Epidemiological and clinical characteristics of immunocompromised patients infected with Pneumocystis jirovecii in a twelve-year retrospective study from Norway

Our study was based on data from St. Olavs hospital, Trondheim University Hospital, which is the only tertiary referral hospital in Central Norway. The health region offers services to approximately 700,000 inhabitants representative of the national population [8, 9]. Until 2017, St. Olavs hospital had the only microbiology laboratory conducting P. jirovecii diagnostics in the region. All patients from central Norway with P. jirovecii detected in one or more respiratory samples by PCR in St. Olavs hospital between 2006 and 2017 were identified and linked to their respective medical records. Only primary episodes were included. All subjects 16 years or older at the time of testing were eligible. Alive patients were included on the basis of informed consent in 2018. There were no minors among these. The need for consent from next of kin or legal guardian of deceased patients was waived by the ethics committee.

We retrospectively reviewed the medical records of the study population and extracted comprehensive epidemiological, laboratory and clinical data. The software Epi Info™ (version 7.2.2.6; Centers for Disease Control and Prevention, Atlanta, GA, USA) was used to record patient data. The number and severity of combined comorbid conditions were assessed according to the Charlson weighted comorbidity index [10]. Corticosteroid exposure pattern 60 days preceding presentation was categorized as daily, intermittent or none. In case of ongoing corticosteroid intake on the date of P. jirovecii detection, the daily dose was converted into the equivalent in methylprednisolone expressed as milligrams per day and the median among users was calculated. Antimicrobial treatments administered after the detection of P. jiroveci, regardless of etiological indication, were also registered. Treatment and documented complications occurring in association with hospitalization were also recorded. Date of death was ascertained through linkage with the Norwegian Population Register through the end of June 2018 for sufficient follow-up.

The PCR analysis was performed as an in-house real-time PCR targeting the beta-tubulin gene of P. jirovecii, as previously described [11], with some modifications. PCR reagents and instruments used varied through the study period, but all changes were validated to ensure equal quality. The laboratory participated in a Pneumocystis jirovecii pneumonia (PCP) DNA EQA Programme (QCMD) from 2012. Semiquantitative estimation of fungal loads was performed on positive samples and results were reported with cycle threshold (C_T) values. C_T values are defined as the replicated number at which the fluorescence generated within a reaction crosses the fluorescens threshold line [12]. Accordingly, a low C_T value corresponds to a high fungal burden and vice versa. Microscopy (direct immunofluorescence (DIF) was performed with MONOFLUO Pneumocystis jirovecii IFA Test Kit #32515 (Bio-Rad). The assay was in use at the laboratory until 2017, mainly on samples resulting positive by PCR whenever positive controls were available. To discriminate cases of PCP (PCP⁺) from colonization (PCP⁻), we applied retrospective case-criteria in line with the European Conference on Infections in Leukaemia (ECIL) guidelines [13] with the available data. According to previous studies, we considered that C_T values above 35 corresponded to colonization and not overt PCP [12, 14] regardless of the host’s HIV status. Thus, the criteria for PCP⁺ among our PCR-positive cohort were i) positive DIF and/or ii) C_T value below 36. Patients with C_T values above 35 and negative or missing DIF result were considered colonized with P. jirovecii (i.e., PCP⁻). Patients with missing C_T value and negative or missing DIF result were classified as “undetermined PCP-status”.

To estimate regional incidence rates, we accessed the online databank of Statistic Norway to retrieve the total number of people 16 years or older living in Central Norway during the study period [15]. These counts represented the denominators in our yearly incidence rate estimates. PCR detection of P. jirovecii was introduced at St. Olavs hospital, our referral laboratory, in late 2006. Thus, estimates were calculated for 2007 to 2017. In 2017, Molde hospital, a local hospital in the health region, established PCR-testing for P. jirovecii too. For completeness, individuals with a positive result at the laboratory in Molde hospital in 2017 were included in the regional incidence estimates for that year.

Continuous quantitative variables are presented as medians with second (q₁) and third (q₃) quartiles or means with standard deviation (+ SD). Discrete variables are given as absolute numbers (percentages). Incidence rate estimates are given with 95% confidence intervals. All analyses were performed using Microsoft Excel (version 16.4; Microsoft Corporation, Redmond, WA, USA) or STATA/MP (version 15.1; College Station, TX, USA).

A final 297 patients (117 F, 180 M) from Central Norway whose samples tested positive for P. jirovecii by PCR in the microbiology laboratory of St. Olavs hospital, were included in the study cohort (Fig. 1). The median patient age was 66 years and a majority (60.6%) of the patients were male. Each patient was classified with respect to their principal immunosuppressive condition associated with P. jirovecii and PCP development (Table 1). Hematological malignancies were the major predisposing conditions, present in more than one third of the patients (36.0%), with non-Hodgkin lymphomas appearing most frequently of these (51.4%). The second largest subpopulation was constituted by patients with solid tumors (25.3%). Therein, lungs including pleural membranes were the most common origin of the primary tumor (36.0%).

Behind malignancies, various non-HIV conditions appeared with decreasing frequencies, including immunological disorders (15.5%), solid organ transplantation (12.5%) and chronic lung disease (6.1%).

In our cohort, only a minority of seven patients (2.4%) presented with PCP in the context of HIV-infection. One patient from South-East Asia died pre-hospitally and resulted HIV positive in the referral hospital post-mortem. Four patients were unaware of their HIV-status and were naïve to anti-retroviral therapy, while the remaining two were not adherent to their anti-retroviral regimen.

Previous known comorbidities were present in 71.4% of patients, with hypertension being the most prevalent (Table 1). Moreover, the Charlson comorbidity index was skewed towards higher values, indicating an old study population and high degree of comorbidities overall.

Nearly all of the non-HIV study subjects had received immunosuppressants or chemotherapy before assessment for PCP. Ongoing drug regimens prescribed in vicinity to presentation were registered and categorized (Table 2).

As expected, the regimens reflected the underlying conditions, namely the etiology for iatrogenic immunosuppression, with chemotherapy with adjuvant steroids for hematological malignancy, being the most common category (22.6%).

Notably, 72.1% had been exposed to systemic corticosteroids in the 2 months preceding evaluation for PCP with a median dose in methylprednisolone of 8 (q₁-q₃ 4–20) milligram per day among users the day of P. jirovecii detection (n = 146). Given the high prevalence of systemic corticosteroid usage, we went on to investigating the indications for prescription among the exposed (n = 214), though some patients had multiple indications. Here immunosuppression for immunological disorders and graft rejection prophylaxis had the highest occurrence (46.3%), followed by systemic corticosteroids as chemotherapeutic agents (35.0%), and other oncological indications combined (31.3%), indicating widespread application in treatment of cancer patients in general.

A subpopulation of 49 patients (16.5%), including the seven HIV-positive patients, were not being prescribed immunosuppressant or receiving chemotherapy at presentation. However, 29 of the non-HIV patients in this group (69.0%) had received iatrogenic immunosuppression, chemotherapy or both the last 5 years.

Only three patients (1.0%) were receiving primary PCP-prophylaxis at presentation.

All but six patients presented with at least one cardinal symptom of pneumonia; cough, dyspnea, or fever prior to detection of P. jirovecii in a respiratory sample. The remaining patients reported reduced general condition, had abnormal findings on their physical or radiographic examinations. Clinical characteristics are summarized in Table 3.

A selection of objective manifestations and laboratory results are also presented though they were not retrievable for all patients. Decreased oxygen saturation (< 95%) was a common baseline finding, documented in 215 (72.4%) patients on presentation. Differential blood counts were incomplete overall, but therein lymphopenia (< 1.0 ×  10⁹ cells/L) dominated (108 of 152 patients; 71.1%) whilst severe neutropenia (< 0.5 × 10⁹ cells/L) appeared infrequently (6 of 224 patients; 2.7%).

Plain chest radiography was performed in 254 cases, and abnormalities were noted in 219 (86.2%), with interstitial (nodular, linear or patchy) opacities being the most frequent features. Among 247 patients undergoing thoracic CT, 242 (98.0%) manifested abnormalities. Ground glass opacities were present in 73.7% of the cases, followed by thickening of interstitial septa (26.7%), both suggestive signs of PCP, but not pathognomonic.

A majority of the patients within our cohort underwent bronchoalveolar lavage for microbiological diagnostics (n = 234, 78.8%), followed by sputum (n = 44, 14.8%), induced sputum (n = 9, 3.0%), tracheal aspiration (n = 5, 1.7%), biopsy upon autopsy (n = 2, 0.7%), nasopharyngeal sampling (n = 2, 0.7%) and lastly transbronchial biopsy (n = 1, 0.3%). DIF microscopy for P. jirovecii was performed on respiratory samples from 118 patients. The examinations resulted positive in 54 of these (45.8%). C_T values were retrievable for 243 patients irrespectively of patient characteristics, mainly from analysis of BAL-fluid samples (n = 192, 79.0%) Table S4 (Additional file 1). Based on the results of the microbiological analyses, 140 patients (47.1%), five of whom were HIV-positive, were diagnosed with PCP (PCP⁺), whereas 116 patients (39.1%) were presumed colonized (PCP⁻) (Figure S1 (Additional file 1)). Epidemiological and clinical characteristics and premorbid iatrogenic exposures of these are summarized in Table S1 and Table S2, respectively (Additional file 1). PCP⁺-patients were generally comparable to the overall population in terms of demographics and predisposition. A tendency of more cardinal symptoms, hypoxia, low lymphocyte counts, elevated lactate dehydrogenase levels, and radiological remarks was noted. The yearly distribution of the three patient categories (PCP⁺, PCP⁻ and “undetermined PCP-status”) is depicted in Fig. 2. There was only one case of PCP in 2006, but the ensuing years saw an increase.

From examining the predisposition and clinical characteristics associated with P. jirovecii-detection, we went on to investigating the outcome (Table 3). Overall, anti-PCP treatment was instituted to 87.9% of the patients. Almost a third (29.6%) required treatment in an ICU and the same proportion received ventilation support (non-invasive and/or invasive). One hundred twenty-one patients (40.7%) experienced at least one complication, primarily respiratory failure or ARDS. Overall, in-hospital mortality was 21.5%, occurring in 64 patients. Cumulative all-cause 30-, 90- and 180-mortality rates for the study population were 20.2, 33.0 and 39.1% respectively. Accounting for surviving non-participants the rates were lowered resulting in an in-hospital mortality of 17.7% and 30-, 90- and 180-day mortality rates of 16.6, 26.8 and 32.0%, respectively. The clinical course of PCP⁺ patients was broadly similar to the population overall. However, a greater proportion received anti-PCP treatment, intensive care, and ventilation support. Moreover, complication and mortality rates were slightly higher (Table S3, Additional file 1).

Our referral laboratory reported upward trends in molecular testing for P. jirovecii during the study period since the introduction of PCR in late 2006. A total of 1790 respiratory samples were referred for PCR analysis; 79 in 2007 compared to 259 in 2017. Accordingly, there was a 3.3-fold increase in analyses from 2007 to 2017. This was accompanied by a 1.8-fold increase in the incidence of samples with a positive PCR result; from 25 in 2007 to 46 in 2017 (Fig. 3). However, the proportion of positive samples remained more or less stable with a mean of 20.8% (SD 4.7) per year. All cases detected within Central Norway gave rise to regional incidence estimates. There were 5.0 cases per 100,000 person years in 2007 compared to 10.8 cases per 100,000 person years in 2017, with an increasing trend during this time interval (Fig. 4).