In the first systematic evaluation of P. jirovecii in Norway we observed an apparent increasing incidence of PCP from 2007 to 2017. The vast majority were constituted by patients with other predispositions than HIV-infection, such as haematological and solid cancers, and immunosuppression in the form of corticosteroids in monotherapy or in combination with chemotherapy. Our research confirms the non-specific, thus challenging, clinical presentation of patients with suspected PCP and the association between P. jirovecii and high risk of in-hospital mortality.
Several studies have reported upward trends of PCP occurring in non-HIV patients [16,17,18,19,20], including one from Denmark [21]. In contrast, a study from Sweden did not register a rise, in spite of an increasing number of cytotoxic treatments, but that study ended in 2011 [22] which is before we saw a clear increase in our study. The authors proposed a more widespread administration of prophylaxis to patients at risk as the reason for this opposing trend [22]. In our study, only three patients were receiving primary prophylaxis at presentation. However, since only subjects who tested positive for P. jirovecii were included in our study population, a selection occurred. As a result, patients receiving prophylaxis without developing PCP, or without undergoing testing during the study period were not included. Nonetheless, our report reveals a gap between patients receiving adequate prophylaxis and those at risk. Concomitantly, a recent Cochrane review showed that trimethoprim/sulfamethoxazole was highly effective in preventing PCP in non-HIV immunocompromised patients with an 85% incidence reduction (95% CI 38 to 96%) and a number needed to treat of 19 patients for PCP prevention [23]. PCP-mortality was also reduced by 83% (95% CI 6 to 97%) without an increase in adverse events [23].
The apparent rise in the PCP-incidence in patients with secondary immunodeficiencies other than AIDS occurs with a concurrent escalation in the administration of immunosuppressants and chemotherapy regimens [1, 24]. Iatrogenic immunosuppression, as well as the impairing effects of the underlying disease itself, are probable explanations for patients developing PCP in this context [25]. Moreover, we report a high prevalence of non-communicable comorbidities which may contribute to the “net state of immunosuppression” [26]. Altogether, this ageing population, cumulating endogenous and exogenous risk factors, may provide favorable conditions for P. jirovecii to re-emerge as an important opportunistic pathogen. Nevertheless, since the proportion of positive PCR results remained stable, it is uncertain whether the observations from our region reflects an actual increase in the number of people infected by the fungus, a changed clinical practice, or increased awareness of PCP, namely detection bias.
Reduction of CD4+ T cells caused by iatrogenic immunosuppression is the most significant risk factor regarding developing PCP in non-HIV immunocompromised individuals [25]. Systemic corticosteroids are hazardous to lymphocyte proliferation and kinetics, especially in high doses [5]. For the grand majority of our study population, preceding exposure to systemic corticosteroids was a common denominator. We report a wide spectrum of indications, exposure patterns and doses at the time of presentation, as well as a diversity in the co-administered chemotherapy and immunosuppressants. Exposure to systemic corticosteroids preceding development of PCP in heterogeneous non-HIV populations has already been described in several studies [19,20,21, 27,28,29,30,31,32,33,34,35,36,37,38]. Even patients receiving systemic corticosteroids in tapering doses are at risk [21]. Moreover, patients with miscellaneous conditions not previously associated to PCP development per se, may develop PCP due to systemic corticosteroids exposure [39]. This was presumably the case for one of our PCP+-patients receiving such treatment for statin-induced myositis.
In spite of the widespread use of corticosteroids and their lymphocytotoxic effects, lymphocyte counts were only documented in about half of the patient records (51.2%). Lymphopenia was present in the majority of these (72.1%), and even more prevalent among PCP+-patients (93.2%). Neutrophil counts, on the other hand, were present in almost all the records. While neutropenic patients occasionally contract PCP, they do not appear to be unproportionally predisposed to PCP [3], though the risk may depend on the intensity and duration of neutropenic states [40]. Perhaps the missing data in our study, namely the incomplete lymphocyte counts, reflect a certain unawareness and unwariness regarding the impairing effects of immunosuppressants on other cell lines than neutrophils. Raised awareness regarding risk factors would probably lead to more patients receiving primary prophylaxis as well as prompter diagnosis in the case of infection. In fact, early treatment is crucial for the outcome since there appears to be a positive association between treatment delay and mortality [33, 41]. Non-HIV patients seem more susceptible to diagnostic delays in spite of more fulminant onset of symptoms [21, 33, 41].
Regarding outcomes, the in-hospital mortality observed in our study is in the lower range, also among the patients retrospectively classified as PCP+. In comparison, it ranges from 15 to 49% for patients without HIV [17, 19, 27, 30, 32, 34,35,36,37, 42,43,44,45], and increases severely above 50% when ICU admission is required for respiratory failure [29, 46,47,48]. The differences in mortality may be due to heterogeneity in inclusion criteria in terms of underlying diseases, respiratory samples and diagnostic techniques. A recent meta-analysis reported a pooled overall in-hospital mortality of 30% for patients without HIV [49]. The prognosis of patients with HIV-infection is reportedly better, with mortality ranging from 10 to 20% during the initial infection, but it increases considerably with the need for invasive respiratory support in this population too [3]. In our study, five out of seven patients diagnosed with PCP in the context of AIDS died, resulting in an in-hospital mortality of 71.4%. This sample is too small to draw any conclusions or comparisons but indicates that PCP in HIV-patients is still a serious and potential life-threatening diagnosis, even in an industrialized country like Norway. Indeed, all the HIV-positive individuals fell outside of UNAIDS’ 90–90-90-treatment target for 2020 [50] in spite of high availability of anti-retroviral treatment.
With respect to the distribution of immunocompromising conditions, our cohort is broadly comparable to other reports [27, 30, 32]. In spite of the seemingly increasing incidence, PCP remains a relatively rare disease in non-HIV immunocompromised patients. This is confirmed by our regional incidence estimates for the study period. Importantly, they represent number of people with positive result for P. jirovecii by PCR. Hence, the incidence of clinical PCP was likely lower. Fillâtre et al. investigated incidence and risk furtherly; reporting incidence rates of PCP related to non-HIV predisposing conditions over two decades from France [39]. Their results demonstrate an apparent dissimilarity in the risk of contracting PCP within this heterogeneous population, presumably related to the underlying conditions and immunosuppressive treatment [39]. The prevalence of predisposing conditions influences how the risk translates into PCP occurrence. For instance, more patients with rheumatoid arthritis (RA) were assessed for PCP than patients with vasculitidies and connective tissue disorders combined in our cohort. This occurred in spite of RA patients’ inferior risk of contracting PCP compared to the latter group [39]. In developed countries, it is estimated that between 0.5 and 1% of the population suffers from RA [51], whilst vasculitidies and connective tissue diseases are much rarer conditions [52]. This may explain our observations.
To diagnose PCP accurately remains a challenge, even with modern technologies. Herein, positive P. jirovecii-PCR was the primary inclusion criterion. To study the epidemiological trends of P. jirovecii in Norway, we believe it was important to describe this population as a whole since all the patients were tested on clinical indication and had a high pre-test probability of PCP. In addition, they represented potential candidates for prophylaxis, mostly unidentified at the time, an important aspect to shed light on per se.
DIF microscopy represents an alternative method for case inclusion and is the current gold standard for PCP diagnosis [3]. However, its sensitivity is known to be unacceptably low, especially in populations dominated by HIV-negative individuals [5]. This seemed to be the case in our population as well. Also, false positives may result due to morphologically interchangeable fluorescent material. Lastly, the validity relies on experienced examinators. In light of this, real-time PCR represents a rapid and objective detection tool, though extrapolation of results is confronted by heterogeneity in PCR-target, respiratory samples, quality of DNA-extraction, host-characteristics, quantification methods and so on [53].
Herein, CT values from semiquantitative real-time PCR analysis and DIF results were collectively used to separate probable cases of PCP from those with presumed colonization. Of note, CT values were not reported in the laboratory information system during the study period. Therefore, CT values were collected retrospectively from the log of the PCR instruments. Unfortunately, some of the PCR-instruments had been replaced and discarded, and consequently CT values for samples run on those instruments were lost. Since the retrievability of CT values depended on which instrument the samples were analyzed, the missing pattern can be considered random and unrelated to patient characteristics. Analysis for beta-D-glucan was not available as a routine assay in our region; thus, such data were unavailable.
Retrospective PCP-classification was a secondary objective to see whether the general trends and characteristics in the overall population were representative. It was performed without considering heterogeneity in respiratory samples, which is a well-known issue [53]. Accordingly, a drawback of this approach is variability in microorganism gradients and volumes across respiratory samples, in addition to intra- and interindividual variability in host-pathogen biology. Collectively, these factors might have resulted in information bias. Yet, regardless of the exact number, the minority of patients with presumed colonization has important implications. Besides the possible role of colonization in chronic diseases, proposed interhuman transmission from individuals harboring P. jirovecii organisms is a concern [6].
To our knowledge, this is the largest study undertaken in a Nordic country regarding testing, epidemiology and clinical characteristics of patients assessed for PCP. However, the study design and methodology have several limitations and may provide grounds for biases. Firstly, the study population was sampled from only one region, thus, findings may not be generalizable to other areas. Secondly, the results of this report are based on retrospective case reviews of medical records, a method associated with certain limitations. Foremost, causal claims cannot be made, for instance regarding corticosteroid exposure and risk of contracting PCP. Also, the case review is a qualitative method. Hind-sight bias is likely to affect all retrospective case record reviews in particular [54]. Further, this design does not allow us to examine unavailable patient characteristics, and we rely on the information provided by the health personnel who treated the patients. Thirdly, we were unable to include all alive patients, which might have resulted in selection bias. In spite of the stigma associated with HIV/AIDS, we have little reason to believe that the request for active consent influenced the recruitment of HIV-positive individuals. In fact, the number of HIV-related PCP cases in our cohort were comparable to the estimated incidence in the region according to the national HIV/AIDS surveillance and health reports [7, 55]. Finally, our approach to identify eligible candidates, using positive PCR might have introduced bias as discussed above.