Can You Get PJP Pneumonia if You're Immunocompetent? No, But You Can Get Diagnosed with it!

When we do a BAL (sometimes against our will, the so-called shut-up bronch: we do it so you will shut up about it), we very often send an "immunocompromise panel" whether the patient has immunocompromise or not. PJP PCR is included in this panel. If it comes back positive, does that mean the patient has PJP?

Firstly, the way to properly diagnose a disease is to build the case for it using natural/history, symptoms, course, risk factors, response to therapy, exclusion of alternative diagnoses. For PJP, a convincing classical picture (outside of HIV/AIDS) is subacute dyspnea on exertion, dry cough, fevers, ground glass opacities on CT (or interstitial opacities on CXR), often in the context of a recent tapering or abrupt discontinuation of corticosteroids. Why on tapering? Because the corticosteroids are suppressing the inflammation/immune response and symptoms of the PJP infection and they are unmasked during tapering or discontinuation. I think of it as the opposite of what we (used to) do in HIV, viz give steroids if the A-a gradient was above a certain level. Because when you started treating them, they got worse as the organisms died off, and steroids ameliorated the clinical worsening.

The fewer the features of the disease you have, the lower the pre-test probability. It is often so low that it falls below the test threshold of Pauker and Kassirer. Why don't I test you for PE right now? You might  have PE, how do you know you don't? I don't test you for PE right now because the probability of you having it without symptoms is so low that I do more harm with a CTPA than the benefit I expect to accrue from finding PE (in a probability-adjusted, EUT style analysis). And, false positives. If I select 1000 random people without symptoms (or with protean or nonspecific symptoms) I may find 10 "PEs". But 9 of them are so-called subsegmental PEs, which is itself a misnomer. It is a subsegmental filling defect in the pulmonary arterial tree on contrast-enhanced imaging. We make an inference that it's a PE. (It could also be an in situ thrombosis from an adjacent process, e.g, pneumonia, atelectasis.) And even if it is a "subsegmental PE", it's unclear whether we should treat it. If it is a false positive, we subject you to a treatment you don't need, won't benefit from, and may be harmed from. And, let's say 0.1% of these 1000 people we "screen" for PE get a contrast reaction. Using this policy of testing people with a negligible probability of PE, we have found 1 segmental PE which we treat (and which may have come to clinical attention later), we found 9 false positives and gave them anticoagulation they don't need, and we created 10 contrast reactions that needed treatment. Oh, yeah: and one dude wouldn't lie down in the scanner so he got sedated and had a respiratory arrest. All that to say that you need to build a case for the disease before you embark on the testing. (Kline et al used the method of Pauker and Kassirer to estimate that the proper test threshold is about 2% PE probability for CTPA.)

Similarly, getting reflexive/routine PJP PCR (or other highly sensitive molecular testing, especially for organisms that are ubiquitous in the environment) on bronchoscopy with BAL, in patients without a compatible presentation or a predisposition for it is a prescription for false positives, because PJP can be a commensal or an environmental contaminant. What's the false positive rate? Well, that depends on the pre-test probability! I can tell you the specificity is about 85%. That means that if you take 100 people who don't have PJP infection but you run the PCR, 15 of them will test positive, and that's exactly what we found when we did BAL on 24 young people with EVALI: 3 of them had positive PCR. Clearly, none of them had PJP, they wound up having EVALI, and we did not treat them (well, we gave them steroids!) and they all did fine on follow-up. You can go to Status Iatrogenicus and play with the Bayes' theorem calculator there to see what the probability of PJP is with varying levels of pre-test probability. Even with a pre-test probability of 10%, sensitivity of 99% and specificity of 85%, the post-test probability of PJP is 42% - you have more false positives than true positives. You may think that 42% is above the treatment threshold - the probability at which the benefits of treatment (in the patient or the population) exceed the harms of treating people without the disease - and that's fine. But I want to establish if we can determine that, in the absence of immunosuppression, the pre-test probability of PJP is basically zero, and all results can be considered false positive. Or, better: determine if, in the absence of immunosuppression, the probability of PJP is uniformly below the test threshold and testing should not be performed. That last statement would please Pauker and Kassirer, but the problem is people (other than yours truly, the Phlegmfighter) don't use thresholds so we will have to deal with false positives. Frequently. 

So, we have to go to the literature. The first widely referenced series of purported PJP in the literature was in the NEJM in 1991. In this series, BAL fluid was centrifuged and then analyzed with Gram-Weigert stan, with the number of "organisms" identified per slide quantified. This stain, like those described in the next series, is highly susceptible to background noise because it stains all cellular material: bacteria, host cells, and debris. So we have a specificity issue, but we shall proceed. Who were the patients?

• A 78-year-old woman with COPD, CHF is brought to hospital for minor trauma, found to have bibasilar opacities, left pleural effusion, WBC 19k, and cultures positive for H. influenza(!). She gets Ticarcillin for this, intubated, gets a BAL, and cytospin of BAL fluid shows 2-5 PJP organisms. She gets Bactrim, but also gets enterococcus bacteremia and dies of complications on hospital day 47. 
• A 66-year-old man has a 3 day history of high fever, malaise, headache, cervical LAD, and a RLL consolidation with air bronchograms. He was treated with cefuroxime and erythromycin, gets intubated, gets a BAL which shows - guess it - 2-5 PJP organisms on Gram-Weigert staining. He's treated with Bactrim, gets an open lung biopsy showing bronchiolitis obliterans, and is extubated on hospital day 49.
• a 73-year-old woman with DM and CHF comes in with a 3 week history of malaise, anorexia, nausea, vomiting, lethargy; 23# weight loss over the past year. CXR showed large heart, RLL infiltrate and pleural effusion. Sputum culture grew E. cloacae, Pseudomonas, Acinetobacter (!). They do BAL and she had 1(!) PJP per five to ten (5-10!) high power slides. They give her Bactrim, she gest worse and dies.
• 70-year-old man with DM and alcohol abuse admitted for hematemesis and dysphagia for 4 months, found to have esophageal carcinoma. Post-op after laryngectomy and other ectomies, he got a fever on the vent and had a RLL opacity and effusion and a tracheoesophageal fistula was also found. BAL showed 2-5 PJP per slide, they gave him Bactrim, he died. 
• 78-year-old woman with AS and AVR on anticoagulation hit her head and was admitted. BP was 280/100 and she was obtunded. She SDH evacuation by neurosurg, had a fever, CXR showed bilateral infiltrates (she aspirated, FFS, he thought to himself). They did a BAL and - you guessed it - 2-5 "organisms" per slide were seen. Bot Bactrim, improved.

These cases hardly deserve any additional comment. They are utterly unconvincing for PJP infection. No patient had a presentation consistent with PJP (at a time when I thought natural history of disease was and history taking and differential diagnosis got more faithful consideration than they do now). The results are false positives/noise/debris or commensals detected in the cytospun BAL fluid.

The second widely referenced series of purported PJP in immunocompetents was in CHEST journal in 1993. They report 5 patients in Spain in 1987 and 1988, nearly 40 years ago. Who were these patients and how were they diagnosed?

The methods for diagnosis of PJP (then PCP) in 1987-88 in Spain were Toluidine Blue O and Giemsa stains. Toluidine Blue O stains all fungal walls with 1,3-beta-D glucan, so the specificity hinges on the pathologist correctly identifying the morphology of the cysts (but not the trophozoites, which are not stained). Candida can look similar in size and appearance, and  stained proteinaceous debris or degenerating macrophages in alveolar exudate can occasionally mimic the size and outline of PJP cysts. Giemsa staining colors nuclear material of both trophozoites (visible as small, pleomorphic, reddish-purple dots [the nuclei] surrounded by a pale blue cytoplasm, measuring 1–5 µm) and cysts (cluster arrangements of up to eight distinct, dot-like internal nuclei [intracystic bodies]). The cyst nuclei stain dark red to purple. The surrounding cyst wall does not stain, creating a clear halo around the cluster of dots. But as with Toluidine Blue O, extracellular blood, platelets or small nuclear fragments can easily be mistaken for free trophozoites or individual intracystic bodies. In summary, the results of these stains cannot be taken as proof positive of PJP, and do not distinguish between colonization (as when samples are concentrated using centrifugation) and true infection. So we have to go back to the patients.

(If you don't feel like reading each case [which I have condensed], I will summarize at the end.)

• 39-year-old male presented with acute (24 hours) of fever, chest pain, and dyspnea, and interstitial opacities on CXR as well as a large exudative pleural effusion (not a feature of PJP). He was acutely ill and toxic. He got intubated and treated with: erythromycin, tobramycin, rifampin and isoniazid. He was not better on day 12 (wonder why!) so they performed "transthoracic needle aspiration" (which I think is a percutaneous lung biopsy) as well as bronchoscopy and the stains were positive for "cysts and trophozoites". Bactrim was started (two weeks in) and a week letter he started getting better and was extubated about a month after initial hospitalization.
• a 30-year-old man had 5 days (acute!) of fever, chills, mucous sputum, diffuse thoracic pain, myalgias, and dyspnea. He was acutely ill and toxic with a WBC of 21k. CXR showed diffuse alveolar infiltrate in mid lung zones. He got intubated and treated with erythromycin and tobramycin. Two day later he got BAL showing cysts and trophozoites on the stains. He got 21 days of Bactrim, repeat BAL was negative, then he got worse 5 days later, PJP detected again, got more Bactrim, got better but spent 50 days on the ventilator.
• 37-year-old woman with 9 days of fever, cough, malaise that did not respond to amoxicillin. Presented acutely toxic with WBC count of 28k. Got intubated, on day 5 she got BAL showing PJP on stain and also pseudomonas, so she got antibiotics changed to Bactrim and ceftazidime. She later was found to have mycoplasma pneumonia on complement fixation (high titer). 
• 37-year-old male smoker with chronic sputum production, presented with three weeks of cough, purulent sputum, right chest pain and fever. He was treated with erythromycin. They found PJP in BAL and thoracentesis. Got better with Bactrim.
• 55-year-old male smoker with COPD has two weeks of fever, chills, dyspnea, and diffuse sweating. Bilateral opacities, WBC 25k. BAL shows PJP along with tons of gram-negative rods. He got better with Bactrim.

In the cases in this series, the presentation is far more consistent with an acute bacterial respiratory illness, which was sometimes confirmed; antibiotic therapy may have been inadequate before starting Bactrim; the PJP which was allegedly detected may have represented colonization, or been a false positive pathological result based on problematic stains used 40 years ago. One or more of these factors is a far more likely explanation for this series than people without immune compromise presenting with toxic illness (today we would call them "septic") and having PJP. Furthermore, we have known for 40 years that the burden of PJP organisms depends on the predisposing disease, with HIV/AIDS patients having what my mentor, Peter Terry, called "sheets of organisms blanketing the alveoli." In these patients, stains such as DFA are often positive because the organism burden is so high - you don't even need to "reflex" to PCR.  By contrast, in patients on corticosteroids, the organism burden is low, and DFA is far less often positive. Thus, even on the most common cause of non-HIV immunosuppression predisposing to PJP, in 2026 we can't see it using DFA, but in 1987 they were seeing it with primitive stains in patients that weren't even immunosuppressed? How were there that many organisms? It is for this reason that I arrive at my conclusion that in 1987 in Spain they may have had a tyro slide tech or neophyte pathologist, or a septuagenarian with failing eyes who made a misidentification of PJP. 

Then they started looking for PJP everywhere and they and the pathologist(s), convinced they had stumbled upon an unrecognized phenomenon then went looking harder and harder for PJP (increased sensitivity), leading to a self-fulfilling prophesy of false positives. In any case, where are the subsequent case series of PJP in immunocompetent patients from Spain? Why does this phenomenon appear to be limited to that one small epoch in 1987-88?

This has gone on for long enough, both today for me, and for the past 40 years for the medical community. Open Evidence, which has been hallucinating like a pothead at a Bob Marley concert today, points out some recent studies from Asia where molecular evidence was taken as prima facie evidence of clinical infection, and some series of COVID-19 patients (almost) all of whom received lots of corticosteroids, but I'm going to pass on reviewing them in this post. I already debunked COVID-Associated Nonpathotenic Aspergillus Respiratory Disease (CANARD) elsewhere on this blog.

My conclusion is that true PJP infection in immunocompetent patients -- if it exists at all -- is so rare as to be non-existent, a rounding error. It is an extraordinary phenomenon, and as Carl Sagan was fond of saying "extraordinary claims require extraordinary evidence." The (admittedly incomplete) evidence I have reviewed here is far from extraordinary. Most PJP results (especially PCR)  in immunocompetent patients represents colonization or environmental contamination, unless proved otherwise by strong collateral evidence of clinical PJP disease. 

Unfortunately, the majority of these false positive results just get treated. Because seeing that PJP PCR Positive! in EPIC is like seeing your grandma naked: you can never unsee it.