Pneumocystis jirovecii (formerly known as Pneumocystis carinii) is a ubiquitous opportunistic fungus, which causes pneumonia [Pneumocystis pneumonia (PCP)]. It was first described in 1940s malnourished infants (2). In the 1980s, it was associated with HIV-infected patients with low CD4 counts (3). PCP continues to be an acquired immune deficiency syndrome-defining illness although the advent of antiretroviral therapy has resulted in incidence reduction (4). PCP has been recognized as a disease in children with primary cell-mediated immunodeficiency, patients receiving chemotherapy for hematological malignancies, solid organ, and bone marrow transplant recipients and in patients requiring immunosuppressants (5). Notably, the incidence in the immunosuppressed non-HIV population is increasing with the escalating use of immunosuppressive agents (4, 6, 7). Inflammatory autoimmune conditions such as IBD account for up to 20% of PCP in HIV-negative patients (8).

Pneumocystis jirovecii exposure appears to occur early in life and is often asymptomatic (9, 10). Reinfection of the immunosuppressed host through environmental or person-to-person transmission rather than reactivation from latency appears to be the major mode of acquisition in immunosuppressed patients (10). Effective host defenses against P. jirovecii are mediated by the innate, T-cell, and to a lesser extent, humoral immune responses. In immunosuppressed patients, the infection results in a dysfunctional immune response, composed of mononuclear cells, CD8 lymphocytes, and activated macrophages, which causes diffuse lung damage (11). Moreover, the inhaled Pneumocystis trophozoites inhibit epithelial repair processes within the alveoli resulting in severe lung damage (12).

Adult data show that unlike in HIV-infected individuals where presentation can be slow and insidious, non-HIV infected immunocompromised individuals can have an aggressive course often culminating in respiratory failure over several days (4, 13–15). PCP is characterized by a non-productive cough, fever, and dyspnea. On chest X-ray, diffuse bilateral interstitial pulmonary infiltrates are most common (4). In equivocal cases, high resolution chest computed tomography is more sensitive and commonly shows diffuse ground glass opacity (16).

Pneumocystis jirovecii cannot be cultured, therefore, definitive diagnosis is made by visualization of the organism’s cysts and trophozoites. This can be achieved using induced sputum, bronchioalveolar lavage specimens, or lung tissue stained with Grocott–Gomori’s methenamine silver or visualized using immunofluorescence. Unfortunately, sensitivity is <67% in the non-HIV infected patient due to low numbers of P. jirovecii (17). PCR testing of P. jirovecii nucleic acid in respiratory samples is more sensitive for diagnosis; however, discrimination between colonization and actual disease has been challenging. Quantitative PCR is thought to help differentiate between these two entities better than qualitative PCR with recent research focusing on oral washes as a less invasive diagnostic tool (18, 19).

First line treatment of PCP is IV TMP–SMX (15–20 mg/kg/24 h of the trimethoprim component in three divided doses) due to excellent tissue penetration, rapid response, and low cost (4). Adverse reactions include leukopenia, thrombocytopenia, and rashes including Stevens–Johnson syndrome (5). However, reactions appear less common in children (20). Alternative therapy includes clindamycin plus primaquine, pentamidine, and dapsone (21). Atovaquone has been used for mild disease as it is less effective than TMP–SMX, but better tolerated (22). Corticosteroids have been used as an adjuvant treatment in severe disease, reducing respiratory failure and the length of intensive care unit stay (23).

PCP mortality rates are between 20 and 60% in immunocompromised non-HIV infected individuals in contrast to rates of 10–20% in HIV-infected patients (4, 7, 13, 15, 24, 25). The more severe course is thought to be secondary to a more disseminated pulmonary inflammatory response and diagnostic delay (4, 26).

A retrospective cohort study determined that IBD patients were at an elevated risk of PCP compared to the general population with an increased relative risk (hazard ratio 2.96; 95% CI 1.75–4.29) but low absolute risk (0.03%) (26). The incidence of PCP in immunosuppressed IBD patients [thiopurines, methotrexate, calcineurin inhibitors, anti-tumor necrosis factor (anti-TNF) agents, or steroids] was 32/100,000 patient-years (PY) compared to 5.5/100,000 PY for non-immunosuppressed IBD patients (26). The risk appeared greater in CD compared with ulcerative colitis (UC) (26). In another population-based IBD cohort, double immunosuppression resulted in a higher risk of PCP than monotherapy (0.6/100 vs. 0.3/100 PY). There were few patients on triple therapy making risk analysis challenging (27).

There is only one published case of a child with IBD developing PCP. The 8-year-old CD patient, on infliximab monotherapy, developed PCP with concurrent disseminated histoplasmosis after 15 months therapy. His disease course was not severe. He did not require respiratory support and responded to TMP–SMX. He was maintained on budesonide with no PCP prophylaxis (28). PCP has been described in adult IBD patients on corticosteroids, calcineurin inhibitors, thiopurines, and anti-TNF agents (8, 26, 29–33).

We performed a literature review of PCP in IBD. A database search for studies on MEDLINE, EMBASE, and the Cochrane Controlled Trials Registry was performed. References of included articles were searched for further studies. Table 1 summarizes the details of published IBD studies reporting patients with PCP. Most sources were case reports [evidence level 5 (34)], and there was likely reporting bias toward immunomodulators, anti-TNF therapy, and patients who died. There were 92 reported cases of PCP in patients with IBD in the English literature, of which 56% were males and 58% had CD. The median age was 45 years [interquartile range (IQR) 30–68, range 8–78]. There was little documented information about CD4 counts at PCP diagnosis; however, 86% (12/14) of patients were lymphopenic. The case-fatality rate (CFR) was 23% (7/31) based on the reported outcome data. Where medication was documented, rates of mono, dual, triple, and quadruple immunosuppression therapy were 35% (18/52), 35% (18/52), 29% (15/52), and 2% (1/52), respectively. The numbers were too small to comment on the effect of incremental risk with increasing numbers of immunosuppressants. Corticosteroids, as mono, dual or triple therapy, were used as IBD treatment, in 88% (46/52) of patients who developed PCP. 42% (22/52) received thiopurines, 44% (23/52) took anti-TNF therapy, and 15% (8/52) used either cyclosporine or tacrolimus.

Corticosteroids are known to reduce CD4 lymphocytes, which predisposes to PCP development (51). Corticosteroids have emerged as a major contributor to PCP in the non-HIV immunosuppressed population and the risk is particularly increased at or above 16 mg of prednisolone (50–52). The median therapy duration prior to PCP was 8–12 weeks (50, 51). Interestingly, in some studies, the disease only became apparent when corticosteroids were tapered (24, 50, 51).

Thiopurines (AZA and 6-mercaptopurine) can inhibit cell-mediated immunity, which influences PCP development. In the literature, two IBD patients on thiopurine monotherapy developed PCP, 10 were on dual therapy, as was the case with our patient, and 10 were on triple therapy. The CFR in this group was 19% (3/16), where outcome data were available.

Cyclosporine works by inhibiting production of IL-2 by helper T-cells and by affecting T-cell, B-cell, neutrophil, and mast cell function. In the literature, 50% (4/8) of IBD patients with PCP died (32, 37, 38). Such cases have prompted a discussion regarding the role of prophylactic antibiotics in patients on cyclosporine; however, the limited cases and reporting bias needs to be considered (37, 53, 54).

Three patients in the IBD literature developed PCP while on methotrexate in combination with corticosteroid and anti-TNF therapy. Methotrexate has been implicated in the development of PCP in rheumatoid arthritis with 28 documented cases, 25% of whom died (55, 56).

Cytokines inhibited by anti-TNF agents are involved in the host response to PCP resulting in reduced PCP clearance. Moreover, anti-TNF therapy can lower CD4 counts making patients more susceptible to PCP (57). Review of the Food and Drug Administration Adverse Event Reporting System data between 1998 and 2003 identified 84 patients with PCP associated with infliximab, with a mean age of 55 years; 19% of patients had IBD. Concomitant immunosuppressive agents included immunomodulators (66%), corticosteroids (50%), and cyclosporine (5%). The CFR was 27% (33). PCP has been reported to occur 9–14 weeks after the first infliximab induction dose (33, 57, 58). Notably, Colombel et al. did not report any cases of PCP associated with 3,160 patients (in six global clinical trials) on adalimumab (59). However, there are two case reports of PCP in IBD patients on adalimumab (8, 40). In total, 162 cases of PCP are reported in the IBD and rheumatology literature associated with anti-TNF therapy, which includes pediatric patients. Of the 138 patients with outcome information available, 20% died. Unfortunately, studies do not always specify whether single, double, or triple immunosuppression was used (8, 26–28, 33, 40–49, 56, 58, 60, 61). Data on the incidence of PCP in patients on anti-TNF agents are largely based on rheumatoid arthritis studies and most are population based. The majority of studies report incidence rates <50 cases per 100,000 PY (62). Results are hampered by heterogeneity in the method of PCP diagnosis, moreover, discrimination between P. jirovecii colonization and actual disease can be challenging.

There are no published data on PCP associated with vedolizumab, ustekinumab, or other novel IBD treatments (63, 64). More long-term data are required to assess safety profiles of these agents.

The use of multiple immunosuppressive agents incrementally increases the risk of opportunistic infection in IBD; moreover, malnutrition and surgery can play a role, although this has not been specifically addressed in PCP (1, 27, 40, 51, 52, 62). As documented in our case, lymphopenia (especially CD4 count <300 cells/mm³) has been associated with increased risk of PCP in 60–95% of cases (5, 24, 50, 65). Notably, not all immunosuppressed patients will present with lymphopenia (24, 65).

Long et al. reviewed a case series of 38 IBD patients who developed PCP and showed that rates of hospitalization at some stage within 60 days of PCP were high (50%) (26). Moreover, patients had higher rates of comorbidities, principally, lung disease and diabetes mellitus, compared to the background population. Advanced age >65 years was found to be an additional risk (26).

A population-based cohort study in juvenile idiopathic arthritis patients reported an incidence of 7/100,000 PY (66). Multiple immunosuppressive agents and lymphopenia are important risk factors in rheumatic diseases (67). The risk of PCP in pediatric cancer depends on the malignancy type and chemotherapy category. Lymphoid malignancies have the highest risk with rates of 22–45% (68). The overall risk of PCP post solid organ transplant has been estimated to be 5–15% in the absence of PCP prophylaxis (5). Risk factors include malnutrition, previous cytomegalovirus infection, and underlying lung disease. Medications such as steroids, antilymphocyte agents, calcineurin inhibitors, and biological agents such as alemtuzumab (anti-CD52 monoclonal antibody) have also been implicated (5).

In a 2012 survey of PCP prophylaxis practice by gastroenterologists, 11% prescribed prophylaxis to their patients with IBD on combination therapy. Gastroenterologists were more likely to prescribe if they had previous practical experience of PCP, or practiced in an academic center (69).

The effects of PCP can be severe, however, prevention entails using drugs with adverse effects that may counterbalance the benefits, as occurred in our patient. The guidelines for PCP prophylaxis in HIV-infected patients have been universally adopted, but there is a lack of consensus on prophylaxis with TMP–SMX in IBD patients. A recent Cochrane meta-analysis of prophylactic treatment with TMP–SMX in non-HIV infected patients included 1,412 patients, of which 520 were children. The authors were unable to find published literature addressing prophylaxis in IBD. In patients with hematological cancers and transplant recipients, they reported an 85% reduction in PCP incidence with prophylaxis and PCP-related mortality reduced by 83% with few adverse events reported (20). No children in the included studies had a severe adverse event compared to 3.1% of adults suggesting a lower probability of harm in children.

There is insufficient evidence to recommend PCP prophylaxis for all IBD patients on immunosuppressive monotherapy; however, with increasing number of immunosuppressive agents, the risk of opportunistic infection is known to increase (1, 27, 62). Large studies addressing the specific incremental increased risk associated with multiple immunosuppressive agents especially triple therapy have not been done. This needs to be considered, as the use of multiple immunosuppressive agents are increasing. A recent study showed that approximately half of patients started on anti-TNF therapy were already on two immunosuppressive agents (40). Dietary therapies (EEN, partial enteral nutrition, and exclusion diets) in CD are increasingly being used and investigated and are attractive options to limit immunosuppression and improve nutrition in this population (70–72).

A study using simulation modeling to address cost and comparative effectiveness of PCP prophylaxis in CD concluded that at the present incidence, routine chemoprophylaxis was not cost effective but, based on limited data, it may be effective in triple immunosuppressive therapy (73). The 2014 European Crohn’s and Colitis opportunistic infection guidelines recommend PCP prophylaxis in IBD patients on triple immunosuppression including either a calcineurin inhibitor or anti-TNF therapy. This recommendation is based on expert gastroenterology and infectious disease opinion (74). Other practical approaches proposed for patients on high-dose steroids and multiple immunosuppressants include measurement of CD4 counts in those patients with a total lymphocyte count of <600 cells/mm³ as a means of identifying patients at risk of infections, although not all patients who develop PCP have lymphopenia, therefore, precluding its use as an isolated risk identification guide (24, 43, 65, 75). Evaluating the utility of measuring CD4 counts in IBD patients on multiple immunosuppressants should be considered.