Evaluation of the corticosteroid dose at Pneumocystis pneumonia onset in non-HIV patients receiving steroids and the period from the discontinuation of trimethoprim-sulfamethoxazole prophylaxis to the onset of Pneumocystis pneumonia

Pneumocystis pneumonia (PCP) is a life-threatening opportunistic infection among non-human immunodeciency virus (non-HIV) immunocompromised patients. The prophylactic use of trimethoprim-sulfamethoxazole (TMP-SMX) reduces PCP incidence. However, it remains unclear when TMP-SMX can be safely discontinued among patients for whom corticosteroid tapering is underway, and occasionally, PCP develops after TMP-SMX discontinuation despite tapering of corticosteroids to considerably lower dose. Methods


Background
Pneumocystis pneumonia (PCP), which is caused by Pneumocystis jirovecii, is a common opportunistic infection among human immunode ciency virus (HIV)-positive patients as well as HIV-negative (non-HIV) immunocompromised patients with hematologic malignancies; solid and bone marrow transplant recipients; and patients administered with immunosuppressive therapies such as corticosteroids, biological drugs, or anticancer drugs. The mortality rate in HIV-positive patients with PCP has declined to 10-20%; however, among non-HIV patients, PCP is life-threatening and the mortality rate remains 30-60% [1].
The use of prophylactic drugs can reduce the incidence of PCP in both HIV patients and non-HIV immunocompromised patients. Although primary prophylactic treatment for PCP is considered in HIV patients with CD4 cell counts of < 200 cells/mm, the treatment has been controversial in non-HIV immunocompromised patients [2][3][4]. Yale and Limper [5] suggested that primary prophylaxis should be considered for patients receiving corticosteroid therapy equivalent to at least 20 mg of prednisolone (PSL) daily for more than 4 weeks in non-HIV patients. However, we have experienced several cases in which PCP developed after the discontinuation of primary prophylaxis despite tapering of corticosteroids to lower than the daily 20 mg PSL equivalent dose. It is necessary to establish a consensus about when prophylaxis treatment against PCP should be started and when it can be safely stopped in non-HIV immunocompromised patients for whom corticosteroid therapy is being tapered.
Trimethoprim-sulfamethoxazole (TMP-SMX) is the rst choice for prophylactic treatment in HIV patients. The prophylactic effect of TMP-SMX in non-HIV immunocompromised patients has also been reported [6,7]. However, the prophylactic use of TMP-SMX is known to be associated with many adverse effects, such as hyperkalaemia, pancytopenia, rash, interstitial nephritis, aseptic meningitis, hepatitis, and pancreatitis [8], which often result in its discontinuation. Although aerosolized pentamidine or atovaquone can be used as the second or third choice when TMP-SMX is unavailable [9], these treatments might not be initiated until the improvement of the adverse event caused by the TMP-SMX treatment. Such delays in restarting prophylaxis could lead to the development of PCP. If it is known how long the prophylactic effects of TMP-SMX prevent PCP after treatment discontinuation, the risk of PCP development could be more precisely evaluated in such conditions. In this study, we retrospectively investigated the dose of corticosteroids at the time of PCP development in non-HIV immunocompromised patients. We also evaluated the period from the discontinuation of TMP-SMX prophylaxis to the onset of PCP to elucidate the prolonged prophylactic effect of this treatment.

Patients
We included non-HIV immunocompromised patients who were diagnosed with PCP from January 2007 to December 2018 in Tazuke Kofukai Medical Research Institute, Kitano Hospital, Osaka, Japan. The study was approved by the ethics committee of Kitano Hospital (approval no. P200501200). The inclusion criteria were as follows: (1) presence of an underlying immunode ciency known to be associated with PCP; (2) clinical symptoms consistent with a lower respiratory tract infection, e.g., cough and/or dyspnoea; (3) presence of new pulmonary in ltrates on chest radiography; (4) the presence of P. jirovecii proven by deoxyribonucleic acid (DNA) detection through polymerase chain reaction (PCR) or a positive uorescent staining result for a bronchoalveolar lavage uid (BALF) or sputum sample; and (5) elevated plasma (1-3)-β-D-glucan (β-D-glucan) levels measured using the Wako β-glucan assay (Fuji lm Wako Chemicals). We excluded patients infected with HIV.

Data collection
Patients' clinical and laboratory data were retrospectively reviewed and collected. These data included age; sex; smoking status; underlying diseases such as autoimmune diseases, malignancies, interstitial pneumonia, diabetes mellitus, and renal diseases; history of renal transplantation; corticosteroid dose and duration; immunosuppressant use; prophylactic TMP-SMX use (if discontinued, the period from discontinuation to the development of PCP); prophylactic aerosolized pentamidine or atovaquone use; initial laboratory ndings; diagnostic methods (BALF or sputum); and in-hospital mortality.

Statistical analysis
Categorical variables are described as counts and percentages, and continuous variables as median values and ranges. Fisher's exact test was used to analyse categorical variables, and the Mann-Whitney U test was used to analyse continuous variables to compare two groups. A two-sided P value of less than 0.05 was considered statistically signi cant. All statistical analyses were performed with the statistical software R (version 3.6.0; R Foundation).

Patient characteristics
In all, 210,414 patients who were hospitalized in our facility during the study period were screened. Of those, 39 patients met all the inclusion criteria ( Fig. 1). Fifteen (38.5%) and 23 (59.0%) patients were diagnosed as P. jirovecii-positive based on PCR results of BALF and sputum samples, respectively. Fluorescent staining of BALF samples was positive for four patients (10.3%) including one patient with a negative PCR result. The patients' baseline characteristics are shown in Table 1. The median patient age was 70 years (range: 27-92 years), and 18 patients (46.2%) were female. Twenty-one patients (53.8%) had autoimmune diseases and 13 patients (33.3%) had malignancies; seven of the 13 patients had had hematologic malignancies. Thirty-two (82.1%) and 14 (35.9%) patients were being treated with corticosteroids and immunosuppressants, respectively, and among them, 11 (28.2%) received both corticosteroids and immunosuppressants at the time of PCP development. Among the 32 patients who were being treated with corticosteroids, the daily corticosteroid dose converted to the PSL equivalent dose ranged from 2 to 60 mg, with the median dose being 14 mg. Further, 24 (61.5%), 11 (28.2%), and six (15.4%) patients were treated with a dose of 20 mg or less, 10 mg or less, and 5 mg or less, respectively (Fig. 2). The corticosteroid administration period ranged from 27 to 2,084 days, with the median period being 121 days. None of 39 patients was being administered TMP-SMX for prophylaxis when PCP developed. Aerosolized pentamidine was administered to only four patients; in three of the patients, it was administered from the beginning of prophylactic treatment, and in the remaining patient, it was administered after TMP-SMX discontinuation. The remaining 35 patients were not being administered any prophylactic drug when PCP developed; 25 of these patients never received prophylaxis, while the remaining 10 had discontinued TMP-SMX before PCP development. Seventeen patients (43.6%) died at the hospital.
Comparison of the characteristics of patients who were never administered and those who discontinued the prophylactic TMP-SMX treatment Twenty-ve patients were never administered TMP-SMX as prophylaxis for PCP and 11 other patients discontinued TMP-SMX before PCP development (Table 2). There were no signi cant differences between the two groups in age, sex, underlying diseases, initial laboratory ndings, corticosteroid use or dose, immunosuppressant use, diagnostic methods, and in-hospital mortality. The duration of corticosteroid treatment was signi cantly longer in the patients who discontinued prophylactic TMP-SMX than in those who were never administered TMP-SMX (

Patients who discontinued prophylactic TMP-SMX
The detailed characteristics of the 11 patients who discontinued prophylactic TMP-SMX before PCP development are shown in Table 3. Only one patient was administered prophylactic aerosolized pentamidine after TMP-SMX discontinuation due to its side effects. A lymphoma patient was never administered corticosteroid therapy, while the remaining 10 patients had been administered corticosteroids for more than 116 days. The daily corticosteroid dose at the time of PCP development, which was converted to the PSL equivalent dose, ranged from 2 to 30 mg, with the median dose was 10 mg. Four patients (36.4%) were administered a dose of 5 mg or less. All of the patients who died at the hospital were administered a dose of 10 mg or higher. The period from TMP-SMX discontinuation to PCP development ranged from 44 to 175 days, with the median duration being 95 days, and in nine patients (81.8%), PCP developed 14 ± 2 weeks after TMP-SMX discontinuation (Table 3 and Fig. 3).

Discussion
In this study, we evaluated the detailed characteristics of 39 non-HIV patients who developed PCP in our hospital. PCP developed in 19 patients (48.7%) treated with a corticosteroid dose lower than the daily 20 mg PSL equivalent dose, which is the threshold for primary prophylaxis as suggested previously [5]. Furthermore, this nding is consistent that reported previously [10][11][12]. PCP may be attributable to the different immune statuses of patients who had different underlying diseases and were treated with different immunosuppressive drugs, and the heterogeneity might have made it di cult to evaluate the necessity or indication of prophylaxis for PCP [5,13].
Moreover, there is little evidence-based guidance about the appropriate timing of discontinuing prophylaxis for PCP [5,14], and Wolfe et al. [15] reported that 52% of infectious disease physicians consider a dose equivalent to a daily dose of 16-20 mg PSL as the criterion for discontinuing prophylaxis. In recent years, however, many patients have been treated with corticosteroids, immunosuppressive drugs, traditional and new anticancer drugs or biological drugs, or combinations of those. These varieties make it di cult to determine when the prophylaxis for PCP can be stopped safely.
In our study, among the 10 patients who had been administered corticosteroid therapy and discontinued prophylactic TMP-SMX before the development of PCP (patient no. 2-11 in Table 3), the corticosteroid dose at the time of TMP-SMX discontinuation varied from 0 to 30 mg equivalent of the daily PSL dose. Three of these patients received TMP-SMX until the corticosteroid dose was tapered to less than 5 mg; one patient (patient no. 3 in Table 3) discontinued TMP-SMX after the cessation of corticosteroid therapy.
However, corticosteroid therapy at a dose equivalent to a daily PSL dose of 2.5 mg was restarted owing to the recurrence of graft-versus-host disease. Our study indicates that physicians should be vigilant about the cessation of prophylaxis for PCP, even if the prescribed corticosteroid dose itself is low.
In this study, no patient was administered TMP-SMX for prophylaxis when PCP developed. This means that there was no breakthrough infection of PCP in non-HIV patients with TMP-SMX prophylaxis regardless of underlying characteristics during the study period in our hospital. These ndings are consistent with previous reports, which show TMP-SMX may inhibit PCP strongly in non-HIV patients [16,17]. There was no signi cant difference in the patient background between the patients who were never treated with prophylactic TMP-SMX and those who discontinued prophylactic TMP-SMX, except for the duration of prior steroid administration. The reason for this difference may be the effect of TMP-SMX. It is understandable that the time from steroid administration to PCP development was longer in the group that received prophylaxis because PCP was prevented during the period under prophylaxis.
Eleven patients developed PCP between 44 and 175 days after the discontinuation of prophylactic TMP-SMX, and nine developed the disease approximately 14 weeks after prophylaxis withdrawal (Fig. 3). This suggests that patients have the maximum risk of developing PCP approximately 3 months after the discontinuation of TMP-SMX prophylaxis. Because the half-lives of both TMP and SMX are shorter than 24 h, the pharmacological prophylactic effect against P. jirovecii may persist for much shorter than 3 months. Therefore, we consider that the reason for the 3-month period might be the life cycle of P. jirovecii. The culture of P. jirovecii is di cult and its life cycle, especially in the human body, is still unclear [18]. Therefore, we cannot prove this consideration by experimental science. However, Yale and Limper [5] reported that the median duration of corticosteroid therapy was 12 weeks before the development of PCP in non-HIV patients with various underlying immunosuppressive disorders. In our study, the median period was 88 days (range: 27-728 days) from the initiation of corticosteroid therapy to PCP development in the patients without TMP-SMX prophylaxis (data not shown). These similarities suggest that it is reasonable to consider that approximately 3 months are needed for P. jirovecii infection to exacerbate and PCP development after the loss of suppression by host immunity or prophylactic agents. Hence, prophylaxis should be restarted within 3 months after the adverse event-induced early discontinuation of the primary prophylaxis. Moreover, we could evaluate the possibility of PCP in non-HIV immunocompromised patients with respiratory symptoms by considering the period from discontinuation of TMP-SMX.

Limitation
There are several limitations to our study. First, P. jirovecii detection using PCR or uorescent staining could provide false-positive results. However, the inclusion criteria comprised clinical symptoms, chest radiography ndings, and laboratory ndings to exclude such false-positive result as much as possible. Second, the present study was a single-centre retrospective study with a small number of cases. However, it might be di cult to conduct prospective studies because of ethical issues or the frequency of PCP development. Third, this study included all non-HIV patients with PCP. Therefore, the results obtained for the overall patient population comprising individuals with various comorbidities may not be applicable in the case of individual diseases.

Conclusions
Our observational study showed that among non-HIV patients, PCP develops under corticosteroid therapy at doses much lower than the daily 20 mg PSL equivalent dose, indicating that clinicians should consider TMP-SMX prophylaxis for PCP even if the dose of corticosteroid administered is low. Furthermore, non-HIV immunocompromised patients are at a higher risk of developing PCP approximately 3 months after the discontinuation of TMP-SMX prophylaxis. To our knowledge, this may be the rst report to elucidate the time required for PCP development after the discontinuation of prophylaxis. Further study is warranted to con rm this result.

Tables
Due to technical limitations, the tables are only available as a download in the supplemental les section. Figure 1 Inclusion chart for our analysis of 39 patients who were diagnosed with non-HIV PCP. In all, 52 patients had the presence of P. jirovecii proven by deoxyribonucleic acid (DNA) detection through polymerase chain reaction (PCR) or a positive uorescent staining result for a bronchoalveolar lavage uid (BALF) or sputum sample. Eight patients were excluded because they were infected with human immunode ciency virus (HIV), four patients were because their plasma (1-3)-β-D-glucan (β-D-glucan) were below sensitivity and one patient was because he was without any clinical symptoms consistent with a lower respiratory tract infection. A total of 39 patients were included and analysed.

Figure 2
The daily dose of corticosteroid converted to the PSL equivalent dose when PCP developed. Thirty-two patients were treated with corticosteroids. The daily dose of corticosteroid converted to the PSL equivalent dose ranged from 2 to 60 mg. Further, 24, 11, and six patients were administered 20 mg or less, 10 mg or less, and 5 mg or less.

Figure 3
The period from discontinuation of TMP-SMX to development of PCP. The period from discontinuation of TMP-SMX to development of PCP ranged from 44 to 175 days, and in nine patients, PCP developed 14 ± 2 weeks after TMP-SMX discontinuation.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Table1.xlsx