Duration of prednisone treatment before development of Pneumocystis jirovecii pneumonia in patients with vasculitis: A case series

Abstract

BACKGROUND

Optimal timing for Pneumocystis jirovecii pneumonia (PCP) prophylaxis among patients with vasculitis is not clear. We set out to characterize the clinical presentation and duration of prednisone use before the development of PCP among these patients.

METHODS

All patients with PCP at The Ottawa Hospital (TOH) between 2006 and 2017 were identified. Using TOH data repositories, the following data were extracted: prednisone dosage, treatment duration, other immunosuppressive medications, PCP prophylaxis, PCP treatment, and death. Data were reported as median and range or as mean and standard deviation.

RESULTS

We identified seven patients (5 men, 2 women) with biopsy-proven vasculitis who developed PCP: six with anti-neutrophil cytoplasmic antibody–associated vasculitis and one with giant cell arteritis. None of the patients were on PCP prophylaxis. The most common symptoms on presentation were cough and dyspnea. At diagnosis, the median lymphocyte count was 0.30 × 10⁹/L (range 0.03–2.10), creatinine was 186 µmol/L (range 78–359), and lactate dehydrogenase was 471 U/L (range 301–1032). All patients were on prednisone at time of PCP diagnosis, with six on doses of ≥20 mg/day for at least 12 weeks. All but one patient were on additional immunosuppressants, with cyclophosphamide being the most common agent for five of the seven patients. Four (57%) required intensive care unit admission, and two (29%) died secondary to complications of PCP.

CONCLUSIONS

PCP is a severe and often fatal opportunistic infection among immunocompromised patients with vasculitis. Frequent evaluation of the need for prophylaxis is required for patients who remain on high-dose steroids and concomitant immunosuppressants.

The immunosuppressive treatment for immune-mediated vasculitides has contributed to increased mortality from opportunistic infections such as Pneumocystis jirovecii pneumonia (PCP) (1). Risk factors for PCP include older age, leuko- and lymphopenia, higher baseline creatinine, dialysis dependency, higher cumulative exposure to steroids, and an aggressive cyclophosphamide regimen (1). Overall, prolonged corticosteroid use (>4 wk) at doses >15–20 mg/day is the best defined risk factor for PCP (2). As such, the recommendation has been to consider PCP prophylaxis when prolonged systemic corticosteroid treatment at ≥20 mg/day for more than 4 weeks is expected (3).

However, PCP prophylaxis prescribing patterns for rheumatology patients remain variable (4,5). To inform a clinical approach to PCP prophylaxis in patients with vasculitis, we describe case series of patients with vasculitis who developed PCP at The Ottawa Hospital (TOH), a tertiary care hospital in Canada.

METHODS

A review of TOH data warehouse identified all patients with detected P. jirovecii in bronchoalveolar lavage fluid or bronchial washing samples by direct immunofluorescent assay between 2006 and 2017. Through review of electronic medical records (EMRs), we identified patients who developed PCP and had underlying vasculitis. Vasculitis diagnosis was made by histopathology, whenever available, or laboratory findings. Clinical data were extracted from EMRs. Descriptive data analysis was performed. For a literature review, we performed a database search of MEDLINE, EMBASE, Scopus, and Web of Science for articles of similar cases series of PCP and confirmed vasculitis from inception to November 2021, using the terms vasculitis and PCP or Pneumocystis jirovecii, prophylaxis, or both. Conference abstracts, reviews, or case series without enough primary data on details of the cases were excluded.

This study was approved by the Ottawa Health Science Network Research Ethics Board (No. 20170859-01H). Informed consent was waived because of the nature of the retrospective case series study.

RESULTS

One hundred twenty-three patients with PCP were identified. Of these, seven (6%; 5 women, 2 men) had vasculitis, and none was receiving PCP prophylaxis (Table 1). Vasculitis was diagnosed according to renal biopsy (n = 4), lung biopsy (n = 1), temporal artery biopsy (n = 1), and positive anti-neutrophil cytoplasmic antibody (ANCA) test (n = 1). Median age at time of diagnosis was 59 years (range 43–76). Of those with ANCA-associated vasculitis (AAV), four had isolated renal involvement and two had both renal and pulmonary involvement.

Table 1:

Clinical features of seven patients with vasculitis diagnosed with PCP

Clinical feature	Patient no.
	1	2	3	4	5	6	7
Sex and age at PCP diagnosis	F, 49	F, 43	F, 76	F, 54	F, 59	M, 72	M, 62
Type of vasculitis	GPA	GPA	MPA	MPA	MPA	MPA	GCA
Organ involvement
Kidney	X (no RRT)	X (no RRT)	X (RRT)	X (RRT)	X (no RRT)	X (no RRT)	* (no RRT)
Lung		X	X				*
Symptoms at presentation	Dyspnea, cough, fever	Dyspnea, cough	Dyspnea, weight loss	Cough, fever	Dyspnea, cough, fever	Dyspnea, cough, hemopytsis	Dyspnea, cough
Length of hospital stay, d	10	8	14	45	5	4	73
Lymphocytes, × 109/L	0.30	0.03	0.10	0.10	0.50	2.10	0.40
Creatinine, µmol/L	78	183	359	253	201	107	186
LDH, U/L	409	471	910	1,032	301	No data	No data
Comorbidities	Seronegative RA, GERD	DM II, Afib, CHF, osteoporosis, reactive airway disease	HTN, DM II, ischemic heart disease, AAA, Afib	HTN, DM II, hypothyroidism, GERD	HTN	HTN, DM II, hypothyroidism, scleroderma, interstitial lung disease secondary to scleroderma, aortic stenosis, history of lung cancer (stage 2 non-small cell carcinoma) treated with only surgical resection 7 y before diagnosis of vasculitis	COPD, CHF, steroid-induced DM, GERD, Afib, polyneuropathies
Dose of corticosteroid at PCP diagnosis, mg/day	20	65	50	30	5	30	50
Other immunosuppressive agents used	CYC, MTX	AZA, CYC, MTX, RTX	CYC	CYC	AZA, RTX	CYC, AZA, MMF, RTX	None
Diagnostic test and location of specimen	BAL, right/left lung	Bronchial washing, RLL	BAL, left lung	Bronchial washings, no data	BAL, RUL	BAL, RUL	BAL, RUL
Time from admission to PCP diagnosis, d	3	4	5	4	4	2	48
Time from diagnosis of vasculitis to PCP, mo	5	79	3	4	3	94	7
Outcome
ICU admission and length of stay, d	ICU required with intubation, 5 d	ICU required with intubation, 8 d	ICU required with intubation, 14 d	ICU required with intubation, 30 d	No ICU	No ICU	No ICU
Mortality at 30 d	Alive	Deceased	Deceased	Alive	Alive	Alive	Alive

*Concomitant COPD and chronic kidney disease

PCP = Pneumocystis jirovecii pneumonia; F = Female; M = Male; GPA = Granulomatosis with polyangiitis; MPA = Microscopic polyangiitis; GCA = Giant cell arteritis; RRT = Renal replacement therapy; LDH = Lactate dehydrogenase; RA = Rheumatoid arthritis; GERD = Gastroesophageal reflux disease; DM II = Diabetes mellitus type 2; Afib = Atrial fibrillation; CHF = Congestive heart failure; HTN = hypertension; AAA = Abdominal aortic aneurysm; COPD = Chronic obstructive lung disease; DM = Diabetes mellitus; CYC = Cyclophosphamide; MTX = Methotrexate; AZA = Azathioprine; RTX = Rituximab; MMF = Mycophenolate mofetil; BAL = Bronchioalveolar lavage; RLL = Right lower lobe; RUL = Right upper lobe; ICU = Intensive care unit

The most common presenting symptoms were cough and dyspnea (Table 1). Fever was present in three (43%) patients. Laboratory investigations at time of PCP diagnosis revealed the following median values: lymphocytes 0.30 × 10⁹/L (range 0.03–2.10), creatinine 186 µmol/L (range 78–359), and lactate dehydrogenase 471 U/L (range 301–1,032). All cases received trimethoprim–sulfamethoxazole (TMP–SMX) for PCP treatment except one, who received clindamycin–primaquine. During admission, four (57%) required intensive care unit admission, and two (29%) died secondary to complications of PCP.

Individual prednisone dose preceding diagnosis of PCP was ≥20 mg/day or higher for 6 patients (Figure 1a-g). All patients with AAV were on at least one other immunosuppressive drug from time of diagnosis of vasculitis to development of PCP, with cyclophosphamide being the most common, with five of seven patients receiving it (Table 1). Mean and median duration of prednisone treatment at a dose of ≥20 mg/day before PCP diagnosis were 21.3 (SD 11.5) and 16 weeks, respectively. At the time of PCP diagnosis, none were receiving PCP prophylaxis. At some point before diagnosis, three patients had been on TMP–SMX, which was discontinued for the following reasons: transaminitis secondary to TMP–SMX (Figure 1a–1b) and query hepatotoxicity secondary to concomitant treatment for latent tuberculosis (Figure 1d). The remaining four cases were at no point on PCP prophylaxis; however, all received TMP–SMX for treatment and secondary prophylaxis with no reported adverse events, thus excluding allergy as a possible reason.

Figure 1:

Prednisone use before PCP diagnosis at time = 0, where shaded regions indicate TMP–SMX prophylaxis (each graph represents one individual case)

PCP = Pneumocystis jirovecci pneumonia; TMP–SMX = Trimethoprim–sulfamethoxazole

DISCUSSION

This study describes the clinical characteristics of seven patients with vasculitis who developed PCP between 2006 and 2017. Only 6% of all PCP cases at TOH during the study period were patients with vasculitis. This observation is likely a result of the known effectiveness of PCP prophylaxis in patients with AAV despite the use of immunosuppressive agents (6). Those who developed PCP in our case series were not on PCP prophylaxis at the time of diagnosis.

The most common presenting symptoms were cough and dyspnea. Unlike among HIV-infected patients, in whom the presence of fever is common, the use of glucorticosteroids in the treatment of vasculitis can inhibit increases in body temperature, making fever a less reliable symptom at presentation (7). This suggests that the presence of dyspnea or cough without fever in an immunocompromised patient with vasculitis should not preclude consideration of an infectious pneumonia and warrants a low threshold for definitive diagnostic testing.

Mortality from PCP was 29%, in keeping with estimates of 30%–60% among non–HIV-infected immunosuppressed patients compared with lower estimates of 10%–20% among HIV-infected patients (8). First-line treatment of TMP–SMX was initiated with six patients, with clindamycin–primaquine shown to be effective in the single patient for whom TMP–SMX was contraindicated.

We report that nearly all identified PCP cases with underlying immune-mediated vasculitides were on ≥20 mg/day of corticosteroids in combination with other immunosuppressive agents for at least 12 weeks before developing PCP. Cyclophosphamide was the most common concomitant immunosuppression used in our case series. Our findings are consistent with those of previous studies of PCP in vasculitis (Table 2). Most cases in the literature occurred among patients not receiving PCP prophylaxis while on significant immunosuppression regardless of time from diagnosis of vasculitis. Late PCP infections are reported and often follow a recent flare and escalation of therapy without re-evaluation of PCP prophylaxis (9). There is also increasing evidence that rituximab is an independent risk factor for PCP (10). These data suggest that PCP prophylaxis be initiated for patients on high-dose steroids, especially in combination with other immunosuppression.

Table 2:

Summary of reported cases of PCP in vasculitis

Author, study type, and study period	Population* and Incidence†	PCP prophylaxis	Time to PCP (median)	Immunosuppression (at time of PCP)	Mortality
Matraiah et al (9), retrospective, 10 y	GPA, n = 12 MPA, n = 2 EGPA, n = 2 Incidence 0.38	14/16 TMP–SMX at induction only	6 mo	15/16 CYC induction Maintenance: PDN + AZA, n = 5 MMF, n = 3 AZA + MMF, n = 3 AZA + MTX, n = 2 RTX, n = 2 AZA + MTX + MMF, n = 1	2/16 (12.5%)
Kermani et al (11), retrospective, 32 y	GCA, n = 7 Incidence not reported	None	3 mo	PDN, 50 (median)	2/7 (28.5%)
Berger et al (12), retrospective, 33 y	GCA, n = 4 Incidence 0.06	None	6.4 mo	PDN, 20–50 MTX, 4/4 cases	1/4 (25%)
Bligny et al (13), retrospective, 15 y	GPA, n = 12 Incidence 13	None	At induction (not specified)	CYC induction PDN (dose not specified)	5/12 (42%)
Assarød et al (14), retrospective, 10 y	GPA, n = 6 Incidence = 5.5	Not reported	Not reported	CYC PDN (dose not specified)	2/6 (33%)
Ward & Donald (15), retrospective statewide analysis, 12 y	GPA, n = 31 PAN, n = 21	Not reported	Not reported	Not reported	62.5% GPA 47.6% PAN
Godeau et al (16), retrospective, 8 y	GPA, n = 12	None	3 mo	CYC induction PDN, 0.7 mg/kg/day	Not reported

*Population of vasculitis with confirmed PCP

† Incidence as cases of PCP in total cases of vasculitis reported

PCP = Pneumocystis jirovecii pneumonia; GPA = Granulomatosis with polyangiitis; MPA = Microscopic polyangiitis; EGPA = Eosinophilic granulomatosis with polyangiitis; TMP–SMX = Co-trimoxazole or trimethoprim–sulfamethoxazole; CYC = Cyclophosphamide, PDN = Prednisone (in mg/day unless otherwise stated); AZA = Azathioprine; MMF = Mycophenolate mofetil; RTX = Rituximab; MTX = Methotrexate; GCA = Giant cell arteritis; PAN = Polyarteritis nodosa

This study has limitations. First, the incidence of PCP among patients with vasculitis not on PCP prophylaxis could not be determined because the total number of vasculitis patients over the same period is not known. Second, the limited sample size was not amenable to inferential statistics, allowing for only descriptive observations.

In conclusion, our results suggest that PCP is a severe and often fatal infection among patients with vasculitis not on PCP prophylaxis who are on corticosteroids in addition to other immunosuppressive therapies. The presentation and laboratory findings of PCP are non-specific, necessitating a high degree of suspicion for this infection among immunocompromised patients. There is a paucity of data to inform specifics regarding when to initiate and discontinue prophylaxis in PCP. PCP prophylaxis is effective and should be considered in patients with vasculitis who are on ≥20 mg/day of corticosteroids in combination with other immunosuppressants or any escalation in immunosuppression for relapse of vasculitis. Further studies are needed to evaluate the timing of prophylaxis to inform the creation of guidelines on PCP prophylaxis in the population with vasculitis.

Ethics Approval:

The manuscript was reviewed by The Ottawa Hospital Research Institute institutional review board.

Informed Consent:

N/A

Registry and the Registration No. of the Study/Trial:

N/A

Funding:

No funding was received for this work.

Disclosures:

J Cowan has received an investigator-initiated research grant from OctaPharma, served as an advisory board member for Takeda on an immunoglobulin product, and received honoraria for lectures and presentations on topics that were unrelated to this study. J Cowan also received financial support to attend the Annual European Hematology Association Meeting and was an editorial board member and medical leader for the Therapeutic Advances in Infectious Disease, Ontario Immunoglobulin Treatment Program. The other authors have nothing to disclose.

Peer Review:

This manuscript has been peer reviewed.

Animal Studies:

N/A

Data Availability:

The data underlying this article are available in the article; there are no additional unreported data.