key: cord-0796835-iw111pg9
authors: Booth, Stephen; Willan, John; Wong, Henna; Khan, Dalia; Farnell, Rachel; Hunter, Alicia; Eyre, Toby; Katz, Harley; Dungarwalla, Moez; Chen, Lucia; Browning, Joe; Polzella, Paolo; Gray, Nicola; Neelakantan, Pratap; Dhillon, Elissa K; Dutton, David; Sternberg, Alex; Prideaux, Steven; Collins, Graham P.; Peniket, Andy
title: Regional outcomes of severe acute respiratory syndrome coronavirus 2 infection in hospitalised patients with haematological malignancy
date: 2020-06-16
journal: Eur J Haematol
DOI: 10.1111/ejh.13469
sha: 14a10207629825f528885f6872c7b2aab2360f81
doc_id: 796835
cord_uid: iw111pg9

OBJECTIVES: We sought to characterise the outcomes of patients with haematological malignancy and SARS‐CoV‐2 infection in hospital in our regional network of 7 hospitals. METHODS: Consecutive hospitalised patients with haematological malignancy and SARS‐CoV‐2 infection were identified from 01/03/2020 to 06/05/2020. Outcomes were categorised as death, resolved, or ongoing. The primary outcome was preliminary case fatality rate (pCFR), defined as the number of cases resulting in death as a proportion of all diagnosed cases. Analysis was primarily descriptive. RESULTS: 66 Patients were included, overall pCFR was 51.5%. Patients ≥ 70 years accounted for the majority of hospitalised cases (42, 63%) and fatalities (25, 74%). Mortality was similar between females (52%) and males (51%). Immunosuppressive or cytotoxic treatment within 3 months of the diagnosis of SARS‐CoV‐2 infection was associated with a significantly higher preliminary Case Fatality Rate (pCFR) of 70%, compared with 28% in those not on active treatment (P = 0.0013, 2 proportions z‐test). CONCLUSIONS: Mortality rates in patients with haematological malignancy and SARS‐CoV‐2 infection in hospital are high supporting measures to minimise the risk of infection in this population.

The pandemic spread of COVID-19 (Coronavirus disease 2019) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has so far caused over 400,000 confirmed deaths worldwide. Preliminary pre-print data from 16,749 hospitalised patients in the United Kingdom (UK) suggest a mortality of at least 33% in unselected hospitalised patients [1] . Risk factors associated with more severe outcomes include increasing age, pre-existing lung disease, diabetes, hypertension and cancer [2] . The risk of SARS-CoV-2 to patients with cancer, and particularly haematological malignancy, is not yet fully clarified.

Early data suggested that patients with cancer are at increased risk of death if contracting this virus [3] [4] [5] [6] . However, numbers of patients with malignancy were low with marked heterogeneity of diagnoses, and little information was presented on treatment history. With regard to haematological malignancies, a cohort study from Wuhan, China, identified 11 patients with haematological malignancy and COVID-19, of whom 8 (72%) did not survive [7] . A further study of 25 patients, of whom 24 had a malignant haematological diagnosis, reported a one-month mortality rate of COVID-19 infection of 40% [8] .

Two recent multicentre retrospective cohort studies of patients with COVID-19 compared patients with cancer with age-matched controls without cancer. Both of these studies have suggested that patients with haematological cancers may be at greatest risk of severe complications of COVID-19 when compared with other malignancies. The first study reported a higher odds ratio of death of 2.3 in 105 patients with cancer compared to age-matched controls [9] . The 9 patients with haematological malignancy had the worst outcomes, with significantly increased risk of intensive care admission, need for ventilation, and death. The second cohort study reported on 218 COVIDpositive patients with a malignant diagnosis, reporting an increased case fatality rate of 2-3 compared to age matched controls [10] . The 54 patients with haematological malignancy had a case fatality rate of 37%, worse than that of solid organ cancers.

Patients with haematological cancers are likely to be at high risk of infectious complications of viral respiratory infections from both immune dysregulation as an intrinsic part of the malignancy, as well as of the immunosuppressive and cytotoxic treatment [11] . In time it will be important to identify from large datasets the diagnoses and treatments which convey the greatest risk to this patient group, and large prospective studies are underway to assemble this information.

This article is protected by copyright. All rights reserved We present outcome data for all hospitalised COVID-19 positive patients with a diagnosis of haematological malignancy from our region, which includes seven hospitals serving a population of 2.8 million patients [12] , as a first attempt to understand in more detail the outcome of these patients.

Consecutive cases were identified prospectively by clinical teams across our regional cancer network from 01/03/2020 to 06/05/2020 and reported to a central database. Patients were required to be hospitalised and positive for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) RNA by reverse transcriptase qPCR of nose and throat swab, or with clinical and radiological features consistent with COVID-19, where the clinical team judged COVID-19 was the most likely diagnosis. All patients had a current haematological malignancy under ongoing treatment or in clinical follow up. COVID-19 was treated according to local practice with many patients entering clinical trials.

Patients who were not admitted to hospital were not included in the analysis because of variation in outpatient testing strategy over time and between hospitals. Patients with asymptomatic nonmalignant conditions, for example monoclonal gammopathy of uncertain significance, were excluded. Patient baseline characteristics collected included age, gender, haematological diagnosis, method of diagnosis of SARS-CoV-2, current haematological treatment and prior lines of treatment.

The primary outcome was preliminary case fatality rate (pCFR), defined as the number of cases resulting in death as proportion of all diagnosed cases [13] . Outcomes were categorised as either death; resolved (patients who were no longer symptomatic and judged to have recovered from the infection by their clinical team); or ongoing (patients remained in hospital with symptoms attributed to SARS-CoV-2 infection).

Analysis was primarily descriptive with the two proportions Z-test used to compare pCFR in the population of patients who received immunosuppressive or cytotoxic treatment in the 3 months prior to SARS-CoV-2 infection and the population who did not receive such treatment.

This article is protected by copyright. All rights reserved In total there were 34 deaths: therefore the overall pCFR was 51.5%, with patients over the age of 70 accounting for the majority of cases 42 (64%) and fatalities 25 (74%) (Figure 1 ). Numbers of deaths in each age group were: <60 years, 4; 60-69 years, 5; 70-69 years 14; >80 years, 11. The pCFR was similar between female and male patients (52% and 51% respectively). Mortality rates were consistently high across diagnostic groups, particularly the myeloid malignancies and 

This article is protected by copyright. All rights reserved Examining outcomes according to modality of treatment, only 6 patients had received anti-CD20 therapy in the past 2 years, of whom 4 died. No patients had received purine analogues in the past 2 years. One patient in the cohort who was 27 months post allogeneic stem cell transplant, having relapsed and received further myeloma therapy. This patient did not survive. A further 2 patients were within 2 years of autologous stem cell transplant, one with myeloma who did not survive and one with primary central nervous system lymphoma, who recovered.

At present the true incidence of COVID-19 in patients with haematological malignancy is not known, since only those with symptoms sufficient for hospital admission, or those already admitted who developed symptoms, were tested, in line with national guidance at the time. The true case fatality rate of COVID-19 in this patient group is therefore likely to be significantly lower than that reported in this paper. In line with other studies, advancing age appears to be a key correlate of poor outcome in patients hospitalised with COVID-19 infection, with a pCFR amongst those with haematological cancers of 59.5% in those over 70 years, as compared to 37.5% in those under 70. (Figure 1) .

A relevant question is what proportion of the regional population with haematological malignancy these 66 patients represents. Hospital data on the numbers of patients fulfilling the UK government's criteria for shielding were available from 2 hospitals covering a total of population of 1.27 million. In these hospitals 3334 patients were identified, equivalent to 263 per 100,000 population, which accords with the UK Haematological Malignancy Research Network prevalence data indicating a prevalence of 167 / 100000 for haematological malignancy diagnosed in the last 3 years or 388 / 100000 diagnosed in the last 10 years [14] . We therefore estimate a regional population of between 4,670 and 10,850 patients with haematological malignancy, indicating the only a small proportion (0.6-1.4%) of patients were admitted with SARS-CoV-2 infection in the period of this series.

It is striking to see that patients receiving chemotherapy within the 3 months preceding their COVID-19 diagnosis have a statistically significantly higher pCFR (62%) than those who have not recently had chemotherapy (28%) (Figure 2 ). This may reflect the immunosuppressive effects of the chemotherapy, or of the underlying condition itself, or may reflect the increased frailty associated with active malignancy. The relatively small number of cases involved precluded

This article is protected by copyright. All rights reserved formal multivariate analysis of potential confounding factors, and it is therefore not possible to determine if receiving therapy is an independent risk factor for mortality. Equally the small numbers of patients receiving any one treatment or class of treatments precludes analysis of the effect of specific treatment types on outcome.

A further important point is that a proportion of this patient group will have a limited prognosis from their haematological malignancy or comorbidities. Of the 37 patients who had received cytotoxic or immunosuppressive treatment only 8 could be classified as being given with curative intent. Of the other 29 patients, 9 would be expected to give relatively durable disease control, for example first line treatment of follicular lymphoma or myeloma; tyrosine kinase inhibitors for chromic myeloid leukaemia; or cytoreduction for polycythaemia vera or essential thrombocythaemia. In the remaining 20 patients prognosis would be expected to be more limited These data do indicate, however, that patients with haematological malignancies requiring hospital admission have a high mortality rate, supporting measures to minimise the risk of SARS-CoV 2 exposure in this patient group. The finding is also in agreement with preliminary data from a very large population cohort study of 17,425,445 adult patients in England, the pre-print data from which suggest a diagnosis of haematological malignancy within 5 years has been associated with at least a 3 times greater risk of death in hospital from COVID-19 during the period 01 February to

This article is protected by copyright. All rights reserved 25 April 2020 [15] . A subsequent UK observational study of 800 patients with cancer included 167 patients with haematological malignancy [16] . Although the risk of death was not significantly increased in patients with haematological malignancy as compared to other cancers in this population with diagnosed infection, it is striking how high a proportion of included patients had a haematological diagnosis, although it is unclear whether this relates to increased risk of infection, risk of developing more severe disease, or likelihood of testing. Interestingly this study did not find a difference in mortality between those who had received cancer treatment in the last 4 weeks and those who had not, although this analysis was of the entire population including nonhaematological cancers and the treatments given were very heterogenous.

Our data support attempts to reduce the contact of individuals from this group with the healthcare system to minimise nosocomial infections [17] . This group of patients need to be prioritised in consideration of how best to use a SARS-CoV-2 vaccine, and included in clinical trials of novel therapies to treat COVID-19.

The submission of patient data to national and international databases is strongly encouraged, since numbers are insufficient at present to answer the questions that clinicians and patients alike are posing, regarding the relative risks of different diagnoses and treatments. These, more complete data will, in time, be fundamental in enabling us to build an informed consensus about future management of haematological malignancy in the era of COVID-19. 

Case Fatality Rate of Cancer Patients with COVID-19 in a New York Hospital System

ECIL-4): guidelines for diagnosis and treatment of human respiratory syncytial virus, parainfluenza virus, metapneumovirus, rhinovirus, and coronavirus

Clinical commissioning group population estimates

Estimating case fatality rates of COVID-19

HMRN Prevalence

OpenSAFELY: factors associated with COVID-19-related hospital death in the linked electronic health records of 17 million adult NHS patients