Preserved C-reactive protein responses to blood stream infections following tocilizumab treatment for COVID-19.

Dear Editor,

In this Journal, Rossotti and colleagues provided early data on tocilizumab utility in COVID-19, later confirmed in randomised studies. Tocilizumab-mediated inhibition of IL-6 signalling can decrease CRP concentrations, potentially confounding the diagnosis of bacterial co-infections in COVID-19 that occur more frequently following longer hospital stays and admissions to the intensive care unit (ICU)3, 4, 5.

In inflammatory arthritides, serial tocilizumab dosing variably attenuates CRP responses following bacterial infections, but the effect following single-dose use in COVID-19 is not defined , . In a small COVID-19 cohort with blood stream infections (BSIs) that had received tocilizumab, CRP was reduced but remained detectable at the time of BSI diagnosis. However, CRP kinetics related to BSI were not assessed, and thus the utility of CRP to guide antibiotic prescribing in this context remains unknown , . We addressed this question by testing the hypothesis that a single dose of tocilizumab for COVID-19 retained CRP responses to bacterial infections, as modelled by BSIs.

We identified patients admitted to Royal Free Hospital (RFH) between 01/03/2020 and 01/02/2021, aged >18 years and diagnosed with COVID-19 by RT-PCR detection of SARS-CoV-2 from nasopharyngeal swabs. Tocilizumab use originated from routine clinical care delivery or randomised clinical trials after unblinding. COVID-19 associated BSIs were defined by isolation in blood cultures of any bacteria, excluding coagulase negative staphylococci, between 14 days prior to and 60 days after COVID-19 diagnosis. We excluded patients that developed BSIs prior to receiving tocilizumab. To assess dynamic CRP responses, we included only patients with blood parameter measurements performed at least 3 days prior to the onset of BSIs. Clinical, laboratory and drug data extraction, and statistical analyses were performed as previously described. The study was approved by the Research and Innovation Group at RFH, which stated that as this was a retrospective review of routine clinical data, formal ethics approval was not required.

Within the COVID-19 patients that met our inclusion criteria, 107 had received tocilizumab, 17 of whom then developed a BSI during their hospital admission (Table 1 ). A separate cohort of 55 COVID-19 patients developed a BSI but had not received tocilizumab (Table 1). Tocilizumab use preceding BSIs was more commonly associated with ICU admission, but the BSI organisms were comparable between the groups (Table 1). In the first week after tocilizumab administration we observed a rapid fall in CRP (Fig. 1 A), but not for total white cell, neutrophil or lymphocyte counts (Fig. 1A & fig S1). The CRP reduction following tocilizumab was short lived, with CRP concentrations rising within 21 days of tocilizumab receipt (Fig. 1A). To exclude confounding by bacterial co-infection, a sensitivity analysis on 90 patients that did not develop a BSI following tocilizumab also showed an early reduction followed by a rebound in CRP (fig S2A). A similar pattern was evident in patients that developed a BSI, although CRP concentrations showed less attenuation and greater heterogeneity within the 21-day period since tocilizumab administration (fig S2B).

Table 1

Baseline demographics and clinical characteristics for patients included in the study.

	No BSI & received tocilizumab(n = 90)	BSI & received tocilizumab(n = 17)	BSI & did not receive tocilizumab(n = 55)	p value*
Age, years, median (range)	63 (28–90)	61 (50–78)	63 (31–100)	p = 0.383
Gender, n (%)MaleFemale	56 (62)34 (38)	9 (53)8 (47)	38 (69)17 (31)	p = 0.253
Ethnicity, n (%)WhiteBlackAsianMixedOther	33 (42)7 (9)15 (19)2 (3)21 (27)	4 (31)3 (23)4 (31)1 (8)1 (8)	27 (55)7 (14)7 (14)0 (0)8 (16)	p = 0.101
Charlson co-morbidity score, n (%)0123+	57 (63)25 (28)6 (7)2 (2)	4 (24)11 (65)1 (6)1 (6)	27 (49)17 (31)7 (13)4 (7)	p = 0.095
ICU admission, n (%)YesNo	52 (58)38 (42)	17 (100)0 (0)	39 (72)15 (28)	p = 0.015
Corticosteroid use, n (%)YesNo	79 (89)10 (11)	15 (88)2 (12)	35 (65)19 (35)	p = 0.076
BSI organism, n (%)Gram-negative bacilliEnterococcus sp.Staphylococcus aureusOther	N/A	10 (40)2 (8)4 (16)1 (4)	32 (62)12 (23)7 (13)1 (2)	p = 0.507

BSI = blood stream infection.

relates to statistical comparison between COVID-19 patients who did or did not receive tocilizumab prior to the onset of BSI. Mann–Whitney test was used to compare age, Fisher's exact test was used to compare gender and microbiology results, and Chi-square test was used to compare ethnicity and Charlson co-morbidities.

Fig. 1

Inflammatory marker responses following tocilizumab administration and onset of BSI in COVID-19. (A) CRP concentration (left panel) and white cell count (right panel) relative to time since tocilizumab administration. All COVID-19 patients receiving tocilizumab included in analysis, independent of a subsequent presence of BSI (n = 107). (B) CRP concentration over time in patients that received tocilizumab and subsequently developed a BSI (n = 17). (C) CRP concentration of time in COVID-19 patients that developed a BSI but with no prior tocilizumab exposure (n = 55). (D) Change (Δ) in CRP induced by onset of BSI in COVID-19 patients stratified by prior receipt of tocilizumab. (E) Relationship between ΔCRP induced by BSI and time between administration of tocilizumab and the onset of BSI. Violin plots represent frequency distribution of all samples, with bold and dashed lines representing median and quartile values. All p values were calculated by Mann-Whitney tests.

To test the hypothesis that CRP would rise following a BSI independent of prior tocilizumab administration, we compared CRP responses in 17 patients that had received tocilizumab prior to a BSI with 55 patients who had not received tocilizumab. Strikingly, in both cohorts, BSIs resulted in clear CRP elevations (Figs. 1B & 1C). We calculated the change in CRP across the time of BSI onset to quantitatively compare this CRP rise. As blood samples were not collected daily in all patients, we derived paired sampling by calculating maximal CRP values 2 or 3 days prior to BSI-detecting blood culture collection and maximal CRP up to 2 days after BSI. This approach revealed an increase in CRP following BSI in 76.5% and 75.0% of patients that had or had not received tocilizumab respectively (Fig. 1D). Moreover, there was no difference in CRP increase between the groups (median CRP change +88 mg/L vs +76 mg/L respectively, p = 0.67 by Mann-Whitney test).

As patients developed BSIs at varying times following receipt of tocilizumab, we tested the hypothesis that BSI-induced CRP increment would be proportional to the time interval between tocilizumab administration and BSI onset. However, in the 17 patients that both received tocilizumab and subsequently developed a BSI, no relationship was observed between the length of the tocilizumab-BSI interval and the change in CRP (r = 0.1069, p = 0.6811 by Rank-spearman correlation) (Fig. 1E).

By inhibiting IL-6 signalling, tocilizumab may impact CRP-guided antibiotic prescribing decisions , . However, we demonstrate that prior administration of a single dose of tocilizumab does not attenuate CRP responses following a BSI, retaining the utility of this biomarker to diagnose bacterial co-infections associated with COVID-19. These findings have important implications for tocilizumab-treated COVID-19 patients: first, clinically-indicated antibiotic prescriptions are unlikely to be delayed, and second, low CRP levels alone are not an indication for continued prescription of unnecessary antibiotics, supporting stewardship efforts. Nevertheless, BSI onset did not initiate CRP elevations in all patients, irrespective of prior tocilizumab use, emphasising that CRP is only one contributor to diagnosing incipient bacterial infections.

Despite preserved CRP responses to BSI, tocilizumab transiently reduced baseline CRP levels, mostly recovering within 21 days. Furthermore, BSI-associated CRP increments were unrelated to time since tocilizumab, indicating that single tocilizumab dosing may not completely neutralise IL-6 responses, although a role for IL-6-independent CRP stimuli cannot be excluded. Measuring IL-6 signalling activity in vivo may predict attenuation of CRP responses and also inform the need for further tocilizumab dosing in COVID-19.

Our study was limited by its single-centre and retrospective nature, constraining patient numbers and negating correction for potential confounders. Nevertheless, increased frequency of corticosteroid use in tocilizumab recipients could have further attenuated CRP responses, counter to our observations. BSIs provided a standardised definition for bacterial infections, but limited extrapolation to non-BSI settings, an area of required future work to confirm the generalisability of our findings.

In conclusion, we show that tocilizumab use in severe COVID-19 preserves elevations in CRP concentration following the onset of a confirmed bacterial co-infection, as modelled by BSIs. Use of tocilizumab should not negate judicious, CRP-guided use of antibiotics in COVID-19.

Contributions

EQW, IB, SB and GP conceived the study. EQW, CB, AN, BOF, JP, ML, SY, SH, DM, MS and GP collected and analysed the data. EQW, SB and GP drafted the manuscript. All authors reviewed and approved the final version of the manuscript.

Declaration of Competing Interest

We declare that all authors have no conflicts of interest