Primary Prophylaxis for Pneumocystis jirovecii Pneumonia in Patients Receiving Rituximab

BackgroundAlthough previous studies suggested that rituximab increases the risk of Pneumocystis jirovecii pneumonia (PJP), it is uncertain whether its primary prophylaxis for PJP is justified.Research QuestionDoes the benefit of primary prophylaxis for PJP in patients receiving rituximab treatment outweigh the potential risk of the prophylaxis?Study Design and MethodsThis retrospective study included 3,524 patients (hematologic diseases, 2,500; rheumatic diseases, 559; pre/post-solid organ transplantation, 465) first exposed to rituximab between 2002 and 2018 in a tertiary referral center in South Korea. Patients were classified into a control group (n = 2,523) and a prophylaxis group (n = 1,001) according to the administration of prophylactic trimethoprim-sulfamethoxazole (TMP-SMX) during the first 28 days after the start of rituximab (intention-to-treat analysis). In addition, exposure to TMP-SMX was examined as a time-varying variable (time-varying analysis). The primary outcome was the prophylactic effect of TMP-SMX on the 1-year incidence of PJP. Inverse probability of treatment weights was applied to minimize the baseline imbalance. The secondary outcome included the incidence of adverse drug reactions (ADRs) related to TMP-SMX.ResultsOver 2,759.9 person-years, 92 PJP infections occurred, with a mortality rate of 27.2%. The prophylaxis group showed a significantly lower incidence of PJP (adjusted subdistribution hazard ratio, 0.20 [95% CI, 0.10-0.42]) than the control group. This result was consistent with the results of time-varying analysis, in which only one PJP infection occurred during TMP-SMX administration (adjusted subdistribution hazard ratio, 0.01 [0.003-0.16]). The incidence of ADRs related to TMP-SMX was 18.1 (14.6-22.2)/100 person-years, and most were of mild to moderate severity. On the basis of 10 severe ADRs, the number needed to harm was 101 (61.9-261.1), whereas the number needed to prevent one PJP infection was 32 (24.8-39.4).InterpretationTMP-SMX prophylaxis significantly reduces PJP incidence with a tolerable safety profile in patients receiving rituximab treatment. Although previous studies suggested that rituximab increases the risk of Pneumocystis jirovecii pneumonia (PJP), it is uncertain whether its primary prophylaxis for PJP is justified. Does the benefit of primary prophylaxis for PJP in patients receiving rituximab treatment outweigh the potential risk of the prophylaxis? This retrospective study included 3,524 patients (hematologic diseases, 2,500; rheumatic diseases, 559; pre/post-solid organ transplantation, 465) first exposed to rituximab between 2002 and 2018 in a tertiary referral center in South Korea. Patients were classified into a control group (n = 2,523) and a prophylaxis group (n = 1,001) according to the administration of prophylactic trimethoprim-sulfamethoxazole (TMP-SMX) during the first 28 days after the start of rituximab (intention-to-treat analysis). In addition, exposure to TMP-SMX was examined as a time-varying variable (time-varying analysis). The primary outcome was the prophylactic effect of TMP-SMX on the 1-year incidence of PJP. Inverse probability of treatment weights was applied to minimize the baseline imbalance. The secondary outcome included the incidence of adverse drug reactions (ADRs) related to TMP-SMX. Over 2,759.9 person-years, 92 PJP infections occurred, with a mortality rate of 27.2%. The prophylaxis group showed a significantly lower incidence of PJP (adjusted subdistribution hazard ratio, 0.20 [95% CI, 0.10-0.42]) than the control group. This result was consistent with the results of time-varying analysis, in which only one PJP infection occurred during TMP-SMX administration (adjusted subdistribution hazard ratio, 0.01 [0.003-0.16]). The incidence of ADRs related to TMP-SMX was 18.1 (14.6-22.2)/100 person-years, and most were of mild to moderate severity. On the basis of 10 severe ADRs, the number needed to harm was 101 (61.9-261.1), whereas the number needed to prevent one PJP infection was 32 (24.8-39.4). TMP-SMX prophylaxis significantly reduces PJP incidence with a tolerable safety profile in patients receiving rituximab treatment. Pneumocystis jirovecii pneumonia (PJP) is a potentially life-threatening infection that occurs mainly in immunocompromised patients.1Thomas Jr., C.F. Limper A.H. Pneumocystis pneumonia.N Engl J Med. 2004; 350: 2487-2498Crossref PubMed Scopus (801) Google Scholar Effective treatment and an established prophylactic strategy in patients with HIV infection have led to a marked fall in occurrence.2Kaplan J.E. Hanson D. Dworkin M.S. et al.Epidemiology of human immunodeficiency virus-associated opportunistic infections in the United States in the era of highly active antiretroviral therapy.Clin Infect Dis. 2000; 30: S5-S14Crossref PubMed Scopus (604) Google Scholar However, the incidence of non-HIV PJP is increasing, with widespread use of immunosuppressive agents for the treatment of hematologic malignancies, rheumatic diseases, and solid organ transplantation.3Morris A. Lundgren J.D. Masur H. et al.Current epidemiology of Pneumocystis pneumonia.Emerg Infect Dis. 2004; 10: 1713-1720Crossref PubMed Scopus (323) Google Scholar, 4Maini R. Henderson K.L. Sheridan E.A. et al.Increasing Pneumocystis pneumonia, England, UK, 2000-2010.Emerg Infect Dis. 2013; 19: 386-392Crossref PubMed Scopus (109) Google Scholar, 5Fillatre P. Decaux O. Jouneau S. et al.Incidence of Pneumocystis jiroveci pneumonia among groups at risk in HIV-negative patients.Am J Med. 2014; 127: 1242.e11-1242.e17Abstract Full Text Full Text PDF Scopus (106) Google Scholar, 6Martin S.I. Fishman J.A. AST Infectious Diseases Community of PracticePneumocystis pneumonia in solid organ transplantation.Am J Transplant. 2013; 13: 272-279Crossref PubMed Scopus (154) Google Scholar Moreover, non-HIV PJP usually has more severe manifestations and carries a higher mortality rate than that in patients with HIV infection.7Yale S.H. Limper A.H. Pneumocystis carinii pneumonia in patients without acquired immunodeficiency syndrome: associated illness and prior corticosteroid therapy.Mayo Clin Proc. 1996; 71: 5-13Abstract Full Text Full Text PDF PubMed Scopus (488) Google Scholar, 8Bienvenu A.L. Traore K. Plekhanova I. Bouchrik M. Bossard C. Picot S. Pneumocystis pneumonia suspected cases in 604 non-HIV and HIV patients.Int J Infect Dis. 2016; 46: 11-17Abstract Full Text Full Text PDF PubMed Scopus (59) Google Scholar, 9Cordonnier C. Cesaro S. Maschmeyer G. et al.Pneumocystis jirovecii pneumonia: still a concern in patients with haematological malignancies and stem cell transplant recipients.J Antimicrob Chemother. 2016; 71: 2379-2385Crossref PubMed Scopus (70) Google Scholar, 10Monnet X. Vidal-Petiot E. Osman D. et al.Critical care management and outcome of severe Pneumocystis pneumonia in patients with and without HIV infection.Crit Care. 2008; 12: R28Crossref PubMed Scopus (130) Google Scholar Many previous studies of the pathophysiology of PJP in patients with HIV infection focus on T cells and show that cell-mediated immunity plays an important role in the clearance of microorganisms.11Kelly M.N. Shellito J.E. Current understanding of Pneumocystis immunology.Future Microbiol. 2010; 5: 43-65Crossref PubMed Scopus (58) Google Scholar,12Kelly M.N. Zheng M. Ruan S. Kolls J. D'Souza A. Shellito J.E. Memory CD4+ T cells are required for optimal NK cell effector functions against the opportunistic fungal pathogen Pneumocystis murina.J Immunol. 2013; 190: 285-295Crossref PubMed Scopus (40) Google Scholar However, accumulating evidence suggests that B cells play a critical role in T-cell-mediated immunity, and that abnormalities in B-cell numbers or function predispose to opportunistic infections such as PJP.13Hu Y. Wang D. Zhai K. Tong Z. Transcriptomic analysis reveals significant B lymphocyte suppression in corticosteroid-treated hosts with Pneumocystis pneumonia.Am J Respir Cell Mol Biol. 2017; 56: 322-331Crossref PubMed Scopus (13) Google Scholar,14Rong H.M. Li T. Zhang C. et al.IL-10-producing B cells regulate Th1/Th17-cell immune responses in Pneumocystis pneumonia.Am J Physiol Lung Cell Mol Physiol. 2019; 316: L291-L301Crossref PubMed Scopus (13) Google Scholar Rituximab, a chimeric anti-CD20 monoclonal antibody, is used widely to treat patients with hematologic malignancies, autoimmune diseases, ABO-incompatible transplantation, and antibody-mediated rejection. It exerts a therapeutic effect by complement and antibody-mediated B-cell depletion; however, this increases the risk of infectious complications.15Gea-Banacloche J.C. Rituximab-associated infections.Semin Hematol. 2010; 47: 187-198Crossref PubMed Scopus (201) Google Scholar,16Aksoy S. Dizdar O. Hayran M. Harputluoglu H. Infectious complications of rituximab in patients with lymphoma during maintenance therapy: a systematic review and meta-analysis.Leuk Lymphoma. 2009; 50: 357-365Crossref PubMed Scopus (84) Google Scholar However, although the prescription information for rituximab published by the US Food and Drug Administration recommends primary PJP prophylaxis for some indications, the incidence of PJP in patients receiving rituximab is unclear; indeed, few studies have investigated the efficacy of primary PJP prophylaxis in such patients.17Wei K.C. Sy C. Wu S.Y. Chuang T.J. Huang W.C. Lai P.C. Pneumocystis jirovecii pneumonia in HIV-uninfected, rituximab treated non-Hodgkin lymphoma patients.Sci Rep. 2018; 8: 8321Crossref PubMed Scopus (10) Google Scholar Therefore, it is uncertain whether PJP prophylaxis is indicated for patients starting rituximab.18Martin-Garrido I. Carmona E.M. Specks U. Limper A.H. Pneumocystis pneumonia in patients treated with rituximab.Chest. 2013; 144: 258-265Abstract Full Text Full Text PDF PubMed Scopus (116) Google Scholar,19Besada E. Routine Pneumocystis pneumonia prophylaxis in patients treated with rituximab?.Chest. 2013; 144: 359-360Abstract Full Text Full Text PDF PubMed Scopus (4) Google Scholar To address this question, we investigated the incidence of PJP in patients treated with rituximab at a large national tertiary referral center over a 16-year period. The aim was to evaluate the effectiveness and safety of PJP prophylaxis and to quantify the precise risk-benefit profile. This retrospective study included patients treated with rituximab for the first time between 2002 and 2018 at Seoul National University Hospital. According to the underlying disorder requiring rituximab treatment, all patients were classified into one of the three disease groups: hematologic disease, rheumatic disease, or pre/post-solid organ transplantation (TPL). Further information regarding patient selection, inclusion and exclusion criteria, the collection of clinical data, and diagnostic evaluation of patients with suspicious PJP is given in the online article (e-Appendix 1 and e-Fig 1). This study was carried out in accordance with the Declaration of Helsinki and was approved by the Seoul National University Hospital Institutional Review Board (Approval No. 1905-173-1036). The need for patient consent was waived due to the retrospective nature of the study. Because there have been no established guidelines on primary PJP prophylaxis in patients receiving rituximab treatment, the selection of patients who would receive prophylaxis and the duration of treatment were mainly at the discretion of the treating physician. In our institution, most physicians have prescribed trimethoprim-sulfamethoxazole (TMP-SMX) for primary PJP prophylaxis at a dose of one single-strength tablet per day, or one double-strength tablet three times per week. The dose of TMP-SMX was adjusted according to each patient's renal function. A review of all prescription data in the study population showed that no patient received second-line prophylactic agents such as dapsone and pentamidine during the observation period. Considering the heterogeneity regarding the starting point and duration of PJP prophylaxis among the study population, efficacy outcomes were assessed using two different time schemes (e-Fig 2). First, an intention-to-treat design (ie, "first exposure carried forward") was used, in which administration of TMP-SMX during the period between baseline (day 0, the day of the first rituximab administration) and day 28 (lead-in period) was necessary to determine whether a patient was to be included in the prophylaxis or unexposed (control) group. Using this scheme, 1,001 patients who received prophylactic TMP-SMX during the lead-in period were classified into the prophylaxis group. The unexposed group comprised 2,269 patients who were never exposed to TMP-SMX during follow-up and 254 patients in whom the start of TMP-SMX treatment was delayed (> 4 weeks). In the latter case, follow-up was censored if patients subsequently received TMP-SMX. Second, a time-varying analysis, in which prophylactic TMP-SMX use was modeled as a time-dependent variable, was performed. In this analysis, follow-up of all patients began at baseline, and each subsequent person-day of observation was classified according to whether the patient received prophylactic TMP-SMX. In the time-varying analysis, patients could be assigned to the exposed and unexposed categories without restriction. Finally, 5,265 episodes (4,006 without TMP-SMX and 1,259 with TMP-SMX) were analyzed (e-Table 1). To detect all PJP cases without misclassification, a two-step algorithm was designed (e-Fig 3). First, because a definite diagnosis of PJP requires identification of the organism, we first captured the 219 cases with positive results from direct immunofluorescence staining and/or PCR assays of induced sputum and BAL during the observation period. Next, two expert investigators (J. W. P. and K. I. J.) independently reviewed all medical records, laboratory data, and therapeutic antibiotic use in the selected cases to confirm PJP. To minimize bias, the investigators evaluated each case without information regarding whether the patient had received the prophylactic TMP-SMX previously. PJP was confirmed on the basis of (1) the presence of clinical and radiographic features suggestive of PJP, (2) the results of microbiologic tests for the identification of organisms other than Pneumocystis jirovecii, and (3) treatment responses to various antimicrobials. The final PJP cases were determined when both assessors agreed that the cases were consistent with PJP. The level of agreement between two investigators was excellent, with a κ value of 0.963 (95% CI, 0.927-0.998). The primary outcome was the prophylactic effect of TMP-SMX on the 1-year incidence of PJP. Secondary outcomes included the effect of prophylaxis on 1-year PJP-related death (defined as death caused by a progression to ARDS or respiratory failure due to uncontrolled PJP) and the incidence of adverse drug reactions (ADRs) related to prophylactic TMP-SMX. All patients were observed from day 29 (intention-to-treat analysis) or baseline (time-varying analysis), and were monitored up until PJP infection, death, loss to follow-up (defined as no subsequent visit for > 6 months from the last visit), or 52 weeks from the start of observation, whichever came first. The safety of prophylactic TMP-SMX was evaluated in two stages: first, all adverse events (AEs) that occurred during the period of prophylactic TMP-SMX administration were captured from the electronic medical database. Next, the probability of causation of each AE was estimated by one author (J. W. P.), based on timing, known AE profile, and improvement of AE after cessation of the agent. AEs showing probable/likely or certain causality were regarded as ADRs related to TMP-SMX.20Edwards I.R. Aronson J.K. Adverse drug reactions: definitions, diagnosis, and management.Lancet. 2000; 356: 1255-1259Abstract Full Text Full Text PDF PubMed Scopus (1893) Google Scholar The type of ADR and its severity were assessed according to the Common Terminology Criteria for Adverse Events, version 5.0.21US Department of Health and Human ServicesCommon Terminology Criteria for Adverse Events (CTCAE). Version 5.0. November 27, 2017.https://ctep.cancer.gov/protocoldevelopment/electronic_applications/docs/ctcae_v5_quick_reference_5x7.pdfDate accessed: January 27, 2022Google Scholar A severe ADR was defined as grade 3 or higher. For risk-benefit assessment, the number needed to treat (NNT) to prevent one case of PJP with prophylaxis, and the number needed to harm (NNH) for one severe ADR, were compared. There were no missing values with respect to clinical characteristics or laboratory findings; thus no imputation was performed. Propensity scores were developed to model the likelihood that a patient would receive TMP-SMX prophylaxis. Then, inverse probability of treatment weights (IPTW) was applied to balance the baseline characteristics between the two groups. This approach creates a pseudo-population in which exposure to TMP-SMX was independent of measured covariates.22Austin P.C. The performance of different propensity-score methods for estimating differences in proportions (risk differences or absolute risk reductions) in observational studies.Stat Med. 2010; 29: 2137-2148Crossref PubMed Scopus (220) Google Scholar Baseline imbalances before and after applying IPTW were estimated, using the standardized mean difference. For time-varying analysis, in which exposure of an individual patient to TMP-SMX can be changed over time, time-varying inverse probability weights were estimated.23Grafféo N. Latouche A. Geskus R.B. Chevret S. Modeling time-varying exposure using inverse probability of treatment weights.Biom J. 2018; 60: 323-332Crossref PubMed Scopus (5) Google Scholar The bootstrap method was used to calculate the 95% confidence rate of NNT (NNH). The efficacy of prophylactic TMP-SMX on outcome was assessed using Fine-Gray models, applying IPTW weights to control for confounders. A competing risk in the analysis of PJP incidence and related mortality was non-PJP-related death.24Austin P.C. Fine J.P. Practical recommendations for reporting Fine-Gray model analyses for competing risk data.Stat Med. 2017; 36: 4391-4400Crossref PubMed Scopus (331) Google Scholar After applying IPTW, the model was adjusted further using prespecified covariates (age, sex, baseline azotemia [glomerular filtration rate < 60 mL/min], baseline lymphopenia [lymphocyte count < 800/μL], and concomitant high-dose steroid [mean daily dose of steroid ≥ 30 mg/d prednisone or equivalent during the lead-in period according to previous studies])25Buttgereit F. da Silva J.A. Boers M. et al.Standardised nomenclature for glucocorticoid dosages and glucocorticoid treatment regimens: current questions and tentative answers in rheumatology.Ann Rheum Dis. 2002; 61: 718-722Crossref PubMed Scopus (342) Google Scholar, 26Park J.W. Curtis J.R. Moon J. Song Y.W. Kim S. Lee E.B. Prophylactic effect of trimethoprim-sulfamethoxazole for pneumocystis pneumonia in patients with rheumatic diseases exposed to prolonged high-dose glucocorticoids.Ann Rheum Dis. 2018; 77: 644-649Crossref PubMed Scopus (104) Google Scholar, 27Park J.W. Curtis J.R. Kim M.J. Lee H. Song Y.W. Lee E.B. Pneumocystis pneumonia in patients with rheumatic diseases receiving prolonged, non-high-dose steroids: clinical implication of primary prophylaxis using trimethoprim-sulfamethoxazole.Arthritis Res Ther. 2019; 21: 207Crossref PubMed Scopus (19) Google Scholar to generate effect estimates robust to misspecification of the model based on IPTW.28Funk M.J. Westreich D. Wiesen C. Sturmer T. Brookhart M.A. Davidian M. Doubly robust estimation of causal effects.Am J Epidemiol. 2011; 173: 761-767Crossref PubMed Scopus (403) Google Scholar A robust sandwich-type variance estimator was used to examine within-subject correlation.29Ellis A.R. Brookhart M.A. Approaches to inverse-probability-of-treatment-weighted estimation with concurrent treatments.J Clin Epidemiol. 2013; 66: S51-S56Abstract Full Text Full Text PDF PubMed Scopus (11) Google Scholar Several sensitivity analyses were performed. First, the efficacy of prophylactic TMP-SMX was estimated after excluding patients with concomitant high-dose steroid treatment (n = 947). Second, multivariable analysis was repeated without IPTW. Third, the main analysis was performed after applying 1:1 propensity-score matching alternatively (n = 1,272; e-Table 2). Fourth, the efficacy of prophylactic TMP-SMX was estimated in the subgroup of patients who were not treated with any concomitant immunosuppressive and/or antineoplastic agents other than rituximab and glucocorticoids (n = 528). Finally, the potential effect of unmeasured confounding was estimated using the E-value, defined as the minimum strength of the association that an unmeasured confounder would need to have with both treatment and outcome to fully explain a specific treatment-outcome association.30VanderWeele T.J. Ding P. Sensitivity analysis in observational research: introducing the E-value.Ann Intern Med. 2017; 167: 268-274Crossref PubMed Scopus (1559) Google Scholar All statistical analyses were performed with R version 3.6.1 (R Foundation for Statistical Computing) and SAS version 9.4 (SAS Institute), and a P value < .05 was considered statistically significant. A total of 3,524 patients treated with rituximab were analyzed. The most common disease group was hematologic disease (n = 2,500; 71.0%), followed by rheumatologic disease (n = 559; 15.9%) and pre/post-TPL (n = 465; 13.2%). The baseline characteristics of the prophylaxis and control groups are presented in Table 1. After applying IPTW, all measured covariates were well balanced (standardized mean difference < 0.1). In the prophylaxis group, the mean (SD) duration of TMP-SMX administration was 153.8 (107.6) days.Table 1Baseline Characteristics of Study Population, Grouped According to Intention-to-Treat AnalysisCharacteristicControl Group (n = 2,523)Prophylaxis Group (n = 1,001)SMD Before IPTWSMD After IPTWAge, mean (SD), y56.2 (15.1)56.9 (14.5)0.0460.032Male, No. (%)1,345 (53.3)523 (52.2)0.0210.005Underlying diseases, No. (%) Rheumatoid arthritis80 (3.2)10 (1.0)0.1520.029 ANCA-associated vasculitis22 (0.9)77 (7.7)0.3420.018 Systemic lupus erythematosus30 (1.2)17 (1.7)0.0430.040 Systemic sclerosis9 (0.4)1 (0.1)0.0540.074 Sjögren syndrome7 (0.3)6 (0.6)0.0490.003 Inflammatory myositis26 (1.0)20 (2.0)0.0790.031 IgG4-related disease10 (0.4)10 (1.0)0.0720.022 Pemphigus55 (2.2)132 (13.2)0.4220.004 Other rheumatic diseasesaIncludes polyarteritis nodosa, anti-phospholipid antibody syndrome, and cryoglobulinemic vasculitis.42 (1.7)5 (0.5)0.1130.013 Diffuse large B-cell lymphoma1,447 (57.4)232 (23.2)0.7430.013 Follicular lymphoma161 (6.4)31 (3.1)0.1550.011 Mantle cell lymphoma79 (3.1)16 (1.6)0.1010.014 Primary CNS lymphoma36 (1.4)124 (12.4)0.4430.025 Chronic lymphoid leukemia64 (2.5)6 (0.6)0.1560.003 Burkitt lymphoma51 (2.0)20 (2.0)0.0020.002 Marginal zone B-cell lymphoma74 (2.9)13 (1.3)0.1140.013 Acute lymphoblastic leukemia10 (0.4)10 (1.0)0.0720.008 MALToma40 (1.6)2 (0.7)0.0830.004 Allo-SCT2 (0.1)3 (0.3)0.051< 0.001 GVHD3 (0.1)14 (1.4)0.1480.012 Other hematologic diseasesbIncludes immune thrombocytopenic purpura, thrombotic thrombocytopenic purpura, intravascular B-cell lymphoma, and Waldenström macroglobulinemia.48 (1.9)8 (0.8)0.0960.006 Liver transplantation9 (0.4)107 (10.7)0.4640.055 Kidney transplantation203 (8.0)117 (11.7)0.1220.002 Other transplantation15 (0.6)14 (1.4)0.0810.006Previous history of chemotherapy, No. (%)129 (5.1)53 (5.3)0.0080.051Azotemia,cDefined as glomerular filtration rate < 60 mL/min. No. (%)360 (14.3)226 (22.6)0.2160.009Baseline lymphopenia,dDefined as lymphocyte count < 800/μL. No. (%)443 (17.6)269 (26.9)0.2250.004Cumulative steroid use (based on prednisone) during past 6 mo, mean (SD), mg445.6 (2,608.1)731.7 (1,929.3)0.1250.014Concomitant high-dose steroid,eDefined as mean dose of steroid ≥ 30 mg/d of prednisone or equivalent during the lead-in period. No. (%)519 (20.6)428 (42.8)0.4910.001ANCA = anti-neutrophil cytoplasmic antibody; GVHD = graft-vs-host disease; IPTW = inverse probability of treatment weights; MALT = mucosa-associated lymphoid tissue; SCT = stem cell transplantation; SMD = standardized mean difference.a Includes polyarteritis nodosa, anti-phospholipid antibody syndrome, and cryoglobulinemic vasculitis.b Includes immune thrombocytopenic purpura, thrombotic thrombocytopenic purpura, intravascular B-cell lymphoma, and Waldenström macroglobulinemia.c Defined as glomerular filtration rate < 60 mL/min.d Defined as lymphocyte count < 800/μL.e Defined as mean dose of steroid ≥ 30 mg/d of prednisone or equivalent during the lead-in period. Open table in a new tab ANCA = anti-neutrophil cytoplasmic antibody; GVHD = graft-vs-host disease; IPTW = inverse probability of treatment weights; MALT = mucosa-associated lymphoid tissue; SCT = stem cell transplantation; SMD = standardized mean difference. Overall, during 2,759.9 person-years, 92 cases of PJP (10 in the rheumatic disease group, 64 in the hematologic disease group, and 18 in the pre/post-TPL disease group) occurred, with a crude incidence rate (per 100 person-years) of 3.33 (95% CI, 2.69-4.09). The cumulative incidence of PJP in the control group was 4.11 (3.26-5.12) with 2.96 (1.19-6.09) for rheumatic disease, 4.50 (3.44-5.78) for hematologic disease, and 7.01 (3.63-12.25) for pre/post-TPL (e-Fig 4). The median (interquartile range) time interval between baseline and PJP infection was 86.0 (80.0) days. PJP-related mortality was 27.2% (25 of 92). The clinical features of these 92 cases at the time of PJP diagnosis were summarized in e-Table 3. A total of 356 non-PJP-related deaths occurred. The prevalence of non-PJP-related death was comparable between the control and the prophylaxis groups (10.3% vs 9.5%; P = .448) (e-Table 4). Univariable analysis identified 1-year incidence of PJP as being significantly associated with age at baseline, disease group, baseline lymphopenia, azotemia, and concomitant treatment with high-dose steroid (e-Table 5). By contrast, concomitant treatment with lower dose of prednisone (20-30 mg/d) was not significantly associated with increased risk of PJP (subdistribution hazard ratio, 1.32 [95% CI, 0.81-2.14]). Multivariable analysis identified azotemia (adjusted subdistribution hazard ratio [SHR], 2.38 [1.75-3.23]) and concomitant treatment with high-dose steroid (adjusted SHR, 3.09 [2.22-4.30]) as the two most important factors that increase the risk of PJP. Intention-to-treat analysis revealed that 80 and 12 cases of PJP occurred in the control and prophylaxis groups, respectively. Prophylaxis significantly reduced the 1-year incidence of PJP (adjusted SHR, 0.20 [0.10-0.42]) and related mortality (adjusted SHR, 0.21 [0.05-0.84]) (Fig 1, Table 2), and its effect was consistent in all disease groups (Table 3, e-Fig 5).Table 2Effect of TMP-SMX Prophylaxis on the 1-Year Incidence of PJP and Related Mortality, Based on Intention-to-Treat and Time-Varying AnalysesParameterIntention-to-Treat Analysis (No. of Patients = 3,524)Control GroupProphylaxis GroupNo. of PJP cases/follow-up period80/1,939.6 person-y12/815.3 person-yCumulative incidence of PJPaPer 100 person-years. (95% CI)4.13 (3.27-5.13)1.47 (0.76-2.57)No. of PJP-related death cases/follow-up period21/2,166.5 person-y4/818.9 person-yCumulative incidence of PJP-related deathaPer 100 person-years. (95% CI)0.97 (0.60-1.48)0.49 (0.13-1.25)Unadjusted SHR for PJP (95% CI)Reference0.22 (0.11-0.46)Adjusted SHR for PJP (95% CI)bAdjusted by age, sex, baseline azotemia, baseline lymphopenia, and concomitant high-dose steroid treatment.Reference0.20 (0.10-0.42)Unadjusted SHR for PJP-related death (95% CI)Reference0.22 (0.06-0.84)Adjusted SHR for PJP-related death (95% CI)bAdjusted by age, sex, baseline azotemia, baseline lymphopenia, and concomitant high-dose steroid treatment.Reference0.21 (0.05-0.84)ParameterTime-Varying Analysis (No. of Episodes = 5,265)Episodes Without TMP-SMXEpisodes With TMP-SMXNo. of PJP cases/follow-up period95/2,472.3 person-y1/509.1 person-yCumulative incidence of PJPaPer 100 person-years. (95% CI)3.84 (3.11-4.70)0.20 (0.005-1.09)No. of PJP-related death cases/follow-up period24/2,518.3 person-y1/509.1 person-yCumulative incidence of PJP-related deathaPer 100 person-years. (95% CI)1.17 (0.67-2.05)0.20 (0.005-1.09)Unadjusted SHR for PJP (95% CI)Reference0.01 (0.003-0.16)Adjusted SHR for PJP (95% CI)bAdjusted by age, sex, baseline azotemia, baseline lymphopenia, and concomitant high-dose steroid treatment.Reference0.01 (0.003-0.16)Unadjusted SHR for PJP-related death (95% CI)Reference0.11 (0.01-0.79)Adjusted SHR for PJP-related death (95% CI)bAdjusted by age, sex, baseline azotemia, baseline lymphopenia, and concomitant high-dose steroid treatment.Reference0.09 (0.01-0.66)PJP = Pneumocystis jirovecii pneumonia; SHR = subdistribution hazard ratio; TMP-SMX = trimethoprim-sulfamethoxazole.a Per 100 person-years.b Adjusted by age, sex, baseline azotemia, baseline lymphopenia, and concomitant high-dose steroid treatment. Open table in a new tab Table 3Prophylactic Effect of TMP-SMX on Cumulative 1-Year PJP Incidence and Related Mortality in Each Disease Group, Using Intention-To-Treat AnalysisDisease Group1-Year PJP Incidence1-Year PJP-Related MortalitySHR (95% CI)SHR (95% CI)Univariable AnalysisMultivariable AnalysisaAdjusted by age, sex, baseline azotemia, baseline lymphopenia, and concomitant high-dose steroid treatment.Univariable AnalysisMultivariable AnalysisaAdjusted by age, sex, baseline azotemia, baseline lymphopenia, and concomitant high-dose steroid treatment.Rheumatic disease0.12 (0.03-0.50)0.10 (0.02-0.38)0.13 (0.02-0.75)0.06 (0.003-0.99)Hematologic disease0.16 (0.04-0.59)0.16 (0.04-0.57)0.22 (0.03-1.81