Clinical, Diagnostic, and Treatment Disparities between HIV-Infected and Non-HIV-Infected Immunocompromised Patients with Pneumocystis jirovecii Pneumonia

A 48-year-old previously healthy man was admitted to our hospital with a 5-week medical history of increasing exercise-induced shortness of breath. He had lost 6 kg of weight within the last 2 months. During the past 3 weeks he had experienced night sweats. He reported increasing dry cough over 4 days prior to admission.

Upon admission his body mass index was 22.8 (180 cm, 74 kg). He was in a trained physical condition. Musculoskeletal system examination and neurological examination were without pathological findings. On the skin of his forehead there was an efflorescence compatible with seborrheic dermatitis. There was no jaundice, anemia, finger clubbing, or cyanosis present. On the neck, both axillas, and groins there were pea-sized palpable lymph nodes. He was febrile with a temperature of 39.5°C. Cottage cheese-like plaques were visible on the hard palate and pharynx, compatible with Candida stomatitis and pharyngitis. His heart rate was 92 beats/min and his blood pressure 120/70 mm Hg. His respiration rate at rest was 23 breaths/min. The breath sounds were vesicular over both lung fields. Arterial blood gas analysis showed a pO₂ of 62 mm Hg, a pCO₂ of 31 mm Hg, and a pH of 7.47 on room air. Oxygen saturation was 92% and dropped to 86% after 3 min of walking horizontally. Lung function test showed a reduced forced vital capacity of 64% with a nonobstructive forced expiratory volume in 1 s/forced vital capacity ratio of 96%, indicating a restrictive spirometry pattern with a reduced diffusing capacity of the lung of 35%.

Routine laboratory values were within normal limits, except for a serum lactate dehydrogenase (LDH) level of 338 U/L (normal < 252 U/L), a hemoglobin concentration of 12.1 g/dL (normal 13.5–17.5 g/dL), and a C-reactive protein level of 2.08 mg/dL (normal < 0.5 mg/dL). An HIV screening test was reactive with a positive HIV-1 Western blot test result. The HIV RNA concentration was 34,618 copies/mL. The number of circulating CD4+ T lymphocytes was 124/μL (7%) and the number of CD8+ T lymphocytes was 1,276/μL (67%), resulting in a CD4:CD8 ratio of 0.1 (normal 1.2–2.7).

Chest X-ray showed bilateral perihilar ground-glass opacities with interstitial thickening. There was no pleural effusion or cavitation visible. A computed tomography (CT) scan of the chest showed multifocal ground-glass opacities, compatible with the diagnosis of alveolitis (Fig. 1).

On bronchoscopy the visual aspect of the trachea and bronchial tree was without pathological findings. Bronchoalveolar lavage (BAL) demonstrated 16 × 10⁶ cells/100 mL (normal 3–12 × 10⁶ cells/100 mL), with a differential cell count of 13% lymphocytes (normal < 10%), 26% macrophages (normal > 85%), 60% neutrophils (normal < 3%), and 1% eosinophils (normal < 0.5%). Grocott-Gomori methenamine silver staining of the BAL fluid identified multiple spherical, oval asci/cysts 5–6 μm in diameter, compatible with the diagnosis of Pneumocystis jirovecii pneumonia (PCP) (Fig. 2). Treatment with trimethoprim/sulfamethoxazole (TMP/SMX) for 21 days as well as antiretroviral therapy (ART) was initiated. After 8 weeks the patient’s symptoms and radiological changes had resolved completely.

While PCP was rarely described before the HIV/AIDS epidemic, the incidence of PCP increased rapidly during the 1980s, with PCP occurring in 75% of individuals with AIDS, resulting in mortality rates up to 40% [1]. Given the fact that > 90% of PCP cases occurred in patients with CD4+ T lymphocyte counts < 200 cells/mm³, the infection became one of the main AIDS-defining illnesses [1]. Since then, the widespread use of PCP prophylaxis with TMP/SMX and early ART have led to a substantial decline in the PCP incidence among individuals with HIV infection. Studies in the early 2000s reported an incidence of PCP among HIV-infected individuals below 1 case per 100 person-years [2-8]. Given the trend towards earlier diagnosis of HIV infection and immediate ART independent of CD4+ T cell count, the incidence of PCP in HIV patients has likely decreased further during recent years [9-12].

Despite its decreasing incidence, PCP is still a serious health concern for people living with HIV/AIDS [13]. Since there is no national surveillance for PCP in the United States and other resource-rich countries, the exact number of cases is difficult to determine. While fewer HIV-infected individuals develop opportunistic infections in resource-rich settings overall, PCP is still one of the most common opportunistic infections in HIV patients in the United States, Canada, and Europe [2, 13]. It mainly affects individuals who are unaware of their HIV infection or who get diagnosed late (i.e., late presenters) or are not receiving ongoing care for HIV infection [14, 15]. Late HIV diagnosis as the major risk factor for PCP is significantly more frequent among populations suffering from healthcare disparities [16]. PCP remains also very common as an opportunistic infection among people living with HIV/AIDS in resource-limited countries (see PCP from a Global Health Perspective with Emphasis on Resource-Limited Settings) [17].

PCP is increasingly diagnosed in non-HIV-infected immunocompromised patients, indicating new challenges for the diagnosis, treatment, and prophylaxis in a larger susceptible population [18-20]. In contrast to HIV-infected patients, there is evidence for a more acute onset of symptoms, faster progression of disease, poorer outcome, higher mortality, and higher risk of coinfections [9, 21-24].

Posttransplant Patients. Both patients undergoing hematopoietic stem cell transplantation (HSCT) and patients undergoing solid organ transplantation (SOT) are at risk of PCP (5–15% of cases), which is higher after allogenic versus autologous HSCT and after heart and/or lung SOT versus for instance renal transplantation [19, 20, 25, 26].

Hematooncologic Patients. Particularly patients with leukemia (such as acute lymphatic leukemia) and lymphoproliferative disorders (such as chronic lymphatic leukemia and non-Hodgkin lymphomas) have a substantial risk of developing PCP. Certain chemotherapeutic agents are more predisposing to PCP, including corticosteroids, cyclophosphamide, methotrexate, vincristine, cytarabine, fludarabine, temozolomide, rituximab, alemtuzumab, ibrutinib, and idelalisib [27-30].

Patients under Immunosuppressive Drugs due to Autoimmune Diseases. Inflammatory diseases, their complications, and immunosuppressive regimens contribute to the risk of PCP. Up to 20% of PCP cases occur in patients with inflammatory diseases, the incidence being especially high in cases of polyarteritis nodosa, granulomatosis with polyangiitis, dermatomyositis/polymyositis, and interstitial lung diseases in patients with rheumatoid arthritis [31, 32]. Furthermore, it depends on the administered immunosuppressant, with following drugs most frequently mentioned: prolonged medium- to high-dose glucocorticoids, cyclophosphamide, rituximab, alemtuzumab, and tumor necrosis factor alpha antagonists [33, 34].

The classic triad of PCP symptoms in AIDS patients includes (1) subacute onset of exertional dyspnea, (2) dry and nonproductive cough, and (3) fever or subfebrile temperatures (Table 1). The subacute course over several days and even weeks often allows differentiation from bacterial pneumonia. Oral thrush and substantial weight loss are also frequently seen in AIDS patients with PCP, as demonstrated in our patient. However, despite the subacute course, deterioration may occur rapidly.

A diagnostic hallmark is respiratory insufficiency, which should be confirmed by arterial blood gas analysis. LDH is often elevated and may have limited use as a predictive parameter for the course of disease. However, a high LDH level is an unfavorable sign and may reflect the severity of the PCP. In contrast, C-reactive protein is often normal, provided there are no other concurrent infections.

Numerous studies have compared the clinical features and the outcome of HIV-infected and non-HIV-infected cases. These comparisons should be considered with caution, as the clinical and therapeutic background of PCP in non-HIV-infected patients with variable immunocompromising disorders is very diverse. In large retrospective cohort studies of PCP cases, however, non-HIV-infected patients were less symptomatic at diagnosis than AIDS patients [19]. On the other hand, the progression of PCP seems to be faster and higher mortality rates have been reported in non-HIV-infected immunocompromised patient compared to HIV-infected patients [35]. The latter may be due in part to the delay of the initiation of a specific anti-infective therapy [9, 19].

X-ray of the chest may show characteristic findings with a perihilar interstitial infiltrate (Fig. 3; Table 1). In non-HIV-infected immunocompromised patients, the initial chest X-ray may be normal due to the acute onset of symptoms and more rapid progression, but generally the features of the chest X-ray are very similar to those found in HIV-infected patients. Sometimes radiological changes are discreet and the X-ray is often inconclusive or does not show lung pathologies seen on CT scan.

High-resolution CT of the chest should be preferred if available, offering a precise description of the radiological pattern with differentiation to other pulmonary diseases [36]. Ground-glass opacities, often with a central distribution as demonstrated by our patient (Fig. 1), are a common radiological pattern of PCP on CT scans, but the distribution can vary from a mosaic pattern to a more diffuse pattern. Upper lobe predominance has been described [36]. Other less common manifestations include nodules, consolidations, or thin-walled cystic lesions, sometimes classified as pneumatocele formation, which are usually associated with the presence of AIDS, whereas non-HIV-infected immunocompromised patients usually present with widespread ground-glass opacities [37-39]. Cystic lesions are often small, bilateral, and multiple, but larger pulmonary cystic lesions with the risk of developing a spontaneous pneumothorax have also been reported (Fig. 4) [40]. The name pneumocystis refers to the microscopic image of the fungus and not to the radiological pattern of the disease. In a prospective study including 30 AIDS patients with respiratory symptoms and an uncertain X-ray, the sensitivity and specificity of high-resolution CT was 100 and 83%, with a positive and negative predictive value of 90.5 and 100%, respectively [36].

In ill patients with clinical suspicion of PCP, empiric treatment should be started immediately. There is no need to delay treatment initiation for diagnostic assessment since cysts persist for several days in respiratory material, even under PCP-specific treatment [41, 42].

If possible, bronchoscopy with BAL should be performed, because analysis of BAL fluid provides not only a higher diagnostic yield, but also allows exclusion of alternative diagnoses or coinfections, e.g., tuberculosis (TB), histoplasmosis, cytomegalovirus (CMV), etc. Induced sputum with hypertonic saline is a less invasive alternative, but has a lower sensitivity of 55–90% [41, 43].

Active pulmonary TB can be excluded either by microscopic smear, by polymerase chain reaction (PCR) (e.g., GenXpert MTB/rifampicin assay; Cepheid Inc., Sunnyvale, CA, USA), and/or by culture from respiratory samples. For most systemic endemic mycoses causing pulmonary disease including histoplasmosis, culture or histology from BAL fluid or lung tissue is the method of choice (Salzer et al., Respiration 2018, accepted for publication). To exclude CMV pneumonitis, histology from lung tissue may demonstrate characteristic inclusion bodies, and BAL fluid can be used for CMV culture or for viral load testing using quantitative PCR (qPCR) assays.

P. jirovecii does not grow on culture media in vitro. Therefore, PCP diagnosis historically relied on the visualization of cysts or trophic forms from respiratory material (Table 2). Direct or indirect immunofluorescence assay (mainly anticyst antibodies) is the most sensitive microscopic method (Fig. 5) and superior to conventional staining methods (e.g., Grocott-Gomori methenamine silver staining) (Fig. 2). The sensitivity of microscopy is lower in non-HIV-infected compared to HIV-infected patients due to a lower fungal load [42, 44].

PCR is an important additional diagnostic method, especially in non-HIV-infected patients, and may add an additional 7% diagnostic yield compared to staining methods alone [45]. Real-time qPCR should be preferred as the only PCR method compatible with diagnosis and the Minimum Information for Publication of Quantitative Real-Time PCR Experiments guidelines, but evidence on the validation of clinical cutoffs is limited [45, 46]. This method enables the detection of very low fungal loads, which can be considered as colonization.

A highly sensitive diagnostic method is the serum (1,3)-β-D-glucan assay characterized by its high negative predictive value that makes a PCP infection unlikely in patients with a negative test result [47]. However, it should not be used as a single diagnostic test for PCP, and positive results should trigger a bronchoscopy with BAL [48, 49]. A limitation is that (1,3)-β-D-glucan testing is frequently not available (at least with time to results of < 24 h). However, it is rather easy to establish the test on a coagulation automate and do single-sample testing [50]. Diagnostic guidelines commonly recommend combining different methods, e.g., microscopy (immunofluorescence and conventional staining) and qPCR [42]. When analyzed in parallel with qPCR fungal load, a correlation between both markers is observable, suggesting that circulating (1,3)-β-D-glucan levels reflect the fungal load in the lungs [51].

TMP/SMX is the treatment of choice for PCP in adults and children, being as effective as parenteral pentamidine and more effective than other regimens. Treatment is generally administered over 21 days (Table 3) [52]. In patients with mild to moderate disease, oral outpatient therapy with TMP/SMX may be considered. The TMP/SMX dosage needs to be adjusted to renal function (Table 4). TMP/SMX is usually also effective in patients who have PCP despite TMP/SMX prophylaxis [42].

Clinical deterioration may occur during the first 4–8 days of treatment. It is important that this deterioration has been frequently seen in the era before the intro duction of highly active ART. Deterioration usually neither indicates immune reconstitution inflammatory syndrome (IRIS) (very rare) nor treatment failure. It is probably due to an inflammatory response caused by antimicrobial-induced lysis of organisms in the lung (that is why steroids may be helpful). Patients can deteriorate rapidly during treatment, so close monitoring of respiratory rate and arterial oxygenation are essential. Nevertheless, physicians must be aware that clinical worsening rebound phenomena after initial improvements have also been observed in HIV-coinfected patients treated with ART that have been attributed to IRIS [53].

Alternative therapeutic regimens for mild to moderate and moderate to severe disease are depicted in Table 3. Dosages for renal impairment and dosage recommendations for children, as well as the most common adverse events and costs of the therapeutic regimens, are displayed in Table 4. All patients with documented or suspected PCP and moderate to severe disease should receive adjunctive corticosteroids as early as possible and always within 72 h after starting specific PCP therapy (Table 3). However, this has only been validated in HIV-infected patients and the classification of mild, moderate, and severe disease is not precisely defined. Early predictors of mortality may be useful to asses the risk of death as shown for HIV-infected patients including older age, second or third episode of PCP, low hemoglobin level, low PaO₂ breathing room air at admission, pulmonary Kaposi sarcoma, and medical comorbidities [54].

In newly diagnosed HIV-infected patients with PCP, ART should be initiated as soon as possible. In a large multicenter trial (ACTG A5164), a total of 282 subjects with an acute opportunistic infection (69% PCP) were randomized to initiate ART immediately or after treatment of the opportunistic infection [55]. Immediate ART resulted in less AIDS progression and deaths, with no increase in adverse events or loss of virologic response compared to deferred ART. This study provided clear arguments for immediate initiation of ART when PCP is diagnosed.

The risk of a PCP-associated IRIS is low. Given the rapid decay of plasma viremia seen with integrase strand transfer inhibitors (INSTIs), the low prevalence of transmitted INSTI resistance, and the favorable safety profile, INSTI-based regimens seem to be a reasonable option. Antiretroviral drugs with a potential allergenic potential and/or high risk of resistance (non-nucleoside reverse transcriptase inhibitors, abacavir) should be avoided in patients with acute PCP.

During therapy of acute PCP, it is crucial to reflect all comedications due to many possible drug-drug interactions. As the treatment of choice in patients with acute PCP according to national and international guidelines (see above) is TMP/SMX, methotrexate should be avoided because of potentially increasing adverse drug effects. If possible, immunocompromising treatment should be reduced, paused, or stopped, but this should be an individual approach (depending for instance on the activity of the underlying disease).

Furthermore, it remains unclear whether adjunctive therapy with glucocorticosteroids in non-HIV-infected patients (e.g., in the intensive care unit setting) leads to any benefit for this subgroup [20, 56, 57]. If patients are already treated with glucocorticosteroids for another reason, there should not be an abrupt discontinuation, though an individual adjustment seems mandatory.

Several guidelines give specific recommendations for prophylaxis in HIV-infected patients, and there are recommendations available for hematooncologic and posttransplant patients as well as for patients with anti-neutrophil cytoplasmic antibody-associated vasculitis receiving cyclophosphamide (Table 5) [30, 32, 58]. Otherwise, there is a lack of consensus for otherwise immunocompromised patients [20]. Most evidence exists for regimens containing TMP/SMX [59]. Alternative prophylactic options (e.g., dapsone, atovaquone, or pentamidine) lack evidence for these patients (Table 5) [58]. One of the major challenges is the identification of patients susceptible to P. jirovecii infection at an earlier stage of risk to start proper prophylaxis [19]. Monitoring CD4+ T lymphocytes, as in HIV-infected patients, appears to be insufficient to predict the risk of PCP [30].

Along with the high burden of HIV infection, particularly in sub-Saharan Africa, PCP infection contributes significantly to HIV-related morbidity and mortality in children and adults. The widespread introduction of PCP prophylaxis as a programmatic intervention as well as the significant rollout of ART has reduced the incidence of PCP infections [60]. Although detailed data are scarce, reports about PCP are available from all continents [61]. In a recent meta-analysis, Wasserman et al. [60] describe a case fatality of 18% for HIV-infected patients with PCP in sub-Saharan Africa.

The diagnosis of PCP is mostly a presumptive diagnosis in resource-limited settings. Clinical signs and symptoms, particularly dyspnea, fever, dry cough, and hypoxia in absence of an alternative diagnosis in an immunocompromised patient with a CD4+ T lymphocyte count < 200 cells/mL and concurrent radiological changes lead to the initiation of PCP treatment [62]. Very limited availability of diagnostic stains and PCP-specific PCR only infrequently allow a definite diagnosis. Bronchoscopy, bronchial wash, BAL, or transbronchial biopsy, which increase the diagnostic yield over sputum diagnosis, are often not available [60]. A high level of awareness and clinical suspicion is required. A South African autopsy study documented that PCP was not recognized prior to death in 89% of PCP cases [63]. Due to the high incidence of TB in many resource-limited settings, patients are treated with anti-TB drugs instead. No data can be found on patients requiring invasive or noninvasive ventilation in resource-limited settings, but limited access to such facilities in many settings contributes to higher mortality rates.

Increased LDH levels support the diagnosis and the test is often available. Rising LDH during treatment is associated with poor prognosis [64]. (1,3)-β-D-glucan is a helpful test to support the diagnosis in absence of a definite diagnosis in HIV-infected and non-HIV-infected patients, but is not widely available [48]. Very promising reports about the S-adenosylmethionine serum level as a diagnostic marker for PCP could not be independently validated [65, 66]. A recent approach used real-time qPCR in oral wash to detect PCP tried to establish a cutoff value for infection versus colonization, but this approach needs further studies [67].

TB is an important differential diagnosis for PCP, but coinfections should also be considered [68]. One autopsy study in South African miners describes cryptococcal pneumonia and bacterial pneumonia as the most frequent coinfections with PCP [63].

Prophylactic treatment with TMP/SMX also reduces the incidence of toxoplasmosis, malaria, and bacterial pneumonias. Particularly in HIV-infected children aged < 5 years, the World Health Organization recommends a continuous PCP prophylaxis, while in areas with a high incidence of malaria and bacterial pneumonias, this is also an option for adults. In other settings, the prophylaxis can be stopped if the CD4+ T lymphocyte count is > 400 cells/µL for more than 3 months or if the viral load is undetectable and the CD4+ T lymphocyte count is 100–200 µL for more than 3 months [69].

PCP is presently at a transitional stage with substantial epidemiologic chances. The widespread use of ART has led to an important decline in the PCP incidence in HIV-infected patients worldwide, while the medical progress leads to a growing number of non-HIV-infected immunocompromised patients susceptible for PCP infection. Therefore, physicians should be aware that PCP is an important infectious complication in hematooncologic and posttransplant patients, including SOT or allogenic and autologous HSCT, and in patients under immunosuppressive drugs due to autoimmune diseases. Considering the rapid course of the disease in non-HIV-infected immunocompromised patients, early diagnosis and treatment of acute PCP infection is crucial and can decrease morbidity and mortality. However, we also would like to draw attention to PCP prophylaxis, which sometimes seems to be disregarded in these heterogeneous patient cohorts, although it is effective and helps to combat rising trends.

In case of HIV infection, especially in late presenters who are unaware of their HIV infection and who often present with very low CD4+ T lymphocyte counts, there is a high risk of developing PCP. Community-based screening methods, partner services, and simple risk scores can be effective tools to find HIV-unaware individuals [70, 71]. PCP prophylaxis is of outmost importance and should be initiated in HIV-infected patients with CD4+ T lymphocyte counts < 200/µL.

It remains unclear why PCP varies between HIV-infected patients and non-HIV-infected immunocompromised patients, but physicians should be aware of the clinical, diagnostic, and management disparities of acute PCP infection in these heterogeneous patient cohorts.

We thank the patients for giving permission to publish the photographs, and Jessica Hofmeister and Franziska Daduna for the Grocott-Gomori methenamine silver staining of the BAL fluid.

The authors declare that no competing interests exist. All authors have submitted the ICMJE Form for Disclosure of Potential Conflict of Interest.