population on May 20, 2020. As we did not screen asymptomatic patients or those with only mild symptoms during the pandemic, we suspected that some of our patients with non-severe SARS-CoV-2 infection have received systemic anti-cancer treatment. We therefore undertook a prospective study between June 12 and July 13, 2020 and offered serological testing to each cancer patient undergoing systemic treatment in our cancer day unit. Our main objective was to report the complication rates of systemic treatment in cancer patients with non-severe COVID-19 infection, identi ed by SARS-CoV-2 RT-PCR or serology, in comparison to SARS-CoV-2-negative patients.
We also aimed to identify the discriminant symptoms and clinical/radiological factors associated with SARS-CoV-2 positivity, and to estimate the seroprevalence of antibodies against SARS-CoV-2 in this population.

Material And Methods
Study design, inclusion criteria, study objectives and endpoints ONCOSARS-1 was a prospective, single center, academic observational study performed at Institut Roi Albert II, Cliniques universitaires Saint-Luc (CUSL), a tertiary cancer center and the largest in Brussels, Belgium. The inclusion criteria were: (i) patients older than 18 years, (ii) with a diagnosis of active solid cancer, (iii) having received systemic anti-cancer therapy between February 15 and May 31, 2020, (iv) admitted to the oncology day unit between June 12 and July 13, 2020, (v) who agreed to have a SARS-CoV-2 serological test and (vi) who signed an informed consent. There were no exclusion criteria. Hematological cancer patients are treated in a separate outpatient unit and did not take part in this study.
In this selected group of patients, our main objective was to investigate the safety of systemic anti-cancer treatment administered to SARS-CoV-2-positive patients in a day unit setting. By doing so, we also aimed to identify the discriminant symptoms and clinical or radiological factors associated with SARS-CoV-2 positivity, and to estimate the seroprevalence of SARS-CoV-2 in this population. The primary endpoint was the rate of adverse events in SARS-CoV-2-positive patients compared to SARS-CoV-2-negative patients according to the Common Terminology Criteria for Adverse Events (CTCAE) version 5.0.
SARS-CoV-2-positive patients were de ned as patients in whom SARS-CoV-2 RNA was detected by RT-PCR in a nasopharyngeal swab between February 15 and July 13, 2020, or as patients with a positive SARS-CoV-2 serological test between June 12 and July 13, 2020. The study was approved by our Ethics Committee (2020/18MAI/278) on June 8, 2020, and all patients signed an informed consent.
In parallel to this study, all cancer healthcare workers had the opportunity to have a SARS-CoV-2 serological test at the same timepoint following informed consent.

SARS-CoV-2 serological test and RT-PCR
Qualitative detection of SARS-CoV-2 total antibodies in human serum was performed on the Cobas e602 using Elecsys anti-SARS-CoV-2 electrochemiluminescent immunoassay (ECLIA) (Roche Diagnostics, Mannheim, Germany). This assay uses a recombinant nucleocapsid antigen for the detection of anti-SARS-CoV-2 antibodies. According to the manufacturer's insert, a cutoff index of ≥ 1.0 indicates a positive result [11].

Data extraction
Data were systematically collected for all patients admitted to our oncology unit. At the beginning of the pandemic, the pre-de ned data template was adapted to highlight COVID-19 symptoms (Supplementary appendix 1). Clinical data regarding patient demographics, comorbidities, cancer type and stage, oncological treatment, adverse events (graded according to CTCAE v5.0), COVID-19 related symptoms (fever, cough, anosmia, dyspnea and rhinitis), thoracic imaging and SARS-CoV-2 testing (RT-PCR and serology) were collected and managed using REDCap (Research Electronic Data Capture) tools hosted at CUSL [12,13]. Data were collected during the initial pandemic outbreak in Belgium (February 15 -May 31, 2020).

Statistical analysis
The distribution of clinical characteristics and outcomes between SARS-CoV-2-positive and -negative patients (as detected through RT-PCR or serology) was compared using the Fisher's exact or Mann-Whitney test, as appropriate. Bonferroni correction was applied for multi-testing. The frequency of treatment-related adverse events was compared using Fisher's exact test. Univariate and multivariate logistic regression modelling was used to identify risk factors associated with (i) adverse events signi cantly more frequent in SARS-CoV-2-positive patients and (ii) SARS-CoV-2 seroconversion. The threshold to include candidate variables in the multivariate model was set at 10%, and a backward stepwise selection was used to obtain the optimal model. In all analyses, p-values were 2-tailed, and Pvalues < 0.05 were considered statistically signi cant. Data analyses were performed using SAS software (version 9.4, SAS Institute Inc., NC, USA).  in Belgium, at CUSL, and in the health care workers of the CUSL ambulatory cancer care unit. Table 1 summarizes the COVID-19 pandemic epidemiology in Belgium and the main actions our cancer center implemented according to national and institutional recommendations. Belgian o cial COVID-19 data are continuously updated on https://covid-19.sciensano.be. Between February 15 and July 13, 2020, only two (7%) out of the 28 physicians who attended the outpatient cancer center daily developed a symptomatic COVID-19 infection documented by SARS-CoV-2 RT-PCR. None of the other physicians nor the unit's 19 oncological nurses developed a symptomatic infection. All healthcare workers were tested for SARS-CoV-2 antibodies during the same period as the ONCOSARS-1 study. Only the two physicians who were positive for SARS-CoV-2 by RT-PCR developed antibodies against SARS-CoV-2.

Patient characteristics and incidence of SARS-CoV-2 detection
Of the 415 patients admitted to our oncology day unit between June 12 and July 13, 2020, 379 (91%) signed the informed consent and 363 (87%) met our inclusion criteria ( Figure 1). Table 2 outlines the main patient characteristics. Per the inclusion criteria, all eligible patients (n=363) had SARS-CoV-2 serology performed. Among these, 141 (38.8%) underwent RT-PCR during the COVID-19 pandemic: 22 (16%) had symptoms suggestive of a SARS-CoV-2 infection that triggered the test and 119 (84%) were tested as part of a systematic screening plan. This plan, introduced by CUSL, from April 1, 2020, mandated that all patients planned for a medical procedure or overnight stay must undergo systematic screening with SARS-CoV-2 RT-PCR (Table 1). However, this systematic screening was not implemented in the outpatient oncology day unit, mainly due to logistical reasons.
Three hundred and forty-nine patients did not undergo any RT-PCR test or had a negative RT-PCR test (Table S2). Eight (2%) of these patients developed antibodies against SARS-CoV-2. Five of these eight had undergone an RT-PCR test, but none experienced COVID-19 symptoms.
SARS-CoV-2-positive patients diagnosed either by RT-PCR or serological test (n=22) did not statistically differ from SARS-CoV-2-negative patients (n=341) with regards to potential risk factors for severe COVID-19 disease (age, comorbidities, smoking, lung cancer, advanced disease, number of lines of systemic therapies for advanced disease, Eastern Cooperative Oncology Group (ECOG) performance status) ( Table  2) [14]. However, the SARS-CoV-2-negative patients had received overall more immunotherapy, generating an imbalance between the treatment groups (P=0.029). In terms of cancer types, head and neck cancers and skin cancers were only observed in the SARS-CoV-2 negative group. SARS-CoV-2-positive patients received more thoracic radiotherapy over the previous six months (P=0.007). These same trends were also seen numerically in the COVID-19 subgroups detected through RT-PCR alone (Table S1) or serology alone (Table S2).

COVID-19 symptoms and thoracic imaging
Among the 363 patients, 150 (41%) developed symptoms suggestive of COVID-19 infection and 213 (59%) remained asymptomatic. Although present in only 5 (23%) SARS-CoV-2-positive patients, anosmia appeared to be the most discriminant symptom with no anosmia reported in the SARS-CoV-2-negative patients (P < 0.001). Dyspnea and rhinitis were not discriminant, being frequently reported in treated cancer patients. SARS-CoV-2-positive patients presented at least one of the more speci c COVID-19 symptoms (fever or cough or anosmia) more frequently during the follow-up period (46% vs 13%, P < 0.001) ( Table 2). Similar ndings were observed for the population described in the COVID-19 subgroups detected through RT-PCR alone (Table S1) or serology alone (Table S2).
Adverse events reported during systemic oncological treatments SARS-CoV-2-positive patients identi ed either by RT-PCR or serology (n=22) presented more hematological adverse events compared to SARS-CoV-2-negative patients (n=341) (73% vs 35%, P < 0.001). Adverse events for neutropenia and lymphopenia were of all grades (Table 3). Only grade 1-2 thrombopenia was observed. There were no signi cant differences between the SARS-CoV-2-positive and -negative patients with regards to other observed adverse events. Bleeding, infections unrelated to COVID-19, and febrile neutropenia occurred at a low rate in the SARS-CoV-2 positive patients (Table 3). Both subgroups of COVID-19 detected only through RT-PCR (Table S3) or only through serology (Table S4) presented similar results. In both univariate and multivariate analyses, SARS-CoV-2 positivity, a lower performance status (ECOG 2-3), and treatment with chemotherapy were signi cantly associated with hematological toxicity. Advanced disease, higher age, the presence of a comorbidity, and symptoms of COVID-19, were not (Table 4).
SARS-CoV-2-positive patients experienced more frequent treatment delays than SARS-CoV-2-negative patients: 55% vs 20%, respectively (P < 0.001). However, the length of delay did not differ between the two groups: median 14 vs 11 days, respectively (P= 0.504). The relationship between treatment delays and hematological adverse events was signi cant (P < 0.001, Fisher exact test).
Two SARS-CoV-2-positive patients, detected by RT-PCR and later con rmed by serology, prematurely ended their ongoing systemic treatment during the study period; one developed complicated sigmoid diverticulitis, while the second developed grade 2 peripheral neuropathy. Three SARS-CoV-2-negative patients also stopped treatment prematurely; two due to adverse events and one to disease progression. No patients stopped treatment due to the pandemic.

Factors associated with seroconversion
To further assess the factors associated with positive SARS-CoV-2 serology, univariate and multivariate analyses were performed (Table 5). SARS-CoV-2 positivity by RT-PCR was the strongest factor associated with seropositivity. The main COVID-related symptoms (fever, cough or anosmia), or a thoracic CT scan with lung in ltrates suggestive of COVID-19 infection, were also signi cantly associated with seroconversion in both the univariate and multivariate analyses. A higher age, the presence of a comorbidity, lung cancer, recent morbid surgery, or advanced disease were not associated with seroconversion. Recent thoracic radiotherapy was signi cantly associated with positive serology in the univariate analysis but not in the multivariate analysis.

COVID-19 outcomes
Five of the 22 (23%) SARS-CoV-2-positive patients were hospitalized due to COVID-19 infection. Of these, three required transient nasal oxygen therapy. All developed antibodies against SARS-CoV-2. Four additional patients were hospitalized with presumed COVID-19, but their status was validated neither by RT-PCR nor by thoracic imaging. These four patients eventually had negative serology.

Discussion
The rapid spread and non-negligible risk of a fatal outcome due to COVID-19 has triggered scienti c societies and healthcare systems to de ne strategies to mitigate the risk of infection in vulnerable patients [3,4]. Here we present the results of a prospective cohort study of ambulatory cancer patients treated with systemic therapy during the main pandemic period in Belgium.
In our study, only 6% of patients (a nding similar to what is seen at the time of writing this article in the general Belgian population, see table 1), and 7% of healthcare workers had laboratory proven contact with the virus (positive RT-PCR or detectable anti-SARS-CoV-2 antibodies). We hypothesize that the rapid implementation of preventative measures (systematic screening of patients with fever or acute airway tract symptoms before entering the oncology day unit, prohibiting visitors unless medically advised, the strong recommendation to wear a face mask, and social distancing) was a key factor, at least in part, in our low rate of SARS-CoV-2 infection [16].
Our SARS-CoV-2-positive patients presented more frequently with fever, cough or anosmia. Although clearly associated with COVID-19 in the general population, [14] dyspnea, a frequent symptom of cancer patients, as well as fatigue and diarrhea, frequently reported as treatment-related adverse events, appeared less discriminant in our cancer patients (Tables 2 and 4). This illustrates just one aspect of the diagnostic challenge during this pandemic. We observed that ve out of eight (63%) patients who were not diagnosed with COVID-19 during the pandemic, but for which a seroconversion was detected afterwards, had presented at least one COVID-19 related symptom. This highlights the need to implement large COVID-19 screening programs in cancer patients to avoid spread of the virus, even among patients with no or very mild COVID-19 symptoms. We found that SARS-CoV-2-positive patients had more characteristic lung in ltrates on CT-scan. This nding was also observed in SARS-CoV-2-positive patients who were not detected during the three rst months of the pandemic but who were subsequently diagnosed through serological testing. This suggests that thoracic imaging could also be a useful screening tool in the event of a second wave.
Delivering optimal systemic anti-cancer treatment while protecting our patients and medical staff were key objectives during this pandemic. Except the implementation of speci c protective measures depicted in Table 1, we did not modify our standard day unit treatment protocols. The safety data we report here are therefore reassuring. We only found a higher rate of hematological adverse events in the SARS-CoV-2positive patients including all grades of neutropenia and lymphopenia, with the latter being the most commonly found biological abnormality in this disease [14]. This led to a higher rate of treatment delays. Importantly, we did not encounter any increase in the other assessed adverse events in SARS-CoV-2positive patients, whereas bleeding, infections unrelated to COVID-19, and febrile neutropenia rates were low. A multivariate analysis con rmed that hematological toxicity was signi cantly associated with the presence of SARS-CoV-2 antibodies after adjusting for patient's performance status and type of systemic treatment. A poorer performance status and the administration of chemotherapy were also signi cantly associated with hematological toxicity. Recently published data is consistent with our ndings: in a cohort study of 309 COVID-19-positive cancer patients, the administration of chemotherapy was not associated with a severe form of the disease [17]. Likewise, in a cohort study of 1,016 patients, SARS-CoV-2 infection rates remained as low as in the general population after the implementation of institutional safety measures [18]. Previous work hypothesized that cancer patients are less likely to develop or maintain antibodies against SARS-CoV-2 [19]. This study adds data to this assumption, as only nine out of 14 (64%) patients with a positive SARS-CoV-2 RT-PCR test had detectable antibody levels at study inclusion. In this population, the presence of symptoms or lung in ltrates on CT-scan suggestive of COVID-19 was signi cantly associated with the development of antibodies. However, we could not nd de nitive risk factors for the development of antibodies against SARS-CoV-2. Administration of radiotherapy to the thoracic area within the previous six months was signi cantly associated with seroconversion in the univariate but not multivariate analysis, warranting exploration in larger cohorts.
Our study has several limitations. First, we prospectively recruited only patients who were still being actively treated with anti-cancer therapy after the peak of the pandemic. Therefore, we cannot exclude that we missed some SARS-CoV-2-negative and non-severe SARS-CoV-2-positive patients treated during the pandemic who had already stopped treatment before the study began. Second, we cannot guarantee that some medical decisions, such as complementary tests and/or treatment options, were not taken with the pandemic in mind. Third, at the beginning of the pandemic, the decision whether to perform SARS-CoV-2 RT-PCR testing was taken in the context of testing kit shortages. Likewise, it was logistically not feasible to compensate for the lack of tests by offering thoracic CT scans. Finally, the rate of false negative results from the nasopharyngeal RT-PCR test is reported to be in the order of 30-40% [20]. This reduces the accuracy of any data ever reported for COVID-19. Likewise, several tests exist to detect antibodies against SARS-CoV-2, with highly variable diagnostic accuracies [21]. However, the electrochemiluminescent immunoassay we used has a high sensitivity (91%) and speci city (100%) at the cut-off pre-speci ed by the manufacturer [22].
In conclusion, this prospective cohort study of patients with solid cancers actively treated during the COVID-19 pandemic demonstrates that when strong preventive measures are taken to protect patients and healthcare workers, systemic anti-cancer therapy can be safely administered to SARS-CoV-2-positive patients not presenting with a severe form of the infection. Attention should especially focus on hematological adverse events. Typical COVID-19 symptoms (fever, cough or anosmia) should trigger SARS-CoV-2 RT-PCR testing to con rm or rule out SARS-CoV-2 infection.

Declarations
Ethics approval and consent to participate  Systematic RT-PCR test before each invasive procedure or admission of a hospitalized patient. Due to logistics, systematic RT-PCR was never recommended for ambulatory oncology patients admitted to the day unit.