Early Corticosteroids for Pneumocystis Pneumonia in Adults Without HIV Are Not Associated With Better Outcome

Accepted Manuscript Early corticosteroids for Pneumocystis pneumonia in adults without HIV are not associated with better outcome Patrick M. Wieruszewski, PharmD, Jason N. Barreto, PharmD, Erin Frazee, PharmD, MS, Craig E. Daniels, MD, Pritish K. Tosh, MD, Ross A. Dierkhising, MS, Kristin C. Mara, MS, Andrew H. Limper, MD PII:

S0012-3692(18)30648-2

DOI:

10.1016/j.chest.2018.04.026

Reference:

CHEST 1704

To appear in:

CHEST

Received Date: 6 November 2017 Revised Date:

7 March 2018

Accepted Date: 2 April 2018

Please cite this article as: Wieruszewski PM, Barreto JN, Frazee E, Daniels CE, Tosh PK, Dierkhising RA, Mara KC, Limper AH, Early corticosteroids for Pneumocystis pneumonia in adults without HIV are not associated with better outcome, CHEST (2018), doi: 10.1016/j.chest.2018.04.026. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Early corticosteroids for Pneumocystis pneumonia in adults without HIV are not associated with better outcome

RI PT

Patrick M. Wieruszewski, PharmD1 Jason N. Barreto, PharmD1 Erin Frazee, PharmD, MS1 Craig E. Daniels, MD2 Pritish K. Tosh, MD3 Ross A. Dierkhising, MS4 Kristin C. Mara, MS4 Andrew H. Limper, MD2 1

SC

Department of Pharmacy, Mayo Clinic, Rochester, MN Division of Pulmonary and Critical Care Medicine, Mayo Clinic, Rochester, MN 3 Division of Infectious Diseases, Mayo Clinic, Rochester, MN 4 Division of Biomedical Statistics and Informatics, Mayo Clinic, Rochester, MN Corresponding Author Prof. Andrew H. Limper, MD Mayo Clinic RO_GO_18_130PUL 200 First Street SW Rochester, MN 55905 [email protected]

M AN U

2

AC C

EP

Conflicts of Interest PMW: None. JNB: None. EF: None. CED: None. PKT: None. RAD: None. KCM: None. AHL: None.

TE D

Running Head: Corticosteroids in Non-HIV PcP

Funding/Support This work was funded in part by a research grant from the Mayo Clinic Department of Pharmacy. Prior Publication This article has been presented in abstract form at the American Thoracic Society International Conference, May 19-24, 2017, Washington, DC. Abbreviations

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

CI = confidence interval HIV = human immunodeficiency virus ICU = intensive care unit IQR = interquartile range LOS = length of stay MV = mechanical ventilation PcP = Pneumocystis jirovecii pneumonia PF = PaO2/FiO2 SF = SaO2/FiO2 SOFA = sequential organ failure assessment score SOFAresp = respiratory component of the sequential organ failure assessment score ∆SOFAresp = change in the respiratory component of the sequential organ failure assessment score from baseline

ACCEPTED MANUSCRIPT

ABSTRACT Background: Evidence supporting adjunctive corticosteroids during the treatment of Pneumocystis pneumonia (PcP) in adults without human immunodeficiency virus (HIV) is

RI PT

minimal and controversial.

Methods: This retrospective cohort study included PcP-positive, hospitalized patients without HIV admitted to Mayo Clinic from 2006 to 2016. Change from baseline in the respiratory

SC

component of the Sequential Organ Failure Assessment score (SOFAresp) at day 5 was compared between early (within 48 hours) steroid recipients and non-recipients utilizing multivariable

M AN U

logistic regression and in a propensity-matched analysis.

Results: Among the 323 included patients (early steroids, N = 258; no steroids, N = 65), the median (IQR) age was 65 (53, 73) years, 63% were male, and 92% were Caucasian. Severityadjusted regression and propensity-matched analyses demonstrated early steroids were

TE D

associated with less improvement in SOFAresp at day 5 compared to no steroids, p = 0.001 and p = 0.017, respectively. No differences were observed in the odds of having a ≥ one-point improvement in SOFAresp at day 5 compared to baseline between groups (adjusted OR 0.76, 95%

EP

CI 0.24, 2.28; p = 0.61). Overall 30-day mortality was 22.9% (95% CI 18.2%, 27.4%). No differences in mortality, length of stay, admission to the ICU, or need for mechanical ventilation

AC C

were found between early steroid recipients and non-recipients. Conclusions: The addition of early corticosteroids to anti-Pneumocystis therapy in patients without HIV was not associated with improved respiratory outcomes. Key words: pneumocystis; steroids; immunocompromised host; respiratory failure

ACCEPTED MANUSCRIPT

MAIN TEXT Introduction Pneumocystis jirovecii pneumonia (PcP) is a life-threatening condition historically affecting

RI PT

patients with human immunodeficiency virus (HIV), and other immune compromised patients. Recent epidemiologic data indicate that the incidence and mortality of PcP among the

immunocompromised population without HIV is rising.1-4 Patients without HIV exhibit a higher

SC

mortality rate relative to their HIV-positive counterparts, likely secondary to greater lung

inflammation with more robust neutrophil recruitment, as well as non-modifiable risks from

M AN U

antecedent disease.5-7 This pathophysiologic difference may predispose patients without HIV to manifest intensified inflammatory response to PcP and consequently more severe hypoxemia, a strong predictor of mortality in this disease.8-11

Corticosteroids are hypothesized to blunt the inflammatory response precipitated by anti-

TE D

Pneumocystis therapy and prevent respiratory decompensation.5-7 Early, limited scope, prospective studies in hypoxemic HIV patients with PcP found that early initiation of corticosteroids reduced the need for mechanical ventilation and conferred the greatest mortality

EP

benefit.12,13 Unfortunately, the data for adjunctive corticosteroids among the immunosuppressed population without HIV is far less compelling and one study even indicated potential harm with

AC C

steroid use.14 Furthermore, the majority of the published data in patients without HIV is derived from small studies which predate contemporary standards of care utilized for refractory hypoxemia, and provide insufficient detail to evaluate their generalizability.14-18 Given the limited and controversial existing evidence, the purpose of this study was to

evaluate the impact of early corticosteroid administration as an adjunct to anti-Pneumocystis therapy during the treatment of PcP in acutely ill immunocompromised adults without HIV.

ACCEPTED MANUSCRIPT

Methods Setting and Participants This retrospective cohort study included hospitalized adult patients (≥18 years of age) with a

RI PT

diagnosis of PcP at Mayo Clinic, Rochester, MN between January 1, 2006 and August 30, 2016. Patients were identified by a primary or secondary diagnosis code consistent with PcP (ICD-9CM Code 136.3 or ICD-10-CM Code B59) and the diagnosis was confirmed by a positive single-

SC

copy Pneumocystis polymerase chain reaction assay or smear from a respiratory specimen.5,19 Patients were excluded if they had documented HIV infection or a history of HIV seropositivity,

M AN U

did not receive anti-Pneumocystis therapy, died within 48 hours of diagnosis, were pregnant, incarcerated, or refused Minnesota research authorization. This study was conducted in accordance with the amended Declaration of Helsinki and it was approved by the Mayo Clinic Institutional Review Board (#16-006232) with a waiver of informed consent.

TE D

At our institution, sulfamethoxazole/trimethoprim is the preferred anti-Pneumocystis therapy and prescription of corticosteroids as an adjunct for PcP is at the discretion of the care team. Mechanical ventilatory care for hypoxemic respiratory failure is protocolized at our

EP

institution to ensure administration of a low tidal volume, lung protective strategy with optimal positive end expiratory pressure, targeting plateau pressures ≤ 30 cm H2O.20,21

AC C

Measures

Abstracted data from electronic medical records included patient demographics, comorbid conditions and prior immunosuppressive regimens (including steroids, chemotoxic agents, and biologics), arterial gas and oximetry data, anti-Pneumocystis treatment regimen, and hospital admission characteristics. All corticosteroid use during the hospital stay was collected including details on agent, daily and total dose, and duration of therapy. Doses were expressed in

ACCEPTED MANUSCRIPT

prednisone equivalents.22 Early corticosteroid use was defined as any corticosteroid administration within 48 hours of institution of anti-Pneumocystis therapy or PcP diagnosis, whichever occurred first. Respiratory function was defined based on the respiratory component

RI PT

of the Sequential Organ Failure Assessment score (SOFAresp). This score considers the

PaO2/FiO2 (PF) ratio or the SaO2/FiO2 (SF) ratio when arterial blood gas data is unavailable. Higher SOFAresp scores represent worse respiratory function (e-Table 1).23,24 The change in

SC

respiratory status at day 5 (∆SOFAresp) was calculated as (baseline SOFAresp – SOFAresp at day 5 or hospital discharge if sooner), where a positive value reflects a clinical improvement.

M AN U

Statistical Analysis

Continuous variables were summarized by medians with interquartile ranges (IQR) and counts and percentages were used for categorical data. For the primary outcomes of ∆SOFAresp at day 5 and day 7 and the secondary outcome of length of stay we fit multivariable linear regression

TE D

models that included terms for early corticosteroid use and baseline severity of hypoxemia according to SOFAresp. Given SOFAresp is not typically considered on a continuum, we also performed a logistic regression analysis where the outcome was defined as at least a one-point

EP

improvement in the ∆SOFAresp at day 5 and day 7. Multivariable Cox proportional hazards models including terms for early corticosteroid use and baseline SOFAresp and the Kaplan-Meier

AC C

method were used for the secondary outcomes of 30-day all-cause mortality, the need for mechanical ventilation, and admission to the ICU. Given corticosteroids are typically prescribed in patients with more severe respiratory

dysfunction, a propensity-matched analysis was also performed based on the probability of receiving steroids within 48 hours to account for these inherent differences in a nonrandomized study.25 This probability was estimated for each patient using logistic regression from relevant,

ACCEPTED MANUSCRIPT

available baseline variables (e-Figure 1). Individuals who received corticosteroids within 48hours were then matched in a two-to-one fashion to non-recipients based on the logit of the propensity score using a caliper of width 0.2 of the standard deviation of the logit of the

RI PT

propensity score.26 The success of the match was characterized according to standardized

differences in baseline characteristics (Table 1, e-Figure 1). This new propensity-matched dataset was then used to re-analyze the aforementioned outcomes using stratified linear, logistic, or Cox

SC

regression models, as appropriate for the outcome, where the matched sets formed the strata. All p-values ≤ 0.05 were considered to be statistically significant. Analyses were

M AN U

performed using SAS version 9.4 (SAS Institute Inc., Cary, NC) and R version 3.2.0 (R Core Team, R Foundation for Statistical Computing, Vienna, Austria, 2015). Results Study Population and Treatment

TE D

Of the 423 patients with a diagnosis of PcP during the study timeframe, 323 patients met eligibility criteria and were included in the study cohort (Figure 1). Two hundred fifty eight (80%) patients received corticosteroids within 48-hours of the index date (early steroid group),

EP

whereas 65 (20%) did not (no early steroid group). Baseline characteristics are shown in Table 1. The majority of included patients had a diagnosis of an immunosuppressive disease at the time of

AC C

PcP (N = 313; 97%), the most common of which was a hematologic malignancy (N = 162; 50%). There were 193 (60%) patients with a history of immunosuppressive medication use prior to hospital admission, and 78 (24%) were receiving more than one agent. The most common immunosuppressants were corticosteroids (N = 174; 54%) followed by cytotoxic agents (N = 73; 23%). A relatively small proportion of patients had a baseline chronic airway disease (N = 67; 21%).

ACCEPTED MANUSCRIPT

All patients received appropriate anti-Pneumocystis therapy, primarily with sulfamethoxazole/trimethoprim (N = 302; 93%). When used early (within 48 hours), the median (IQR) time to steroid initiation and duration of therapy was 6 (-13, 25) hours and 7 (5, 12) days,

RI PT

respectively (Table 2). Respiratory Outcomes

At baseline, the median (IQR) SOFAresp was 2 (1, 3), which corresponded to PF (N = 181) and

SC

SF (N = 142) ratios of 145 (95, 220) and 387 (307, 424), respectively. Patients in the steroid group had worse baseline respiratory function as evidenced by lower PF and SF ratios, which

M AN U

corresponded to higher SOFAresp scores (Table 1).

The median (IQR) day 5 ∆SOFAresp was 0 (0, 1). One hundred thirty one (41%) patients experienced at least a one-point improvement in day 5 ∆SOFAresp, whereas 130 (40%) patients exhibited no change, and 62 patients (19%) worsened by at least one-point (e-Figure 2). When

TE D

stratified by group, most patients in both the early steroid and no early steroid groups demonstrated no change in ∆SOFAresp at day 5 (Figure 2). After adjustment for baseline severity of hypoxemia defined by SOFAresp, the mean

EP

difference between the early steroid and no early steroid (reference) groups in ∆SOFAresp at day 5 was -0.46 (95% CI, -0.73, -0.21; p = 0.001). This suggested that the no early steroid group

AC C

exhibited a significantly greater improvement in respiratory function at day 5 compared to early steroid patients. In contrast, the odds of at least a one-point improvement in ∆SOFAresp at day 5 did not differ between the groups (OR 0.75, no early steroids as reference group, 95% CI 0.24, 2.28; p = 0.61). These results were no different when subgroups of patients were used taking in consideration baseline steroid use at the time of PcP (e-Table 2, e-Table 3). Additionally, these results were similar when separating out patients who received steroids within 48 hours of

ACCEPTED MANUSCRIPT

presentation and 48 hours of initiation of anti-Pneumocystis therapy (e-Table 2). Furthermore, a comparison of change in SF ratios at day 5 from baseline revealed less improvement in the early steroid group (p = 0.038, e-Table 4). Day 7 analyses in the 169 patients who remained

RI PT

hospitalized showed no difference between the groups (Table 3, e-Table 4).

Secondary Outcomes

SC

When adjusting for baseline severity of hypoxemia, the patients who received early steroids were not significantly more likely than the no early steroid patients to die within 30 days (p = 0.091,

M AN U

Table 4). These results were no different when subgroups of patients were used taking in consideration baseline steroid use at the time of PcP (e-Figure 3). While those who received early steroids were observed to be more likely to be admitted to the ICU, require mechanical ventilation, and stay in the hospital longer, statistically only admission to the ICU was

TE D

significantly greater in the early steroid group (HR 3.02, 95% CI 1.05, 8.65; p = 0.049, Table 4).

Propensity Matched Cohort

EP

According to the propensity score, 162 patients (108 early steroid, and 54 no early steroid) were successfully matched. The variables utilized for matching patients in the propensity cohort are

AC C

listed in e-Figure 1. Groups were well balanced after matching as shown in Table 1 and e-Figure 1. Specifically, in the propensity matched cohort SF, PF, and SOFAresp were similar between early steroid and no early steroid groups. Analyses revealed similar findings to the unmatched cohort for the mean difference in ∆SOFAresp at day 5 and odds of at least a one-point improvement in SOFAresp at day 5 between groups (Table 3). Additionally, the change in SF at day 5 from baseline was not different between the early steroid group and the no early steroid

ACCEPTED MANUSCRIPT

group (p = 0.25, e-Table 3). No statistically significant differences were noted in any of the secondary endpoints between the two groups in the propensity matched analysis (Table 4, Figure 3).

RI PT

Discussion

This retrospective cohort analysis of PcP in the non-HIV population did not demonstrate benefit associated with early corticosteroid use on respiratory function at day 5 compared to baseline.

SC

After adjustment for baseline severity of hypoxemia, no significant differences in 30-day allcause mortality were found between the early steroid group and no early steroid group.

M AN U

Propensity-matched analyses revealed no significant differences in mortality, length of stay, admission to the ICU, or the need for mechanical ventilation between the early steroid group and the no early steroid group.

While the use of adjunctive steroids for HIV patients diagnosed with PcP has shown a

TE D

significant mortality benefit, the 1990 consensus statement, which recommended their use in this population acknowledged the absence of evidentiary support for their recommendation to extrapolate this data also to patients without HIV.12 Variability of host immune response to

EP

infection and destruction of Pneumocystis suggests that these two populations may have distinct responses to the intervention.6,11 Since the 1990 publication, despite the increasing incidence of

AC C

non-HIV associated PcP, few studies have captured the impact that adjunctive steroids have on clinical outcomes. An earlier investigation reported on a group of thirty non-HIV patients, retrospectively analyzed, and demonstrated that steroids were associated with shortened time on the ventilator and ICU length of stay.18 Two subsequent retrospective analyses, each with less than 100 patients, revealed no associations between steroid use and clinical outcomes.15,17 Interestingly, a third study demonstrated worse ICU mortality in patients who received

ACCEPTED MANUSCRIPT

corticosteroids, which was not attributable to ICU-acquired infection.14 Further, a systematic review and meta-analysis of these small studies failed to demonstrate an overall survival benefit with the use of steroids.27

RI PT

Given the consensus recommendation, there may be an inherent provider bias toward prescription of steroids in more severely ill patients with PcP regardless of HIV status. Indeed this is supported by the baseline differences observed in our study in PF and SF ratios, and

SC

corresponding SOFAresp scores between the early steroid and no early steroid groups. The small sample sizes of the published studies in the non-HIV population have prohibited a rigorous

M AN U

attempt at adjustment for this and other underlying differences between groups. As such, it is difficult to interpret whether their results are indicative of treatment effect or just a baseline predisposition toward worse outcomes. To our knowledge, our study is the largest to date, including more non-HIV-associated PcP cases than all prior studies combined. Our multivariable

TE D

models adjusted for baseline severity of hypoxemia, as well as analyses in our propensitymatched cohort, confirmed the previous studies, which found no observable benefit with early corticosteroid administration. It is unclear why the association between steroids and outcomes

EP

differs according to baseline HIV status. We speculate however that these differences may be due to reduced response among those with prior exposure to corticosteroids immediately

AC C

preceding PcP presentation, which may be a consideration for future studies. Although the adverse consequences of using steroids are generally known, specific safety

data in the non-HIV population infected with PcP are lacking. Whereas a previous study showed worse mortality when corticosteroids were used in this population, we found no difference in the present analysis. Future research should also consider other safety endpoints associated with steroid use including hyperglycemia, ICU-acquired weakness, delirium, and the risk for

ACCEPTED MANUSCRIPT

coinfections. Insight into these risks could aid clinical decision making and the risk/benefit evaluation in this unique population. The present study has several limitations. As previously noted, there is an inherent bias

RI PT

toward use of steroids for more severe cases in clinical practice. Given the limited evidence among individuals without HIV, we suspect this predilection for use either reflects a belief in the putative benefits of corticosteroids or a lack of knowledge about the limited external validity of

SC

the previous studies. To address this limitation, multiple techniques were used to adjust for baseline severity of hypoxemia both through direct regression adjustment and propensity

M AN U

matching. No benefits of steroids were found despite these efforts, though we cannot exclude the possibility that other unmeasured variables associated with steroid use may have been unintentionally omitted from the models.25 The uniquely immunosuppressed population studied is highly susceptible to limitations of care such as “do not intubate” or “do not resuscitate”

TE D

orders, which may potentially bias the results. Because acquisition of these variables retrospectively was not possible, we excluded patients who expired within 48 hours of diagnosis. These efforts decrease, but cannot completely exclude, the likelihood that limitations of care may

EP

have negatively contributed to the outcomes of this study. Also, in routine clinical practice there is variability in the timing, dose, and duration of steroids administered. To standardize the

AC C

definition of early steroid use for analyses, we selected a cut off of 48 hours from the index date. During the course of PcP, initiation of anti-Pneumocystis therapy triggers lysis of Pneumocystis pathogens in the lungs, which is believed to evoke a profound pro-inflammatory response through exposure of surface β-glucans on the organism.5,12 This temporal relationship of immune and inflammatory response was substantiated by the original corticosteroid studies in the HIV population, where early steroids, within 24–72 hours, were found to have the greatest mortality

ACCEPTED MANUSCRIPT

benefit. Although 20 patients in the present study received steroids after 48 hours, the median (IQR) time to initiation among these was 73 (68, 107) hours, likely beyond the window of opportunity theorized. Furthermore, when separating out patients who received steroids within

RI PT

48 hours of presentation and 48 hours of initiation of anti-Pneumocystis therapy, there was no change in respiratory outcome. While the doses of steroid used in our study varied, the relatively narrow interquartile range suggests the variability may have contributed only in a limited manner

regimen has on the course of these patients.

SC

to the study. Future studies should continue to explore the impact that manipulation of the steroid

M AN U

Finally, we chose to utilize the respiratory component of the SOFA score as a standard criterion to assess change in respiratory function. Historical definitions of hypoxemic respiratory failure in PcP studies are inconsistent. These often utilize the PaO2, without regard for the amount of oxygen administered at the time of measurement, or the alveolar-arterial gradient.28 In

TE D

retrospective studies such as the present, this would limit the study to patients with arterial lines in place at baseline. Furthermore, invasive monitoring is discontinued when no longer needed. Therefore some patients may not have an arterial line at day 5, which makes use of previous

EP

definitions difficult for assessment of trajectory over time. The SOFA score was originally developed to utilize the PF ratio; however, Pandparihande et al. established and validated SF

AC C

ranges for SOFA score assignment when arterial blood gas data is not available,24 which allowed for a standardized definition of respiratory function in this study. Additional analyses to evaluate the robustness of this hypoxemic respiratory failure definition included the need for intubation and SF ratio as a continuous variable, neither of which yielded differences between the early steroid group and the no early steroid group in this study.

ACCEPTED MANUSCRIPT

Conclusion In this propensity matched cohort study, the administration of adjunctive corticosteroids within 48 hours of anti-Pneumocystis therapy did not provide any benefit in short-term

RI PT

respiratory outcomes, need for ICU admission or intubation, survival, or hospital length of stay among patients without HIV treated for PcP. Our findings should prompt clinicians to re-

evaluate routine administration of steroids as adjunctive therapy for PcP among patients without

AC C

EP

TE D

M AN U

studies are warranted to further explore these findings.

SC

HIV in the absence of other compelling indications for receiving steroids. Robust prospective

ACCEPTED MANUSCRIPT

Acknowledgements

RI PT

Guarantor statement AHL had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the analysis, and had final responsibility for the decision to submit for publication.

M AN U

Financial/nonfinancial disclosures PMW: No financial or personal interests to disclose. JNB: No financial or personal interests to disclose. EF: No financial or personal interests to disclose. CED: No financial or personal interests to disclose. PKT: No financial or personal interests to disclose. RAD: No financial or personal interests to disclose. KCM: No financial or personal interests to disclose. AHL: No financial or personal interests to disclose.

SC

Author Contributions PMW, JNB, EF, and AHL contributed substantially to the conception and design of the study. All authors contributed substantially to analyses and interpretation of results, and drafting of the manuscript for important intellectual content.

AC C

EP

TE D

Role of the sponsors The funding source had no role in the design or conduct of the study; data collection, management, analysis, or interpretation; writing the manuscript, or the decision to submit the manuscript for publication.

ACCEPTED MANUSCRIPT

References

6. 7.

8.

9.

10.

11.

12.

13.

14. 15.

16.

RI PT

SC

5.

M AN U

4.

TE D

3.

EP

2.

Coyle PV, McCaughey C, Nager A, et al. Rising incidence of Pneumocystis jirovecii pneumonia suggests iatrogenic exposure of immune-compromised patients may be becoming a significant problem. J Med Microbiol. 2012;61(Pt 7):1009-1015. Kofteridis DP, Valachis A, Velegraki M, et al. Predisposing factors, clinical characteristics and outcome of Pneumonocystis jirovecii pneumonia in HIV-negative patients. J Infect Chemother. 2014;20(7):412-416. Sepkowitz KA. Opportunistic infections in patients with and patients without Acquired Immunodeficiency Syndrome. Clin Infect Dis. 2002;34(8):1098-1107. Tasaka S, Tokuda H. Pneumocystis jirovecii pneumonia in non-HIV-infected patients in the era of novel immunosuppressive therapies. J Infect Chemother. 2012;18(6):793-806. Carmona EM, Limper AH. Update on the diagnosis and treatment of Pneumocystis pneumonia. Ther Adv Respir Dis. 2011;5(1):41-59. Hahn PY, Limper AH. The role of inflammation in respiratory impairment during Pneumocystis carinii pneumonia. Semin Respir Infect. 2003;18(1):40-47. Yale SH, Limper AH. Pneumocystis carinii pneumonia in patients without acquired immunodeficiency syndrome: associated illness and prior corticosteroid therapy. Mayo Clin Proc. 1996;71(1):5-13. Arend SM, Kroon FP, van't Wout JW. Pneumocystis carinii pneumonia in patients without AIDS, 1980 through 1993. An analysis of 78 cases. Arch Intern Med. 1995;155(22):2436-2441. Curtis JR, Greenberg DL, Hudson LD, Fisher LD, Krone MR, Collier AC. Changing use of intensive care for HIV-infected patients with Pneumocystis carinii pneumonia. Am J Respir Crit Care Med. 1994;150(5 Pt 1):1305-1310. Festic E, Gajic O, Limper AH, Aksamit TR. Acute respiratory failure due to pneumocystis pneumonia in patients without human immunodeficiency virus infection: outcome and associated features. Chest. 2005;128(2):573-579. Zahar JR, Robin M, Azoulay E, Fieux F, Nitenberg G, Schlemmer B. Pneumocystis carinii pneumonia in critically ill patients with malignancy: a descriptive study. Clin Infect Dis. 2002;35(8):929-934. Consensus statement on the use of corticosteroids as adjunctive therapy for pneumocystis pneumonia in the acquired immunodeficiency syndrome. The National Institutes of Health-University of California Expert Panel for Corticosteroids as Adjunctive Therapy for Pneumocystis Pneumonia. N Engl J Med. 1990;323(21):1500-1504. Ewald H, Raatz H, Boscacci R, Furrer H, Bucher HC, Briel M. Adjunctive corticosteroids for Pneumocystis jiroveci pneumonia in patients with HIV infection. Cochrane Database Syst Rev. 2015(4):Cd006150. Lemiale V, Debrumetz A, Delannoy A, Alberti C, Azoulay E. Adjunctive steroid in HIVnegative patients with severe Pneumocystis pneumonia. Respir Res. 2013;14:87. Delclaux C, Zahar JR, Amraoui G, et al. Corticosteroids as adjunctive therapy for severe Pneumocystis carinii pneumonia in non-human immunodeficiency virus-infected patients: retrospective study of 31 patients. Clin Infect Dis. 1999;29(3):670-672. Force ADT, Ranieri VM, Rubenfeld GD, et al. Acute respiratory distress syndrome: the Berlin Definition. JAMA. 2012;307(23):2526-2533.

AC C

1.

ACCEPTED MANUSCRIPT

22.

23. 24.

25.

26.

27.

28.

RI PT

SC

21.

M AN U

20.

TE D

19.

EP

18.

Moon SM, Kim T, Sung H, et al. Outcomes of moderate-to-severe Pneumocystis pneumonia treated with adjunctive steroid in non-HIV-infected patients. Antimicrob Agents Chemother. 2011;55(10):4613-4618. Pareja JG, Garland R, Koziel H. Use of adjunctive corticosteroids in severe adult nonHIV Pneumocystis carinii pneumonia. Chest. 1998;113(5):1215-1224. Wilson JW, Limper AH, Grys TE, Karre T, Wengenack NL, Binnicker MJ. Pneumocystis jirovecii testing by real-time polymerase chain reaction and direct examination among immunocompetent and immunosuppressed patient groups and correlation to disease specificity. Diagn Microbiol Infect Dis. 2011;69(2):145-152. Acute Respiratory Distress Syndrome N, Brower RG, Matthay MA, et al. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med. 2000;342(18):1301-1308. Amato MB, Meade MO, Slutsky AS, et al. Driving pressure and survival in the acute respiratory distress syndrome. N Engl J Med. 2015;372(8):747-755. Mager DE, Lin SX, Blum RA, Lates CD, Jusko WJ. Dose equivalency evaluation of major corticosteroids: pharmacokinetics and cell trafficking and cortisol dynamics. J Clin Pharmacol. 2003;43(11):1216-1227. Ferreira FL, Bota DP, Bross A, Melot C, Vincent JL. Serial evaluation of the SOFA score to predict outcome in critically ill patients. JAMA. 2001;286(14):1754-1758. Pandharipande PP, Shintani AK, Hagerman HE, et al. Derivation and validation of Spo2/Fio2 ratio to impute for Pao2/Fio2 ratio in the respiratory component of the Sequential Organ Failure Assessment score. Crit Care Med. 2009;37(4):1317-1321. Brookhart MA, Wyss R, Layton JB, Sturmer T. Propensity score methods for confounding control in nonexperimental research. Circ Cardiovasc Qual Outcomes. 2013;6(5):604-611. Austin PC. Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies. Pharm Stat. 2011;10(2):150-161. Injean P, Eells SJ, Wu H, McElroy I, Gregson AL, McKinnell JA. A systematic review and meta-analysis of the data behind current recommendations for corticosteroids in nonHIV related PCP: Knowing when you are on shaky foundations. Transplant Direct. 2017;3(3):e137. Burt CC, Arrowsmith JE. Respiratory failure. Surgery (Oxford). 2009;27(11):475-479.

AC C

17.

ACCEPTED MANUSCRIPT

RI PT

Tables

Table 1. Baseline patient characteristics and demographics

Propensity-Matched Cohort Early No Early Steroids Steroids Standardized (N = 108) (N = 54) Difference 67 (59, 76) 65 (54, 71) 0.141 68 (63) 36 (67) -0.078 100 (93) 49 (91) -26 (23, 30) 25 (23, 30) 0.033

123 (48) 22 (9) 83 (32) 52 (20) 16 (6)

20 (6) 54 (17) 54 (17) 1 (0.3) 193 (60) 174 (54) 68 (21) 14 (4) 24 (7) 8 (3)

18 (7) 44 (17) 44 (17) 0 (0) 163 (63) 152 (59) 55 (21) 10 (4) 20 (8) 5 (2)

39 (60) 4 (6) 17 (26) 9 (14) 4 (6)

-0.250 0.093 0.121 0.172 0.004

58 (54) 6 (6) 32 (30) 15 (14) 8 (7)

31 (57) 4 (7) 15 (28) 9 (17) 3 (6)

-0.075 -0.075 0.041 -0.077 0.075

2 (3) 10 (15) 10 (15) 1 (2) 30 (46) 22 (34) 13 (20) 4 (6) 4 (6) 3 (5)

0.181 0.049 0.049 -0.177 0.341 -----0.150

4 (4) 17 (16) 17 (16) 0 (0) 57 (53) 53 (49) 19 (18) 4 (4) 10 (9) 2 (2)

2 (4) 8 (15) 8 (15) 0 (0) 26 (48) 18 (33) 10 (19) 4 (7) 3 (6) 1 (2)

0 0.026 0.026 0 0.093 ----0

EP

TE D

162 (50) 26 (8) 100 (31) 61 (19) 20 (6)

M AN U

SC

Unadjusted Cohort Early No Early Steroids Steroids Standardized (N = 258) (N = 65) Difference 65 (52, 73) 65 (55, 72) -0.134 157 (61) 45 (69) -0.167 238 (92) 59 (91) -27 (24, 31) 26 (23, 30) 0.081

AC C

Characteristica Age (y) Male, n (%) Caucasian, n (%) BMI (kg/m2) Immunosuppressive disease, n (%) Hematologic malignancy Bone marrow transplantation Inflammatory of connective tissue disease Solid organ tumor Solid organ transplantation Airway disease, n (%) Asthma COPD Cystic fibrosis Other Chronic immunosuppressive regimen, n (%) Steroid Chemotoxic Biologic Other PcP prophylaxis, n (%) Baseline respiratory status SOFAresp

Total Cohort (N = 323) 65 (53, 73) 202 (63) 297 (92) 26 (24, 31)

2 (1, 3) 2 (2, 3) 1 (1, 2) 0.706 2 (1, 3) 2 (1, 2) 0.047 145 143 212 213 200 PF ratio (N = 181) --(95, 220) (94, 214) (104, 254) (137, 275) (100, 233) 387 338 414 381 414 SF ratio (N = 142) --(307, 424) (291, 419) (329, 438) (314, 419) (321, 438) a Median (IQR) unless otherwise specifiedBMI = body mass index, COPD = chronic obstructive pulmonary disease, PcP = Pneumocystis pneumonia, PF = PaO2/FiO2, SF = SaO2/FiO2, SOFAresp = respiratory component of the sequential organ failure assessment score

ACCEPTED MANUSCRIPT

Table 2. Treatment characteristics No Early Steroids (N = 65)c 59 (91) 73 (68, 107)c 60 (40, 80)c 60 (40, 68)c 4 (1, 10)c

RI PT

Early Steroids (N = 258) 247 (96) 6 (-13, 25) 60 (40, 100) 60 (42, 79) 7 (5, 12)

AC C

EP

TE D

M AN U

SC

Characteristica Sulfamethoxazole/trimethoprim, n (%) Time to steroid initiation (h) Starting steroid dose (mg)b Daily steroid dose (mg)b Duration of steroids (d) a Median (IQR) unless otherwise specified b Prednisone equivalents c Steroid recipients after 48 hours, N = 20

ACCEPTED MANUSCRIPT

Table 3. Analysis of respiratory function in the early steroid vs no early steroid group Adjusted for Baseline SOFA Propensity Analysis Effect size Effect size N (95% CI) p-value N (95% CI) p-value ∆SOFAresp Mean changea -0.47 (-0.73, -0.21) Day 7 169 -0.31 (-0.72, 0.10) ≥ one point improvementb

0.001

162

0.14

57

323

-0.46 (-0.84, -0.08) -0.15 (-0.74, 0.44)

0.017 0.61

0.75 0.61 162 0.78 0.48 (0.24, 2.28) (0.39, 1.55) Day 7 169 0.62 0.41 57 1.00 1.00 (0.20, 1.94) (0.33, 3.05) a Effect size is the average difference (“no early steroids” as the reference group) wherein a positive value favors improvement in the early steroid group and a negative value favors improvement in the no early steroid group b Effect size is the odds ratio (“no early steroids” as the reference group) wherein a value > one favors an improvement in the early steroid group CI = confidence interval, SOFAresp = respiratory component of the Sequential Organ Failure Assessment score

AC C

EP

TE D

M AN U

SC

Day 5

323

RI PT

Day 5

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Table 4. Clinical outcomes according to early vs no early steroid use Adjusted for Baseline SOFA Propensity Analysis Effect size Effect size Variable N (95% CI) p-value N (95% CI) p-value 30-day all-cause 2.08 0.091 162 2.56 0.06 323 a mortality (0.89, 4.87) (0.96, 6.81) Need for MVa 250 5.47 0.098 156 5.42 0.11 (0.73, 41.0) (0.69, 42.58) ICU admissiona 175 3.02 0.049 135 1.75 0.32 (1.05, 8.65) (0.58, 5.32) Hospital LOS (d)b 318 -1.26 0.65 165 -2.85 0.35 (-6.82, 4.29) (-8.84, 3.13) a Effect size is the hazard ratio (“no early steroids” as the reference group) b Effect size is the average difference (“no early steroids” as the reference group) CI = confidence interval, ICU = intensive care unit, LOS = length of stay, MV = mechanical ventilation, SOFA = Sequential Organ Failure Assessment Score

ACCEPTED MANUSCRIPT

Figure Legends

RI PT

Figure 1. Study flowchart. HIV = human immunodeficiency virus, ICD = International Classification of Diseases, PcP = Pneumocystis jirovecii pneumonia, PCR = polymerase chain reaction

Figure 2. ∆SOFAresp distribution of early steroid recipients vs. no early steroids

AC C

EP

TE D

M AN U

SC

Figure 3. Kaplan-Meier curves for 30-day survival curves in the unadjusted (a) and propensity matched (b) cohorts

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

e-Table 1. Breakdown of the SOFAresp score SOFAresp SaO2/FiO2 PaO2/FiO2 1 <512 <400 2 <357 <300 3 <214 <200 4 <89 <100 FiO2 = inspired oxygen fraction, PaO2 = partial pressure arterial oxygenation, SaO2 = oxygen saturation, SOFAresp = respiratory component of the sequential organ failure assessment score

e-Figure 1. Standardized differences in the unadjusted and propensity-matched groups

Online supplements are not copyedited prior to posting and the author(s) take full responsibility for the accuracy of all data.

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

e-Figure 2. Percentage breakdown of ∆SOFAresp

AC C

EP

TE D

e-Table 2. Sensitivity analyses of respiratory function in specified subgroups Adjusted for Baseline SOFA Effect size ΔSOFAresp N (95% CI) p-value a Mean change at day 5 Baseline steroid use 174 -0.41 (-0.61, -0.20) <0.001 No baseline steroid use 149 Reference Steroids within 48h of admission 161 -0.15 (-0.36, -0.06) 0.17 Steroids within 48h of PcP therapy 245 -0.38 (-0.62, -0.13) 0.003 ≥ one point improvement at day 5b Baseline steroid use 174 0.62 (0.36, 1.07) 0.085 No baseline steroid use 149 Reference Within 48h of admission 161 1.51 (0.88, 2.61) 0.13 Within 48h of anti-Pneumocystis therapy 245 1.07 (0.55, 2.07) 0.84 Effect size is the average difference (“no early steroids” as the reference group) wherein a positive value favors improvement in the early steroid group and a negative value favors improvement in the no early steroid group b Effect size is the odds ratio (“no early steroids” as the reference group) wherein a value > one favors an improvement in the early steroid group CI = confidence interval, SOFAresp = respiratory component of the Sequential Organ Failure Assessment score a

Online supplements are not copyedited prior to posting and the author(s) take full responsibility for the accuracy of all data.

ACCEPTED MANUSCRIPT

-0.34 (-0.77, 0.09) Day 7 112 -0.35 (-0.91, 0.22) ≥ one point improvementc

0.12

149

0.23

58

-0.43 (-0.78, -0.08) -0.29 (-0.96, 0.37)

0.017 0.38

174

1.08 0.91 149 0.57 0.25 (0.31, 3.72) (0.22, 1.49) Day 7 112 1.07 0.94 58 1.20 0.83 (0.21, 5.32) (0.22, 6.38) a Adjusted for baseline SOFA b Effect size is the average difference (“no early steroids” as the reference group) wherein a positive value favors improvement in the early steroid group and a negative value favors improvement in the no early steroid group c Effect size is the odds ratio (“no early steroids” as the reference group) wherein a value > one favors an improvement in the early steroid group CI = confidence interval, SOFAresp = respiratory component of the Sequential Organ Failure Assessment score

EP

TE D

M AN U

Day 5

174

SC

Day 5

RI PT

e-Table 3. Sensitivity analyses of respiratory function in subgroups of baseline steroid use Baseline Steroid Usea No Baseline Steroid Usea Effect size Effect size ΔSOFAresp N (95% CI) p-value N (95% CI) p-value b Mean change

AC C

e-Table 4. Comparison of change in SaO2/FiO2 Adjusted for Baseline SOFA Effect size SaO2/FiO2 N (95% CI) p-value N Changea Day 5 323 -33.3 0.038 162 (-64.9, -1.8) Day 7 170 -26.5 0.32 57 (-79.1, 26.0) a Effect size is the average difference (“no early steroids”

Propensity Analysis Effect size (95% CI) p-value -23.4 0.25 (-62.8, 16.1) -42.0 0.23 (-110.5, 26.5) as the reference group)

Online supplements are not copyedited prior to posting and the author(s) take full responsibility for the accuracy of all data.

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

e-Figure 3. Kaplan-Meier curves for 30-day survival in the sensitivity analyses of those with baseline steroid use: unadjusted (a) and adjusted for baseline SOFA (b), without baseline steroid use: unadjusted (c) and adjusted for baseline SOFA (d).

172644 Online supplements are not copyedited prior to posting and the author(s) take full responsibility for the accuracy of all data.