Is Tocilizumab an effective therapy for Severe COVID19: a retrospective cohort study

To compare the outcome of severe COVID-19 patients treated with Tocilizumab (TCZ). Methods: A retrospective cohort study comparing the clinical characteristics and outcomes of patients who received TCZ with those who did not, was conducted at The Indus Hospital, Karachi. A sub-group analysis was conducted on the TCZ group to identify predictors of mortality. Results 88 patients including 41 patients in the TCZ group and 47 in non-TCZ group were recruited. Baseline characteristics were comparable. TCZ group patients presented with worse clinical features including median SpO2 82% vs 88%, p<0.05 and CRP 193 vs 133.9 mg/L, p<0.05. TCZ group showed severe bilateral chest x-ray ndings (92% vs 31%, p<0.05) compared to non-TCZ. In the TCZ group 85.4% were admitted in ICU compared to 69.8% in non-TCZ group, p>0.05. Mortality was not different among the groups (46% in TCZ group vs 51.1% in non-TCZ group, p>0.05). Median length of hospital stay, days of intubation, use of inotropic agents, use of invasive ventilation or in-hospital complications were not different between the groups. Sub-group analysis revealed that mortality within TCZ group was associated with high IL-6 levels (173 vs 69.66 pg/ml, p<0.05), ICU admission (100% vs 72%, p<0.05), need for mechanical ventilation (100% vs 13.6%, p<0.05) and higher incidence of in-hospital complications, p<0.05. TCZ group had more critical patients and TCZ failed to demonstrate any mortality benet in these patients. Non-survivors within the TCZ group were more critical compared to survivors and developed higher proportion of in hospital complications


Introduction
The coronavirus disease 2019 (COVID-19) originated from Wuhan, China in December 2019. It was declared a pandemic by World Health Organization (WHO) in February 2020 (1) and has infected more than 24,916,716 people till date leading to 841,372 deaths globally (2). Approximately 10% of cases develop a severe pneumonia and progress to acute respiratory distress syndrome (ARDS) (3). A mortality of 65% has been reported in critical COVID-19 cases (4). Similar to SARS and MERS infection, severe COVID-19 is characterized by both direct cytopathic and cytokine-mediated in ammatory damage, a phenomenon termed as "cytokine release syndrome" (CRS) (3,(5)(6)(7)(8). Lung biopsies of fatal COVID-19 cases demonstrate severe alveolar oedema, proteinaceous exudate and dense lymphocytic in ltration (9). Interleukin-6 (IL-6) is postulated to be a major player in the cytokine mediated damage (10). In ammatory markers including C-reactive protein (CRP), ferritin, lactate dehydrogenase (LDH) and D-dimer are also raised in CRS, which hence, bears a striking resemblance to the CRS seen in chimeric antigen receptor CAR T-cell therapy. Tocilizumab (TCZ) is an IL-6 receptor antagonist approved by FDA for CRS in patients with CAR T-cell therapy (11). In lieu of the biochemical and clinical similarity between the two cytokine release syndromes, clinicians in China rst attempted the use of TCZ in COVID-19 related CRS with elevated IL-6 levels (12). Published reports of experience with TCZ seem promising and encourage the use of TCZ in severe COVID-19 with elevated IL-6 levels (13,14). However, evidence yet derived seems to be weak as most lack comparison groups and could only analyze small samples. Moreover, the preliminary results from phase III, randomized, double-blind placebo controlled COVACTA trial failed to demonstrate any mortality bene t in patients treated with TCZ as compared to placebo at week 4 (15).
Pakistan recorded its rst case on 26th February, 2020 and The Indus Hospital Karachi, responded to the health emergency by establishing a dedicated 20 bedded COVID-19 facility on 19th March 2020 later expanding it to a 60 beds. It was led by a team of infectious disease experts, pulmonologists and anesthesiologists. Following success stories from other parts of the world, our experts also offered TCZ to severe COVID-19 patients. Later, TCZ was also incorporated in national COVID-19 management guidelines in China, Italy and Pakistan (12,16,17). The aim of this paper is to compare the outcome of severe and critical COVID-19 patients who received TCZ with those who did not. Predictors of mortality in the TCZ subgroup were also assessed.

Methods
This is a sub-study from a larger on going prospective study on all admitted con rmed (Nasopharyngeal PCR positive) COVID-19 patients under IRB approval number IRD_IRB_2020_04_002 at The Indus Hospital, Karachi. This is a retrospective cohort study on severe and critical COVID-19 patients older than 18 years admitted between 24th March and 19th June 2020. Patients were classi ed according to national guidelines (16) as "severe COVID-19" if on admission they have respiratory rate > 30/min, SpO2 < 90% on room air or Chest X ray showing > 50% of lung elds involved along with a positive nasopharyngeal PCR for COVID-19 and "Critical COVID-19" if along with the criteria for severe disease they manifested ARDS, Septic shock or multisystem organ failure (MSOF). Likewise, CRS in COVID-19 pneumonia was de ned as per national guidelines as presence of any of the following; Ferritin > 2000 ng/ml Or > 1000 ng/ml and rising for 24 hours; lymphocyte percentage < 20% and /or neutrophil to lymphocyte ratio (NLR)of > 5 along with ferritin > 700 ng/ml or LDH > 300 U/L or D-dimer > 1000 ng/ml or CRP > 70 mg/L. TCZ was administered to severe and critical COVID-19 patients who demonstrated CRS according to the mentioned criteria. Most patients received a single intravenous dose of TCZ of 4-8 mg/kg (maximum 800 mg) and some exhibiting severe CRS received 2 doses 12 hours apart. Factors affecting the use of TCZ were availability of the drug, presence of contra-indications (e.g. liver disease, kidney failure, presence of concomitant bacterial/ fungal infection) or physician's choice. Demographic information, clinical presentation, laboratory abnormalities including in ammatory markers, imaging results were recorded. Patient's hospital course including management, in-hospital complications, need for ventilation, length of stay and nal outcomes was also noted. The primary outcome was mortality in the two groups while we also compared length of stay, occurrence of in-hospital complications, use of inotropic support and use of mechanical ventilation. We performed a subgroup analysis based on mortality in the TCZ group. Data was recorded on a REDCap database and analyzed on SPSS v 24. Continuous variables were expressed as mean (SD) or median (IQR) as appropriate. Categorical variables were expressed as number (%). Student's t-test or Mann Whitney U test were used to compare continuous data. Chi square or Fischer's Exact test was used to compare categorical data. Length of hospital stay was compared using log-rank test and Kaplan Meir survival curves were made to show the survival function and p-value ≤ 0.05 was considered signi cant.

Characteristics of the study groups
There were 41 patients in the TCZ group compared with 47 patients in the Non-TCZ group with a mean age of 57 years (Table1). Most patients in TCZ group (73%) were critical at presentation compared to non-TCZ group (53%), p=0.053. Clinical features at presentation were similar in both groups except median oxygen saturation 82% in the TCZ group vs 88% in the Non-TCZ group, p=0.012. Laboratory parameters were also comparable in both groups except median Absolute lymphocyte count 1.02 x 10 9 in TCZ group as opposed to 0.75 x 10 9 in non-TCZ group, p<0.05. Baseline median CRP in the TCZ group was 193 vs 133.9 mg/L in non-TCZ group, p<0.05. 9/41 patients received 2 doses of TCZ while 32/41 received a single dose.
Most patients in TCZ group showed bilateral radiographic ndings including multi-lobar in ltrates (92% vs 31%) and consolidations (78% vs 31%) compared to non-TCZ group, p<0.05. Clinical severity scores for pneumonia like SOFA and CURB 65 scores were also comparable in the two groups except median MulBSTA score which was 9 in TCZ group while 11 in non-TCZ group, p<0.05. Most patients in the TCZ group received non-invasive ventilation within the Emergency Department 75.6 % vs 35.6 % in Non-TCZ group, p<0.05. There was no statistical difference in ICU admissions, 85.4% in TCZ vs 69.8% in Non-TCZ group, p>0.05. Regarding other compassionate therapies, there was no difference in steroid use and use of Hydroxychloroquine, while 31.7% received IVIG in the TCZ group compared to 10.6% in Non-TCZ group, p<0.05.

Comparison of outcomes
There was no difference in mortality; 46% in TCZ group vs 51.1% in non-TCZ group, p>0.05 ( Table 2). Length of ICU stay was 9 days in TCZ vs 8 days in Non-TCZ, p>0.05. No signi cant difference were reported in median length of hospital stay (10 days in TCZ group vs 8 days in Non-TCZ group days; p>0.05: Fig 1), median days of intubation (4.5 days in TCZ group vs 1 day in Non-TCZ group; p>0.05), use of ionotropic agents (43.9% in TCZ group vs 47.8% in Non-TCZ group; p>0.05) or use of invasive ventilation (53.7% in TCZ group vs 52.2% in Non-TCZ group; p>0.05). Cause of death was MSOF in 89.5% of patients in TCZ group vs 47.4% in Non-TCZ group, p>0.05. In hospital complications including nosocomial infections, were also not signi cantly different in the two groups.

Predictors of mortality in TCZ group:
Sub-group analysis for mortality within the TCZ group showed that D-dimer, LDH, Ferritin, pro-calcitonin and IL-6 were signi cantly elevated among those who died after TCZ therapy, p<0.05 (Table 3). There was no difference in median CRP level at baseline (167.8 in survivors vs 219.4 mg/L in non-survivors; p>0.05) and the median maximum change in CRP level (baseline value -lowest post therapy value) after TCZ therapy in the two groups (26.05 in survivors vs 13.2 mg/L in non-survivor; p>0.05). Those who died had elevation of LDH after TCZ therapy as compared to survivors where levels decreased after therapy, p=0.028. Baseline albumin level was signi cantly low in those who died compared to those who survived, p<0.05. Although there was no signi cant difference in disease category at presentation (severe/ critical), those who died had signi cantly higher mean SOFA scores (4.38 vs 5.39; p<0.05). Those who died were all admitted to the ICU (72.7% in survivors vs 100% in non-survivors; p<0.05). Need for invasive ventilation (13.6% in survivors vs 100% in nonsurvivors, p<0.05) median days of intubation (0 in survivors vs 7 in non-survivors, p<0.05) and use of inotropes (9.1% in survivors vs 84.2% in non-survivors, p<0.05) were higher in those who did not survive compared to those who survived. The survivors showed fewer in-hospital complications while those who died had more nosocomial infections, refractory shock and acute kidney injury (p<0.05). Repeat chest x-ray showed improvement in 95.5% of survivors compared to 52.2% of those who died (p<0.05). When compared the use of other compassionate therapies with TCZ like intravenous immunoglobulin (IVIG), Azithromycin (AZT), HCQ and methyl prednisolone, it was seen that all drugs except methyl prednisolone were associated with mortality (p<0.05). Therapeutic anticoagulation was also associated with mortality in our data (p<0.05).

Discussion
In this retrospective cohort study comparing outcomes of hospitalized severe and critical COVID-19 patients, there was no difference in mortality in those who were given TCZ therapy. The two groups in this analysis were largely analogous in their presentation and clinical features. Disease severity markers, laboratory parameters and demographic features such as age and gender were comparable at baseline. Our results corroborate the initial ndings of the phase III randomized, placebo controlled COVACTA trial which failed to demonstrate any bene t of TCZ and showed similar mortality and time to discharge by week four in the two groups (15) should be highlighted that many studies on the use of TCZ lack comparison groups, have small sample sizes, used multiple compassionate therapies and do not have data on confounders (13,(22)(23)(24)(25) .The time to intervention and disease severity varies and side effect reporting is variable and may be related to concurrent treatment too. In our data, those who received TCZ were signi cantly more critical with lower peripheral oxygen saturation, requiring higher non-invasive ventilation (NIV) support in Emergency room, worse radiological ndings and higher CRP levels but eventually showed no difference in mortality after treatment with TCZ.
TCZ causes profound immunosuppression, as Emily et al showed that nosocomial infections in critical COVID-19 patients were twice in TCZ group as compared to non-TCZ group (20). On the contrary, there was no difference in incidence of nosocomial infections between the TCZ group and Non-TCZ group in our study.
However, we saw 2 cases of candidaemia and 3 cases of gram negative multidrug resistant organisms (MDRO) in the TCZ group. Modest reversible raise in liver enzymes was seen in some patients but severe hepatic injury was not seen. Since Pakistan is endemic for Tuberculosis (TB), it is premature to comment on the risk of TB re-activation in survivors of COVID-19 treated with TCZ.
Use of other compassionate therapies like Intravenous Immunoglobulins (IVIG) was signi cantly higher in the TCZ group. However, the use of Azithromycin (AZT) and AZT in combination with HCQ was higher in Non-TCZ group. Since this was a desperate situation, physicians had a low threshold for administering any potentially bene cial therapy.
In the sub-group analysis, we demonstrated important mortality predictors within the TCZ group. Nonsurvivors had signi cantly higher serum levels of D-dimer, ferritin, IL-6 and pro-calcitonin and SOFA scores with lower serum albumin levels as compared to survivors. Hence, non-survivors were comparatively sicker on presentation, all requiring ICU admissions and mechanical ventilation. Wu et al found raised LDH and D-dimer to be risk factors for progression to ARDS which led to death in 200 patients in Wuhan(26). Sara et al showed that non-survivors have twice normal D-dimers, higher median SOFA scores and more frequently need respiratory, inotropic and renal support (27). High IL-6 levels secreted by pathogenic T cells and in ammatory monocytes are associated with impaired gas exchange, lung brosis, platelet aggregation and coagulopathy due to vascular endothelial injury and angiotensin II receptor microvascular dysfunction (28, 29). A metaanalysis reported mean IL-6 levels to be 2.9 times higher in patients with complicated COVID-19 disease compared to those with un complicated disease (30). Similarly, Wu et al reported higher IL-6 levels 6.05 in survivors vs 10.7 pg/ml among non-survivors, p < 0.05 (26). Hence, IL-6 can be used as a guide to initiating therapy with TCZ while also serving as a predictor of mortality in severe COVID-19.
The sub group analysis further demonstrated a sharp reduction in serum CRP levels post TCZ as the most marked immune-modulating effect, consistent with other studies (25, 28), with no difference in mortality. In contrast, serum LDH levels were seen to rise in non-survivors after TCZ and may be used as an indicator of therapy failure. Radiological improvement after 48 hours of TCZ therapy was observed in 95.5% of survivors as compared to 52.6% of non-survivors, while 36.8% had no change on chest X-ray and 10.6% showed deterioration. It is worthy to note that despite radiological improvement, non-survivors showed clinical deterioration. Xu et al reported resolution of radiographic lesions following TCZ therapy in 19/21 (90%) of his patients (28). Non-survivors fared worse compared to survivors with respect to in-hospital complications like refractory shock, Acute Kidney Injury (AKI), nosocomial infections and thromboembolism.
Our study is unique in being among the few comparative reports of TCZ therapy in severe and critical COVID-19 patients. Prospectively collected data was used for this research and captured most confounders which can affect the response to TCZ therapy.
Our study is limited by its sample size, but declining number of cases limit the feasibility of further augmenting the data, hence we decided not to wait. The study is observational in nature and selection of groups was largely dependent on availability of drug or physician choice which may have introduced bias.
However, most baseline characteristics in our study groups are well balanced at presentation minimizing bias. Further application of advanced statistical methods to control for confounders would be desirable but is limited by sample size. This is not a randomized clinical trial, however, TCZ appears to have a limited role in the management of patients with severe COVID-19 based on our results. There is a crucial need for more adequately designed randomized controlled trials on the role of TCZ in severe COVID-19.

Ethics approval and consent to participate
The study was approved by the Institutional review board (IRB approval number IRD_IRB_2020_04_002). The IRB has exempted taking consent from the participants because this was a prospective chart review of already collected data.

Not applicable
Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request

Competing interests
The Authors declare no Con ict of Interest.