Immune checkpoint inhibitor therapy and outcomes from SARS-CoV-2 infection in patients with cancer: a joint analysis of OnCovid and ESMO-CoCARE registries

Background As management and prevention strategies against COVID-19 evolve, it is still uncertain whether prior exposure to immune checkpoint inhibitors (ICIs) affects COVID-19 severity in patients with cancer. Methods In a joint analysis of ICI recipients from OnCovid (NCT04393974) and European Society for Medical Oncology (ESMO) CoCARE registries, we assessed severity and mortality from SARS-CoV-2 in vaccinated and unvaccinated patients with cancer and explored whether prior immune-related adverse events (irAEs) influenced outcome from COVID-19. Findings The study population consisted of 240 patients diagnosed with COVID-19 between January 2020 and February 2022 exposed to ICI within 3 months prior to COVID-19 diagnosis, with a 30-day case fatality rate (CFR30) of 23.6% (95% CI 17.8 to 30.7%). Overall, 42 (17.5%) were fully vaccinated prior to COVID-19 and experienced decreased CFR30 (4.8% vs 28.1%, p=0.0009), hospitalization rate (27.5% vs 63.2%, p<0.0001), requirement of oxygen therapy (15.8% vs 41.5%, p=0.0030), COVID-19 complication rate (11.9% vs 34.6%, p=0.0040), with a reduced need for COVID-19-specific therapy (26.3% vs 57.9%, p=0.0004) compared with unvaccinated patients. Inverse probability of treatment weighting (IPTW)-fitted multivariable analysis, following a clustered-robust correction for the data source (OnCovid vs ESMO CoCARE), confirmed that vaccinated patients experienced a decreased risk of death at 30 days (adjusted OR, aOR 0.08, 95% CI 0.01 to 0.69). Overall, 38 patients (15.8%) experienced at least one irAE of any grade at any time prior to COVID-19, at a median time of 3.2 months (range 0.13–48.7) from COVID-19 diagnosis. IrAEs occurred independently of baseline characteristics except for primary tumor (p=0.0373) and were associated with a significantly decreased CFR30 (10.8% vs 26.0%, p=0.0462) additionally confirmed by the IPTW-fitted multivariable analysis (aOR 0.47, 95% CI 0.33 to 0.67). Patients who experienced irAEs also presented a higher median absolute lymphocyte count at COVID-19 (1.4 vs 0.8 109 cells/L, p=0.0098). Conclusion Anti-SARS-CoV-2 vaccination reduces morbidity and mortality from COVID-19 in ICI recipients. History of irAEs might identify patients with pre-existing protection from COVID-19, warranting further investigation of adaptive immune determinants of protection from SARS-CoV-2.


INTRODUCTION
The efficacy of immune checkpoint inhibitors (ICIs) strongly relies on their capacity of inducing T-cell immune reconstitution. 1 T-cell exhaustion is a contributory mechanism underlying the severity of SARS-CoV-2 infection, 2 leading on one hand to the investigation of programmedcell death-1 inhibitors as a therapeutic strategy in severe COVID-19. 3 On the other hand, given the pathological immune-mediated mechanisms underlying COVID-19 and the risk of immune-pathology stemming from ICI use, there has been growing concern around the use of ICI in patients with COVID-19 and cancer. 4 5 Clinical data in support of a protective, as opposed to detrimental, effect of ICI in the prognosis of COVID-19 in patients with cancer have been inevitably biased by patient selection and underlying clinical characteristics. Initial reports revealed inconsistent results ranging from worse outcomes, 6 7 to no difference in COVID-19 severity 8 9 in ICI-exposed patients compared with ICIunexposed patients.
Large metanalyses have suggested no differential impact of ICIs on COVID-19 morbidity and mortality in comparison to other systemic anticancer therapies. 10 11 However, COVID-19 outcomes in patients with cancer have substantially evolved over time. Improved management of COVID-19, 12 immunization campaigns, 13 14 changes in community transmission and the emergence of new SARS-CoV-2 variants 15 have considerably changed the clinical impact of SARS-CoV-2 infection on patients with cancer since March 2020.
To date, a significant gap in knowledge remains as to whether the positive effect of SARS-CoV-2 vaccination observed in the general population extends to patients with cancer treated with ICI. Recent evidence suggesting that ICI may precipitate subclinical cytokine release following SARS-CoV-2 vaccination 16 strengthens the need to understand the relationship between COVID-19 vaccination and clinical outcomes.
With the aim of providing a contemporary description of COVID-19 morbidity and mortality in patients with cancer who were receiving ICIs at COVID-19 diagnosis and to assess the protective role of SARS-CoV-2 vaccination in this population, we developed this joint analysis of the OnCovid and European Society for Medical Oncology (ESMO) CoCARE registries.

Study design
OnCovid (NCT04393974) is a European registry study approved by the UK Health Research Authority (20/ HRA/1608) collecting data from consecutive patients fulfilling the following inclusion criteria: (1) age ≥18 years; (2) Reverse transcription polymerase chain reaction (RT-PCR) confirmed diagnosis of SARS-CoV-2 infection; (3) history of solid or haematological malignancy either active or in remission at the time of COVID-19 diagnosis.
The ESMO-CoCARE is an observational prospective study, based on a longitudinal multicenter survey of patients with cancer with any solid or hematological malignancy who were diagnosed with COVID-19.
For both registries, data from consecutive, all-comer patients were collected using electronic case report forms designed with the Research Electronic Data Capture software (Vanderbilt University, Nashville, Tennessee, USA). Study details and procedures, patients' eligibility, and clinical endpoints for both studies have already been extensively presented. 12-14 17-24 A list of participating centers with eligible patients for the present analysis is provided as online supplemental table 1.

Objectives and endpoints
The main objective of this analysis was to assess the protective role of SARS-CoV-2 vaccination in patients with cancer treated with a unique immunotherapy strategy, by comparing COVID-19 morbidity and mortality between unvaccinated and vaccinated patients.
In addition, we aimed to describe differences in COVID-19 severity and mortality depending on prior history of immune-related adverse events (irAEs) captured from ICI initiation until COVID-19 diagnosis.
Data of patients who received ICI within 3 months prior to COVID-19 diagnosis were merged from the OnCovid and ESMO CoCARE registries. Patients on chemotherapy-ICI and targeted therapy-ICI combinations were excluded from the analysis.
All-cause case fatality rate at 30 days (CFR 30 ) was chosen as the main clinical endpoint, to differentiate early COVID-19-related mortality, from late, likely cancerrelated deaths. As measures of COVID-19 morbidity, we evaluated the all-cause hospitalization and intensive care unit (ICU) admission rates, the rate of COVID-19 complications (at least one among acute respiratory failure, ARDS, kidney injury, secondary infections, sepsis, septic shock, acute cardiac injury, acute liver injury and others including thrombo-embolic events and other Open access coagulopathies, autoimmune diseases, gastrointestinal reactions), the receipt of at least one COVID-19-oriented therapy (including antivirals, chloroquine-based treatment, antibiotics, corticosteroids, interleukin-6 inhibitors and others) (yes vs no), and supplemental oxygen therapy requirement (yes vs no).
Patients who received two doses of the BNT162b2, mRNA-1273, ChAdOx1-S, and CoronaVac vaccines prior to COVID-19, or in case of infection diagnosed at least 28 days after a single dose of the Ad.26.COV2.S vaccine, were defined as fully vaccinated. Patients who received one vaccination, without meeting the above-mentioned time criteria, were considered partially vaccinated, while patients who received a third dose of either the BNT162b2 or mRNA-1273 vaccine (or a second dose after the Ad.26. COV2.S vaccine) were considered boosted. Considering the limited sample size of vaccinated patients with breakthrough infections in the study population, and that the electronic case report form of the ESMO-CoCARE registry was not designed to collect information on booster doses, patients were grouped as unvaccinated (including partially vaccinated) and fully vaccinated (either doubledosed or boosted patients) for all the comparative analyses, while patients with unknown vaccination status were excluded.
For the irAEs analysis, we evaluated COVID-19 outcomes according to the experience of any grade (National Cancer Institute Common Toxicity Criteria for Adverse Events, V.5.0) treatment-related side effects with a putative immune-mediated mechanisms at any time prior to COVID-19. These were previously evaluated by clinicians at participating sites during routine consultations as clinically indicated, without predefined time points, and collected retrospectively by investigators.
Considering the recognized role of lymphopenia as prognostic biomarker in patients with COVID-19, 25 we explored the association between the absolute lymphocyte count at COVID-19 (within 1 week of diagnosis) and the experience of prior irAEs in the subset of patients from the OnCovid registry. A detailed description of statistical analysis is provided as online supplemental methods. The most frequent primary tumor was lung cancer (47.1%), the majority of patients were male (67.5%), aged ≥65 years (62.1%), with at least one comorbidity (77.1%) and presented an active (76.7%), and advancedstage (80.2%) tumor (table 1).

SARS-CoV-2 vaccination is associated with improvement in COVID-19 outcomes in ICI recipients
After the exclusion of 13 patients with unknown vaccination status, 227 patients were included in the SARS-CoV-2 vaccine analysis.
None of the baseline demographics and oncological characteristics were associated with SARS-CoV-2 vaccination status, with the exception of a higher proportion of patients with at least one comorbidity among unvaccinated patients (80.5% vs 64.3%, p=0.0230) (online supplemental table 3).
Distribution of baseline patient characteristics prior to and after inverse probability of treatment weighting (IPTW) is reported in online supplemental table 5. Given the suboptimal balancing ability, country, comorbidities, tumor status and tumor stage were included in all IPTW-fitted multivariable logistic regression models for each COVID-19 outcome, which are reported in full as online supplemental table 6 and are summarized in the forest plot graph provided in figure 3. Compared with unvaccinated patients, full vaccination was associated with a decreased risk of death at 30 days (adjusted OR, aOR 0.08, 95% CI 0.03 to 0.26), of hospitalization (aOR 0.15, 95% CI 0.07 to 0.36), of COVID-19 complications (aOR 0.24, 95% CI 0.12 to 0.49) and of need for COVID-19-specific therapy (aOR 0.25, 95% CI 0.13 to 0.46). However, after clustered-robust correction for data source, the upper limit CI crosses one for all COVID-19 outcomes except for CFR 30 (aOR 0.08, 95% CI 0.01 to 0.69).
History of irAEs prior to COVID-19 is associated with decreased COVID-19 mortality in patients receiving ICI Overall, 38 patients (15.8%) experienced any grade irAEs at any time prior to COVID-19, which are summarized in online supplemental table 7. The median time from occurrence of irAEs and COVID-19 diagnosis was 3.2 months (range 0. 13-48.7, computed on data of 27 patients from the OnCovid registry).
The occurrence of irAEs was not associated with any of the baseline demographics and oncological characteristics, including the disease status (active vs remissive/in response) at COVID-19 (p=0.5339), with the exception of the primary tumor (p=0.0373) (online supplemental Distribution of baseline characteristics distribution prior to and after the IPTW is reported in online supplemental table 10. Given the suboptimal balancing ability, country, tumor stage, primary tumor and vaccination status were included in the IPTW-fitted multivariable logistic regression model for COVID-19 mortality, which confirmed that patients who experienced any grade irAEs prior to COVID-19 had a decreased risk of death at 30 days (aOR 0.47, 95% CI 0.23 to 0.99). Clustered-robust correction for data source further strengthened this finding (aOR 0.47, 95% CI 0.33 to 0.67) (online supplemental table 11).
Lastly, in a subset of patients from the OnCovid cohort, we revealed that the median absolute lymphocyte count within 1 week of COVID-19 diagnosis was significantly higher among patients who experienced any grade irAEs prior to COVID-19 than in those who did not experience irAEs (1.4 vs 0.8 10 9 cells/L, p=0.0098) (figure 4B).

DISCUSSION
Our study is the largest analysis on patients with cancer on ICIs diagnosed with COVID-19 to date. With the inclusion of patients diagnosed up until February 2022, it provides a more contemporary picture of COVID-19 outcomes in this specific population. Although merely descriptive due to the limited sample size of subgroups, the reducing CFR 30 across the pandemic phases suggests a time-dependent improvement of COVID-19 mortality, especially during the more recent Omicron outbreak, as already reported for the OnCovid population. 13 Even considering the time requirements for the delivery of immunization campaigns since the first SARS-CoV-2 vaccine approval, 26 and that most of the included Open access patients were diagnosed during the prevaccination phase, we consider 17.5% of full vaccination a relatively low rate, and a possible impact of vaccine hesitancy, as initially reported in early 2021, 27 28 cannot be excluded.
Although preliminary evidence from clinical trials supports the safety and immunogenicity of SARS-CoV-2 vaccines in patients with cancer on active ICI-based treatments, 16 29 30 this study demonstrates the efficacy of anti-COVID-19 vaccination in patients receiving ICI in routine clinical practice. The ~83% reduction in the CFR 30 in fully vaccinated patients along with COVID-19related morbidity is confirmed after adjustment for major prognostic confounders in IPTW-fitted models, a process made necessary by the inherent differences existing in study procedures and data collection modalities between the two registries.
The convergence of COVID-19 and ICI-toxicity in eliciting unopposed T-cell activation and downstream cytokine excess has been highlighted suggested as a hypothetical source of clinical risk to patients with cancer ever since the beginning of the pandemic. 5 31 Contrary to initial concerns, we document an association between the occurrence of irAEs and reduced CFR 30 : a novel finding of potential interest in the development of COVID-19specific therapeutics.
In our study, the protective role of irAEs of all grades on COVID-19-related mortality was independent of common clinicopathological features relating to cancer and COVID-19 prognosis, including SARS-CoV-2 vaccination status. It has been established that patients experiencing irAEs are those capable of mounting a more vigorous anticancer immune reconstitution, resulting in   Open access longer survival. 32 Because T-cell exhaustion is not solely a hallmark of cancer progression but a mechanism of COVID-19 severity, 25 31 we speculate whether history of prior irAE might be a surrogate of more functional T-cell immunity, leading to improved mortality from COVID-19 irrespective of vaccine status.
In keeping with this view, we found that the absolute lymphocyte count at COVID-19, was significantly higher among patients who experienced prior irAEs. It is well known that patients with severe COVID-19 show reduced counts of peripheral CD4+and CD8+ T cells 31 , and that reduced CD4+/CD8+T cells, B cells, NK cells, and absolute lymphocyte cell count levels are significantly associated with COVID-19 mortality in the general population. 25 At the same time, the known mechanisms leading to irAEs involve expansion of intratumoral and peripheral T-cell receptor repertoires along with a mobilization of large numbers of T cells 33 34 and, to a lesser extent, activation and exhaustion of CD21 low B cells. 35 On the other hand, a decrease in the absolute lymphocyte count has been reported with severe ICI-associated myocarditis. 36 While OnCovid and ESMO CoCARE registries lack information on T-cell phenotype at COVID-19 diagnosis, our findings are provocative in suggesting that prior irAE might represent a hallmark of protection from COVID-19 mortality through invigorated T-cell immunity. These findings deserve further mechanistic studies to fully elucidate the immunological links between irAEs and COVID-19 outcomes in patients with cancer.
Our study acknowledges several limitations, including lack of data regarding the smoking status and more detailed information regarding irAEs duration and management. Of note, previous irAEs and their putative immune-mediated mechanism were assessed at participating site in routine practice, without predefined time points. This might have impacted the quality of data with risks of underreporting, as the 16.7% and 3.1% rates of all grade and ≥G3 irAEs, respectively, are lower than those reported in interventional clinical trials with ICIbased regimens, 37 but comparable to reports from clinical practice. 38 In addition, inherent differences between the two registries significantly impacted the accuracy of the estimates from the vaccination analysis: information about booster doses only recently started to be collected for patients entered in the ESMO CoCARE registry and was not available for our analysis. Furthermore, for ~24% of vaccinated patients, the specific type of vaccine could not be reconstructed. While constituting an important limitation, this is unlikely to have affected our results, given recent evidence suggesting largely comparable efficacy of commonly available SARS-CoV-2 vaccines. 39 Lastly, despite the inclusion of a significant proportion of more recently diagnosed patients, the lack of availability of viral genomic sequences across the pandemic phases did not allow us to make conclusive considerations about new SARS-CoV-2 variants, while the limited sample size of the 'alpha-delta' and 'omicron' phases subgroups prevented us from running adequately powered timeadjusted analyses.
Despite the mentioned limitations, our results collectively support the notion that ICI recipients are not especially vulnerable to COVID-19, with mortality rates that are in keeping with the general population with COVID-19 and cancer. In these patients, SARS-CoV-2 vaccination leads to significantly improved outcome from COVID-19, comparably with other oncological patient populations. 13 14 40 Considering the continuously expanding indication for ICI therapy, 41 our findings are of utmost importance to ensure effective utilization of this therapy during and beyond the SARS-CoV-2 global epidemic.
Patient consent for publication Not applicable.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available on reasonable request. Individual, deidentified participant data and data dictionary may be made available at the request of investigators whose proposed use of the data has been approved by the OnCovid consortium and ESMO CoCARE steering committees following review of a methodologically sound research proposal.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.
Open access This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any