Immunocompromised patients with SARS-CoV-2 infection in intensive care units, outcome and mortality

Background: The new severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) outbreak severely hit Northeastern France from March to May 2020. The massive arrival of SARS-CoV-2 positive patients in the intensive care units (ICU) raised the question of how immunocompromised patients would be aﬀ ected. Therefore, we analyzed the clinical, biological and radiological features of 24 immunocompromised ICU patients with severe SAR-CoV-2 infection. Results: The mortality rate was signiﬁ cantly higher for immunocompromised patients compared with other patients (41.7% versus 27.3%, respectively, p = 0.021). Mortality occurred Mortality rate of SARS-CoV-2 acute respiratory syndrome in immunocompromised patient is high. No treatment was associated with survival improvement. Prolonged full-code management is required for these patients.


Background
The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) epidemic affected Northeastern France early on, with an uncontrolled cluster arising near our hospital in March 2020 [1,2]. Our Intensive Care Units (ICU) were confronted with a massive number of patients in need of critical care. In our ICU, critically ill and immunocompromised patients (ICP) have been treated for decades with the same consideration as other critically ill patients. However, with the massive arrival of patients with severe SARS-CoV-2 infection, we feared an overwhelming in lux of infected immunocompromised patients. We were concerned that these patients would be severely affected by SARS-CoV-2. Therefore, we conducted a monocentric observational study of all immunocompromised patients admitted to ICU during this outbreak. The objective was to describe their clinical characteristics and outcome. treatment with immune suppressive drugs for any medical condition were considered to be immunocompromised (ICP). Information on demographics, medical history, SARS-CoV-2 infection, biological parameters, treatments and outcome were collected. Simpli ied acute physiology score (SAPSII) [3] and sequential organ failure assessment score (SOFA) [4] were calculated. Radiological severity was assessed by an independent radiologist, according to the Fleischner Society's consensus statement [5] as mild, moderate-to-severe or critical lesions. A SARS-CoV-2 swab sample was performed at admission and then weekly during the ICU stay and analyzed using Real Time Polymerase Chain Reaction (RT-PCR). All patients received a notice of information and this study was approved by the local institutional review board. The patients were not participating to any other trial concerning SARS-CoV-2 during the study period.
Continuous variables are expressed as medians with interquartile ranges (IQR). Categorical variables are described as counts with percentages. Comparisons were performed using Fisher's exact test, Wilcoxon rank-sum test and chi-square test of independence. Overall ICU survival was de ined as the time between ICU admission and the date of death or last follow-up. Survival curves were established with a Kaplan-Meier's method and the difference between immunocompetent and immunocompromised patients was analyzed by a log-rank test. Analysis were performed on R studio software (4.0.3 version).
characteristics of ICP patients with SARS-CoV-2 infection are reported in Table 1. Twelve (50%) patients had a history of active cancer in the last ive years (see histological types Table 1). Three of the oncology patients also had a history of hematological malignancies (lymphoid malignancies). Eight patients had a history of hematological malignancies, consisting of lymphoproliferative disorders in 6 of the 8 cases (75%). Within 3 months previous to admission, 2 patients had undergone autologous stem-cell transplantation. Among the onco-hematological patients, 12 (50%) received chemotherapy in the previous 3 months. The four remaining patients were solid organ transplant recipients, who had received kidney (n = 3) and liver (n = 1) transplants. No HIV patients were hospitalized during this period. Interestingly, six patients (25%) displayed monoclonal gamma-pathology. The median time between the irst symptoms and ICU admission was 9 (6.2 -11) days.
The severity of the acute respiratory distress syndromes was illustrated by the presence of moderate to critical radiological in iltrates in most of the patients (n = 14, 82%), low PaO 2 /FiO 2 ratios, especially in the deceased group, high ICU prognostic scores and the need for invasive mechanical ventilation in all but 2 patients. Of note, non-invasive ventilation was frequently insuf icient to treat acute respiratory failure in our ICP patients. One patient successfully underwent ECMO for 32 days.
Eleven patients (79%) had a positive lupus anticoagulant. All but one received anticoagulant therapy, 11 (48%) of them receiving therapeutic anticoagulation. We observed venous thromboembolism events in 3 patients and one patient, with newly diagnosed promyelocytic leukemia, died from a carotid artery occlusion.
Healthcare-associated infections developed in 9 (37%) patients, mostly ventilator-associated pneumonia (78%). Only one patient had airway colonization by Aspergillus fumigatus, without proof for invasive infection. Interestingly 67% and 50% of the patients still had a positive SARS-CoV-2 PCR result from the SARS-CoV-2 swab sample at days 21 and 28 of ICU stay, respectively, suggesting prolonged viral carriage in ICP patients.
The median length of stay in the ICU was 9 days (IQR 7-26) for the patients who recovered and 4 days (IQR 4-12) for the deceased patients, respectively (p = 0.4). Figure 1 shows the survival curve of all SARS-CoV-2 infected patients, ICP or not. At day 28, the ICP patients had a signi icantly lower survival rate compared with other patients (41.7 % versus 27.3%, p = 0.021). The causes of death were respiratory failure (50%), sepsis (40%) and thrombosis (10%). No detectable factors were signi icantly different between SARS-CoV-2 ICP patients who recovered and those who did not, with the exception of classic scores (SAPSII, SOFA), which were higher in the deceased group (Table 1).

Discussion
Here we describe the global characteristics, severity and outcome of immunocompromised patients with severe SARS-CoV-2 infection admitted in two medical intensive care units. These patients represented 10% of the total number of patients admitted in our ICU during the irst wave of the SARS-CoV-2 outbreak, an equal proportion to what was described by the COVID-ICU Group [1]. The 28-day mortality rate in the intensive care unit observed for immunocompetent patients with SARS-CoV-2 infection was a bit lower to that reported in the literature for ICU patients with ARDS due to SARS-CoV-2 infection [1,6], in comparison with immunocompetent patients, the mortality of immunocompromised patients was signi icantly higher ( Figure 1). This was a striking difference from what is expected for immunocompromised patients in ICU, whose survival rate is nowadays comparable to that of immunocompetent patients, particularly in the context of respiratory infections, such as in luenza [7,8]. In their recent review, Fung, et al. [9] also point out this outstanding mortality of immunocompromised patients with SARS-CoV-2 infection, and being immunocompromised is associated with 90-day mortality in the COVID-ICU group publication [1].
In addition, patients with lymphoid malignancies, especially those with monoclonal gamma-pathy, seem to be particularly susceptible to SARS-CoV-2 infection, a particularity also highlighted by Malard, et al. [10]. This observation once again distinguishes the SARS-COV-2 infection since patients with lymphoid malignancies are not hematological patients usually admitted to ICU [8].
It is also important to notice that despite a clear severity of SARS-CoV-2 infection in immunocompromised patients, accompanied by high mortality in the early days, no ICP patient died beyond day 14 in ICU, highlighting a possible interest in prolonged full-code management for these patients in the ICU.
The clinical and biological characteristics of immunocompromised patients were comparable to the overall characteristics of all patients with SARS-CoV-2 infection [1,11]. The immunocompromised patients also developed ventilatoracquired pneumonia which negatively affects their prognosis. However, although at high theoretical risk, our patients did not undergo invasive fungal infections, in contrast with what has been reported during in luenza infections [8] or in other publications on SARS-CoV-2 [12]. Thromboses were observed in immunocompromised patients requiring the use of curative anticoagulation [13,14], which is dif icult to implement in these patients who are readily thrombocytopenic. In addition, no treatment (in particular no anti-viral therapy), showed any signi icant impact on the mortality rate, in accordance with recent publications [9,15].
One inal striking result is the prolonged viral excretion. Indeed, one month after the onset of symptoms, more than half of patients had still a detectable viral load in the respiratory luids, raising the issue of maintaining isolation measures.
Although the number of patients in our study is low, but in full accordance with the expected rate of admissions of such patients in our facility over a two-month period, we did not identify signi icant differences between immunocompromised and -competent groups except for mortality rates. This study nevertheless provides a detailed description of the characteristics of ICP patients, and the management challenges of these patients admitted to intensive care with severe SARS-CoV-2 infection, from a single center point of view.

Conclusion
The population of immunocompromised patients, especially patients with lymphoid malignancies like monoclonal gammopathy, seem to be those most affected by SARS-CoV-2 infection, with an unusually high mortality rate. Special attention should be paid to these patients for whom prolonged resuscitation seems required.

Declarations
Ethics approval statement: All patients received an information notice and the study was approved by the local institutional review board.

Consent for publication:
The authors accepted the manuscript for publication.

Authors' contributions:
All authors contributed to data collection. CD, HM, MB, MS, TA, TL, RCJ, YG and GM analyzed and interpreted the patient data. CD, TA, FM, VC, FS and GM were major contributors in writing the manuscript. All authors read and approved the inal manuscript.