Does COVID‐19 pandemic impact cancer outcomes in metastatic setting? A comparative cohort study among metastatic patients treated at day care hospital

Abstract Introduction COVID‐19 outbreak rapidly spread since early 2020 leading to the implementation of nationwide lockdowns. To cope with this sudden change, management guidelines were quickly published to adapt oncological care, with potential impact on cancer outcomes. Methods We conducted a retrospective comparative cohort study to assess the impact of the COVID‐19 outbreak in 2020 on cancer outcomes in metastatic patients. Two cohorts of metastatic patients receiving intravenous (iv) therapy in a French oncological day care hospital were assessed: a 2020 cohort during the first French lockdown, and a 2018 historical cohort before the COVID‐19 pandemic. We performed a propensity score analysis to match patients from the two cohorts. After one‐year follow‐up, we compared progression‐free survival (PFS) and overall survival (OS) between cohorts. Adaptations of medical oncological treatments in 2020 were also analysed. Results The 376 patients of the 2020 cohort were matched with 376 of the 2018 cohort. No SARS‐CoV‐2 infection was observed in the 2020 cohort. The adjusted PFS was significantly shorter in 2020 compared to 2018 (HR = 1.23; 95% CI: 1.03–1.46), as well as among patients without treatment adaptation compared to matched patients of the 2018 cohort (HR = 1.33; 95% CI: 1.10–1.61). We did not observe any significant difference of PFS among the group with treatment adaptations. OS was not significantly different. Conclusion Metastatic cancer patients treated during the first lockdown had a higher risk of disease progression 1 year after COVID‐19 outbreak. However, oncological treatment adaptations or SARS‐CoV‐2 infections do not explain these results. A longer follow‐up is needed to observe the impact on OS.


| INTRODUCTION
The new coronavirus disease 2019 (COVID-19) first emerged in late 2019 in Wuhan, China, and rapidly grew into a global pandemic.As of late 2022, there were over 634 million confirmed cases of COVID-19 and 6.6 million deaths reported. 1In France, the first severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection was diagnosed in January 2020. 2 The pandemic rapidly spread afterward, leading to the implementation of a first nationwide lockdown from 17 March to 11 May, 2020.
1][12] In particular, hospital admissions were avoided and management at home was encouraged.In this context, the use of telemedicine instead of on-site monitoring visits was encouraged, as well as replacement of intravenous drugs with oral drugs, adjustment of chemotherapy regimens or shortening of radiotherapy treatments.Moreover, patients with slowly evolving metastatic cancers could have temporary breaks in their treatment.Besides, when healthcare facilities were challenged, priority was given to patients either with curative-intent therapeutic strategies or with the longest life expectancy.Furthermore, cancer supportive care was adapted in order to avoid hospitalisations due to oncological treatment complications (e.g.extended use of granulocyte-colony stimulating factor in primary prophylaxis). 13,14ue to these pandemic-related adaptations, cancer outcomes may have been affected, but remain poorly documented to date. 15In particular, there are no data dedicated to prognostic of metastatic cancer patients managed during the COVID-19 pandemic.Yet, it may be assumed that these patients were particularly vulnerable to the effects of COVID-19 on the healthcare pathway and, as a consequence, more likely to progress in the short to medium term.
We have previously conducted a prospective study (NCT04366154) in adult patients with solid/hematologic malignancy with ongoing medical treatment during the first COVID-19 lockdown in the outpatient clinics of two French comprehensive cancer centres.It highlighted that medical oncology practice was modified for 27% of the 734 patients enrolled, especially 29% for the 435 patients in metastastic setting. 16Clinical outcomes of these medical adjustments were not investigated.In this context, we aimed to assess the impact of the COVID-19 outbreak on cancer outcomes in metastatic patients undergoing an anticancer treatment per intravenous (iv) route in the day care hospital, as compared with a historical cohort of metastatic patients treated before the COVID-19 pandemic.

| Study design and patients
This study is a retrospective comparative cohort study of adults with metastatic solid tumour who received an iv anti-cancer treatment (either chemotherapy, immunotherapy and/or targeted therapy) at the day care hospital of the comprehensive cancer Centre François Baclesse, Caen, France.Patients receiving exclusive oral therapy or already participating in a clinical trial were not eligible.The study compared a cohort of patients treated during the first nationwide COVID-19 lockdown from March to June 2020 (the 2020 cohort) to a historical cohort of patients treated from March to June 2018 (the 2018 cohort) in the outpatient department.
The 2020 cohort consisted of patients from the COVIPACT study who met the eligibility criteria.The COVIPACT study was a one-year longitudinal prospective study initiated at the beginning of the first French nationwide lockdown: it aimed to assess adjustments in oncological practice, as well as psychological consequences. 16t enrolled 734 participants who were at least 18 years old, were treated for solid or hematologic cancer in two regional cancer centres during the first COVID-19 lockdown (Clinical trial NCT04366154), of whom 435 patients were in metastatic setting.
The 2018 historical cohort was constituted retrospectively from medical records of the day care hospital.It included all eligible metastatic patients receiving an anti-cancer treatment per iv route between March and June 2018.

| Data collection
This retrospective study was approved by the institutional review board.It was conducted in compliance with the French Research Standard MR-004 'Research not involving Human participants' (compliance commitment to MR-004 for the Centre François Baclesse number 2214228 v.0, dated from 07/03/2019).It is registered with the French Health Data Hub under the reference F20210415165210.All patients received information and none of them expressed opposition to the use of their data.
Collected data included sociodemographic parameters and clinical information, such as patient characteristics, cancer diagnosis and treatments, including ongoing treatment at the time of inclusion (between March and June for each cohort), clinical cancer outcomes and survivals.For the 2020 cohort, data were extracted from the database of the COVIPACT study for patients.They also included information on adjustments in oncological treatments, based on the physician decision after a multidisciplinary committee discussion considering clinical characteristics of the patients, ongoing treatment and status of the disease.Survival outcomes (disease progression and/or death) were also collected within 1 year after inclusion for the purpose of our study.For the 2018 cohort, data were collected from the registry database and clinical records of patients.

| Study purposes and statistical analysis
The primary endpoint was to compare the progressionfree survival (PFS) of patients between the 2020 cohort and the 2018 cohort, with a one-year follow-up.Secondary objectives were to assess and compare overall survival between the two cohorts, and to evaluate the PFS outcomes of patients with and without cancer treatment adaptations in the 2020 cohort, as compared to matched patients in the 2018 cohort.
Data were described by using number and proportion for categorical variables, mean and standard deviation for continuous variables.Comparisons of patient's characteristics between cohorts were assessed by the use of Pearson's chi-squared test for categorical data, and by independent sample t-tests for continuous data.
PFS was defined as the time elapsed from the patient's start date of the ongoing treatment at inclusion to disease progression, or death whatever the cause.OS was defined as the time elapsed from the patient's start date of the ongoing treatment at inclusion to death.Patients who remained without an event by the end of the study period (30 June 2019 for the 2018 cohort and 30 June 2020 for the 2020 cohort) were censored.PFS and OS were estimated by the Kaplan-Meier method.Univariate and multivariable Cox models were performed to measure the difference of survival outcomes between cohorts with adjustment on patient's characteristics and treatment (age, sex, body mass index (BMI), performance status, primary tumour location, number of treatment lines, and treatment protocol (distinguishing four categories: chemotherapy, chemotherapy + targeted therapy, immunotherapy, targeted therapy); time from diagnosis being closely related to number of treatment lines was not included in the multivariable model).
To reduce the effects of confounding variables due to the non-randomised data in the estimation of COVID-19 pandemic on survival outcomes, we conducted a propensity score matching.The propensity score, here defined as the probability of cohort assignment, was estimated by using a multivariable logistic regression model, containing age, sex, primitive tumour location, European Cooperative Oncology Group Performance Status (ECOG PS) and treatment protocol.Patients of the 2020 cohort were matched with patients of the 2018 cohort in a 1:1 ratio, without replacement, by using the nearest neighbour method.Additionally, separated analyses were then conducted among patients for which treatment was adjusted in 2020 due to the pandemic, and patients for which treatment was administered as usual, each of these two sub-samples being matched with samples of the 2018 cohort.
All statistical analyses were carried out using R statistical software (4.1.2).A two-side p-value less than 0.05 was considered statistically significant.

| Screened population
Among the 435 patients in metastatic setting enrolled in the COVIPACT study, 377 patients undergoing iv treatment at day care hospital of the Centre François Baclesse between March and June 2020 met the eligibility criteria.From medical records of our centre, 501 metastatic patients treated with iv treatment at day care hospital at the same period in 2018 were identified (Figure 1).The sociodemographic and clinical characteristics of these patients are presented in Table S1.

| Analysed population
Propensity score analysis including age, sex, primitive tumour location, ECOG PS and treatment protocol allowed to match 376 patients of the 2020 cohort with 376 patients of the 2018 cohort, the remaining patient of the 2020 cohort being not matched because of missing ECOG PS (Figure 1).The sociodemographic and clinical characteristics were globally balanced between the two cohorts (Table 1).Women accounted for 61.8% of patients, the median age was 65.4 years and the three main cancer locations were breast (31.1%), digestive (25.5%) and lung (20.9%).Most of the patients had an ECOG PS of 0 or 1 (87.4%) and were in first or second line of treatment (76.6%).Chemotherapy with or without targeted therapy was the most prescribed treatment (70.2%), followed by targeted therapy alone (17.7%) and immunotherapy (12.1%).
None of the enrolled patients experienced a SARS-CoV-2 infection and no patient died from COVID-19 during the first French lockdown.
The characteristics of the patients with treatment adaptations are presented in Table 2.In comparison to the whole 2020 cohort, the subgroup with treatment adaptations was particularly different.Breast cancer was still the most represented but with a greater proportion (37.5% vs. 28.7%).Lung and head and neck cancers were also more represented (31.7% and 12.5%, respectively vs. 21.8% and 8.2%), whereas digestive and gynaecological cancers were less concerned (6.7% both vs. 24.5% and 10.1%, respectively).Patients were still mostly in first line of treatment but with a higher predominance on the other lines (66.3% vs. 51.3%).While 68.1% of patients included in the 2020 cohort received chemotherapy with or without targeted therapy, only 36.5% of those with treatment adaptations were treated with these types of treatment.Targeted therapy alone (e.g.Trastuzumab) and immunotherapy alone (e.g.Pembrolizumab) were the main treatment protocols (40.4% and 23.1%, respectively vs. 17.8% and 14.1%).
The characteristics of the subgroup without treatment adaptation were more similar to the whole 2020 cohort, especially concerning the proportion of digestive cancers and chemotherapy (Table 2).
In the multivariable model, after adjustment on sex, ECOG PS, number of treatment lines, cancer location  and treatment protocol as well as age and BMI, PFS was significantly decreased in the 2020 cohort (HR = 1.23; 95% CI: 1.03-1.46;p = 0.019; Table 3).

| Treatment adaptations
One hundred three patients with pandemic-induced treatment adaptations of the 2020 cohort were matched with 103 patients of the 2018 cohort.After adjustment on sex, ECOG PS, BMI, age, number of treatment lines, cancer location and treatment protocol, there was no significant difference in PFS according to time period (HR = 0.71; 95% CI: 0.48-1.07;p = 0.10; Table 4).For patients without treatment adaptation, using the same matching and adjustment, we found a significant difference in PFS according to time period, with a higher risk of progression in 2020 (HR = 1.33; 95% CI: 1.10-1.61;p = 0.004; Table 4).

| DISCUSSION
In this retrospective comparative cohort study, we investigated for the first time the impact of COVID-19 outbreak on cancer outcomes among metastatic patients undergoing iv treatment at a day care hospital.With one-year follow-up, the adjusted PFS was significantly shorter in the 2020 cohort compared to the 2018 cohort.In the subgroup of patients without cancer treatment adaptation due to the pandemic, PFS was shorter as well.None of the patients included in 2020 was infected by SARS-CoV-2 and no difference of PFS was observed among the subgroup with treatment adaptations as compared to a matched sample of the 2018 cohort.OS was not significantly different between the two cohorts.
[19] T A B L E 3 Univariable and multivariable analysis for PFS in the analysed population.For example, in a French study, there were poorer OS rates in 2020 for patients diagnosed with colorectal cancer and infected with SARS-CoV-2, compared to 2018 and 2019. 20n contrast to these studies, the COVID-19 pandemic was not involved in the survival outcomes of our patients.This may be partly explained by the fact that no patients in the 2020 cohort experienced SARS-CoV-2 infection during the first French lockdown.The absence of SARS-CoV-2 infection may result from the implementation of adapted healthcare pathway for these patients at higher risk of contamination, while limiting hospital admissions and favouring management at home.Another before-after study conducted in Portugal enrolled patients diagnosed with cancer during the first lockdown period in 2020 with different cancer locations and compared them to patients diagnosed in the same period of 2019.Among the 763 patients diagnosed in 2020, only nine were infected by SARS-CoV-2.After a follow-up of 4 months, there was an increase in overall short-term mortality of diagnosed cancer patients after COVID-19 outbreak, especially for patients with stage III cancer and those undergoing surgical treatment or radiotherapy. 21Increase of short-term mortality was partly explained by adaptations of healthcare system during the COVID-19 pandemic.Indeed, diagnosis delays were observed in the first half of 2020 as well as presentations with more clinically advanced conditions. 22urthermore, some modelling studies have predicted a long-term effect on survival outcomes, up to 5 years. 23,24n our study, we included metastatic cancer patients undergoing iv treatment, which represents a different  oncology population.With a longer follow-up of 1 year, OS was not significantly different after the COVID-19 outbreak.Nevertheless, there was a higher risk of disease progression in the 2020 cohort as well as in the subgroup without treatment adaptation, possibly because of similar characteristics as compared to the 2018 cohort.At first, similarly to the Portuguese study, changes in access to care may be involved, especially due to delays in routine follow-up reported in 2020. 25Follow-up of metastatic disease may have been impaired by the pandemic, with, for example, delays in follow-up imaging or appointments.

Univariable analysis
Hospital avoidance may also have caused patients to wait to seek medical attention in case of symptoms.Moreover, psychosocial effects of COVID-19 outbreak on cancer patients may also be considered.Indeed, anxiety and depression symptoms during the COVID-19 pandemic have been reported in several studies from different countries. 16,26,27he potential role of psychosocial factors on cancer outcomes has been studied.In a meta-analyse, chronic daily life event stress, severe life events, depression and social isolation favoured cancer growth and metastasis, concluding that psychosocial factors may represent risk factors for cancer incidence and evolution. 28eassuringly, we did not observe significant change in PFS for patients in metastatic setting with adapted treatment in 2020.This probably reflects the relevance of guidelines and appropriate patient identification for implementation of adapted cancer management.According to guidelines, if the healthcare system was overwhelmed, a triaging could be done to prioritise oncological emergencies and curative treatments over palliative treatment for advanced pathologies. 11,29In other cases, the adaptations in oncological treatment program were weighted by the need to continue therapies aiming at curing and treating cancer, as mortality from untreated active malignancy remains extremely high and by the modest risk of severe COVID-19 due to systemic anticancer therapy. 30,31n an American cohort study that focused on metastatic solid cancers, there were no delays in systemic treatment initiation or preference against use of myelosuppressive therapies. 32In our study, capacities of the day care hospital, although modified, were not overwhelmed and adaptations of treatment plans were made for a selected group of metastatic patients of good prognosis (mainly monotherapy with immunotherapy/targeted therapy for patients with a good performance status) while patients with a poorer prognosis treated with chemotherapy or combined treatments continued their program.Likewise, in other studies, short-term cancer outcomes have not been worsened by modifications of treatment plans. 20,33imilarly, recent observations by Tran et al. from a large retrospective cohort addressing the impact of the COVID-19-related health care disruptions on pathologic cancer staging: no statistically significant differences in pathologic stage or features between cancers staged in the first year of the pandemic as compared with cancers staged in the 2 years before the pandemic were found. 15hese various findings addressing different facets of cancer care pathway thus provide reassuring insight into cancer care patterns during the pandemic, from diagnosis to management of metastatic cancer patients.They support the insightfulness of the recommendations by learned cancer societies for adaptations of cancer care to cope with such a pandemic.They also demonstrate that despite disturbances in delivery of health care services, the prioritisation of cancer procedures during times of reduced capacity allowed to maintain a similar quality of cancer management.
We constituted a large cohort of 752 patients with multiple cancer sites, which is, to our knowledge, the first to explore the impact of COVID-19 outbreak on cancer outcomes of metastatic patients undergoing an iv treatment at day care hospital.In contrast to many other studies evaluating the relationship between cancer and SARS-CoV-2 infections, there was no patient infected in our study, which allowed us to exclude any involvement on cancer outcomes.Although it was not feasible to realise a 'gold standard' control study without COVID-19 involvement, we formed a historical cohort consisting of patients treated in 2018 over the same period as the 2020 cohort.A propensity score matching was made to circumvent the lack of randomisation between the two groups and to enhance comparability between them.
However, this study has some limitations.First, our data came from a single French cancer centre so that it could not reflect great variations of the SARS-Cov-2 infection rate and healthcare overwhelming by regions and countries.In addition, we did not collect information on socioeconomic status of patients, or other relevant characteristics, including qualitative data, that may have provided some additional relevant context and may have been helpful to nuance the interpretation of the findings.Moreover, we collected whether patients had progressed or died but we did not collect disease-related or treatmentrelated complications that may have impacted PFS.In addition, the follow-up of our study was short and a longer follow-up seems necessary to assess further information on long-term survival outcomes.The natural evolution of the pandemic and the implementation of vaccination campaigns since 2021 should also be considered.

| CONCLUSION
In this comparative retrospective cohort study, the adjusted PFS at one-year follow-up was significantly shorter in the 2020 cohort compared to our historical 2018 cohort.PFS was similarly shorter in the subgroup without treatment adaptation, arguing for a potential effect of the pandemic on metastatic disease progression of patients receiving iv treatment at day care hospital.The changes in access to care and psychosocial issues may be involved.In the subgroup with treatment pandemicrelated adaptations in 2020, PFS was no more significantly different.The impact of treatment adaptations was reassuringly limited by following relevant guidelines to select patients with a better prognosis.OS was also not significantly different.Our findings could serve as valuable feedback in the event of future pandemic.Furthermore, it may be also relevant to involve patient advocacy groups in discussions to better integrate the patient perspective in the care prioritisation and to facilitate patient's adhesion to the therapeutic options and care management.Nevertheless, a longer follow-up is necessary to better understand the impact on long-term survival outcomes. .

a:
ECOG PS, Eastern Cooperative Oncology Group performance status

T A B L E 4
PFS (multivariable model) for matched samples of patients of the 2020 cohort with or without treatment adaptations with patients of the 2018 cohort.Matched samples -treatment adaptations in 2020 (N = 206) a Matched samples -no treatment adaptations in 2020 (N = 546) b Characteristics of the analysed population (N = 752).Characteristics of patients of the 2020 cohort according to pandemic-induced treatment adaptations (N = 376).
T A B L E 1 T A B L E 2Abbreviation: BMI, body mass index.
in the 2020 cohort with treatment adaptations matched to another group of 103 patients from the 2018 cohort.b 273 patients in the 2020 cohort without treatment adaptations matched to another group of 273 patients from the 2018 cohort.
a 103 patients