Occasionally being confronted with a pneumonia-like illness that just doesn't want to get better, the intensivist needs to be ready to contemplate diagnostic possibilities outside of the normal range of differentials. This chapter presents an impossibly long list of unusual pneumonias, pneumonia-like illnesses, tumours collections and unrelated respiratory conditions which could produce the impression of pneumonia on an initial cursory inspection.
This scenario was explored in Question 21 from the second paper of 2017, which ended up being classified as an Infectious Diseases question. It was because when the trainees were asked to "list the factors that might be responsible for the slow resolution", the college answer suggests they were first expected to discuss infectious causes. the antibiotics were wrong, or they were insufficient, or the bug is resistant, or the patient is a malnourished chain-smoking alcoholic. In short, Question 21 was intended to discuss the approaches to something which is still clearly just recalcitrant community-acquired pneumonia of some sort, and which simply requires a more robust approach to antimicrobial interrogation and chemical warfare.
However, a part of the question deals with the apparently remote possibility that the patient's respiratory failure is due to something other than an infectious pathogen. Perhaps originally it was a CAP, but over two weeks of mechanical ventilation, it has developed into some sort of hideous noninfectious condition, which is described in obscure journals using sinister acronyms like COP and AIP. What the hell is this, one asks oneself. This difficult question probably falls into the territory of respiratory medicine and mechanical ventilation, because for diseases like this antimicrobial therapy is useless.
This chapter is broadly dedicated to discussing the characteristics and disease patterns of these noninfectious, postinfectious, inflammatory and noninflammatory not-pneumonias. The author shares with the reader the caveat that, him being an intensivist and not a respiratory physician, any discussion here suffers from a diminished authenticity, in the same way as a might affect a circus clown's article about the Australian Parliament - insofar as there is obvious overlap between the two areas of expertise, the writer obviously has minimal formal training in the subject matter, and this may degrade the credibility of their statements.
There is no one specific paper that could be used as a reference here, though the ATS/ERS consensus statement on classification of interstitial pneumonias (Demedts et al, 2002) and its update (2013) were illuminating, and Black (2016) and Lynch et al (1993) also have excellent articles in the press, both titled "Noninfectious mimics of community-acquired pneumonia". The rest had to be pieced together bit by bit from random sources listed in the references. In his effort to educate himself about this subject, the author has probably written too much, and the reader with finite time is referred to the summary table below, as this list of differentials is probably going to be enough for most people.
Causes of Non-resolving Pneumonia
Infectious, but not "typical"
- Intracellular organisms
- Legionella sp.
- Chlamydia and Chlamydophila (psittacosis, from birds)
- Coxiella burnetii (Q fever, from cattle)
- Francisella tularensis (tularemia, from rodents)
- Aspergillus sp.
- Cryptococcus (neoformans or gattii)
- Protozoan and parasitic
- Toxoplasma gondii
- Viral pneumonia
Not infectious but still inflammatory
- Inflamed by something:
- Drug toxicity pneumonitis
- Chemical pneumonitis
- Radiation pneumonitis
- Lymphangitis carcinomatosis
- Inflamed all on its own (i.e. autoimmune):
- Eosinophilic pneumonia
- Pulmonary vasculitis
- Pulmonary sarcoidosis
- Langerhan's lymphohistiocytosis
- Idiopathic interstitial pneumonias (IIPs):
- Usual interstitial pneumonia (UIP), likely IPF
- Cryptogenic organising pneumonia (COP)
- Acute interstitial pneumonia (AIP)
- Non-Specific Interstitial Pneumonia (NSIP)
- Rare IIPs (eg. LIP, DIP, and so on)
Not even an inflammatory state
- Bog-standard garden variety atelectasis
- Endobronchial mets, or invasive primary malignancy
- Pulmonary oedema
- Trapped loculated fluid in a fissure
- Pulmonary infarction
- Pulmonary contusion
- Alveolar haemorrhage
- Alveolar proteinosis
So, this whole time you thought you were treating some kind of community-acquired infection. But it did not go away with standard therapy, and now ten-fifteen days later the patient remains mechanically ventilated, with the consolidation unchanged, still laughing at you from the chest Xray. You even escalated the antibiotics, all to no avail. The sputum cultures remain negative, other than a scant smear of Candida albicans, presumably because that's the only thing that's left in the bronchi after you carpet-bombed the patient with meropenem. Could it be "atypical", you wonder.
Yes, reader, it absolutely could, and in fact "atypical" pathogens make up a significant proportion of pneumonia in the community. For example, some Australian data from 2008 suggests that around 15% of community-acquired pneumonia there was due to these "atypical" pathogens. The term "atypical" had originally referred to an unusual clinical presentation (for example, a predominance of constitutional symptoms over respiratory ones), but now has become colloquialised to the point where people usually just mean "the organism does not appear on gram staining". That means,
Characteristic features which might suggest "atypicalness" include:
The likelihood of this being missed and turning into a longstanding non-resolving pneumonia by accident is quite small, as most standard antibiotic recommendations for community-acquired pneumonia tend to contain a tetracycline or a macrolide which would be effective against these bugs. Of course, having no peptidoglycan in their wall, these organisms are entirely immune to the effects of β-lactam antibiotics or basically anything else that uses the bacterial cell wall as a target. This means that regimens targeting "hospital-acquired pneumonia", usually caused by gram-negatives, may not cover an atypical organism.
Fungus is on the menu wherever the patient is not perfectly immunocompetent, or where the practice is taking place in some tropical country (where the burden of fungal disease is generally higher). Davies (1994) and Shamim et al (2014) do an excellent job of covering this moist and spongy ground, which means we do not need to revisit it here, except to say that there are two main patterns: fungal pneumonia among people missing T-cell mediated immunity, and fungal pneumonia among those whose phagocytes are few or useless (eg. the neutropenic population of oncology patients). To organise the pathogens in this way probably has some merit:
Radiology of the pneumonia often raises fungus as a possibility because they tend to be rather creative in their lesions, with unique patterns of invasion (such as the cavitating abscesses and "halo" inflammatory changes of Aspergillus, for example). Some, like Pneumocystis, invade along interlobular septae and create ground glass infiltrates which spare the peripheries (preferring the hila). Rarely, yeasts and fungi will coexist with bacteria inside a boring standard lobar consolidation (like Candida in aspiration pneumonia). In short, there are some characteristic patterns to fungal pneumonia, and the reader comfortable with German is invited to review Heuel et al (2000), who have collected some of the most excellent images of any Lungenpilzerkrankung you could possibly think of.
Viruses form a large proportion of community lung pathogens, and it is absolutely plausible that the non-resolving culture-negative consolidation you are looking at is a viral phenomenon. Even before COVID, coronaviruses were on the list of possible pathogens, which also included:
There are about a hundred others, and memorising their names is pointless, as there are no pathogen-specific antiviral therapies at this stage, apart from COVID-specific antiinflammatory drug cocktails. Moreover, the patterns of disease are usually very similar from virus to virus. The best use for your memory would be to make a note of the differences between the clinical features of viral and bacterial pneumonia.
Viral pneumonia tends to:
It's not parasitic. If you're practising in the developed Western world and you're contemplating a parasitic or protozoan cause for your non-resolving consolidation, you can be fairly certain that you're wrong. Cool climate amoebas are not interested in your lung, and even the tropical protists need to be encouraged in by a carelessly lazy immune system leaving the door slightly ajar, or by some hideous systemic failure of basic sanitation. Toxoplasma is probably the only one that would be encountered in routine practice, provided that practice cares for a large number of patients with poorly controlled AIDS. Other organisms which fit this description include:
Additionally, multicellular parasites such as Strongyloides and Ascaris can produce pulmonary infections. This list comes from an excellent article by Martínez-Girón et al (2008) and the reader is directed there for further information with the warning that this is unlikely to ever become relevant for their exam preparation. In summary, abscess formation is a characteristic feature for many of these organisms, but many others can present as diffuse pneumonitis or interstitial nodular changes, making it difficult to present them in summary.
There is a whole lot of different ways drugs can do damage to the lung; for example, the manifestations of pulmonary toxicity can resemble any of the interstitial pneumonia variants listed below. Rather than trying to describe the massive spectrum of possible acute and chronic drug-induced lung injury, perhaps it would be better to just list some drugs which are known to produce it, and leave links to Rossi et al (2000) and Schwaiblmair et al (2012), whose excellent reviews will satisfy the needs of most casual readers. For the rest, there's pneumotox.com, which lists all possible and impossible pulmonary toxins and their common pathological reactions. A curated list is offered here:
The radiological findings in these scenarios are often as impressive as they are uninformative. Surveying sixty cases of mainly chemotherapy-induced pneumonitis, Akira et al (2002) observed basically every pattern you could possibly think of: unilateral or bilateral, patchy or focal, ground glass and dense consolidation - and often changing from one form to another over the course of the illness. Only a few characteristic features seemed to repeat across the cases:
Here is a representative CT demonstrating these features, in case any of the readers share the author's difficulty in visualising what "centrilobular" means. For a bit of contrast, this link will take you further down the page where you can see what "perilobular" inflammatory change looks like.
Gemcitabine-induced pneumonitis from Akira et al (2002). Note the white arrow pointing to thickened interlobular septa with nodules along them. The black arrow also points to a thickened septum. Within the lobules (regions bounded by the septa) the ground glass inflammatory change is distributed fairly evenly, one might even say centrally.
The common theme here seems to be the intrapulmonary biotransformation of a previously relatively benign substance into something directly toxic. In some circumstances, a co-factor is required (oxygen in the case of bleomycin, or radiation exposure for others). Reaction products from this biotransformation are either directly toxic, causing diffuse alveolar damage (DAD), or highly immunogenic, causing hypersensitivity pneumonitis. Valiant efforts to remove the offending substance via the normal lymphatics of the lung only results in spreading the inflammation along those interlobular drainage pathways, producing interlobular septal thickening.
Whereas drug-induced pneumonitis usually requires the biotransformation of an (probably already pretty toxic) agent into a (probably only slightly more toxic) metabolic product, chemical pneumonitis occurs from exposure to something that is straight-up corrosive, often as it is inhaled or ingested and then aspirated. Generally speaking, in order for an acute inhalation injury to produce pneumonitis, the inhaled agent needs to be either poorly water-soluble, or to be present in particles smaller than about 5 μm. From Gorguner & Akgun (2010), examples include:
Realistically, the history of inhaling something like chlorine is going to be the most exciting thing about a presentation of this sort, and you will never be confused about what is happening in the chest, but one can conceive of some scenarios where the list of differentials is broad and chemical pneumonitis is near the top. The classical case is the unconscious person found collapsed in some small enclosed shed-like space, with no history and with suspicious chemicals all around. Their pulmonary infiltrates: are they a normal pneumonia, or aspiration, or something they were breathing while they lay there in a coma?
Anyway. Two generic features can be commented on:
Only a few agents, eg. heavy metal fumes, oxides of nitrogen, and sulfur dioxide, tend to produce a respiratory insult lasting weeks and transitioning into something that looks like COP. By that stage, only supportive and empirical management will be possible. Thus, though theoretically after much digging one can arrive at the diagnosis retrospectively, in practice identifying the substance that caused the injury only has value in the context of the case report you are writing.
If the patient has a pneumonia-like illness which is failing to respond to treatment and there is a history of radiotherapy to the chest, radiation pneumonitis is definitely on the list of differentials. Chest radiotherapy could cause radiation pneumonitis where the cancer being treated was:
The treatment would have to had been curative in intent, i.e. a large dose of radiation would have been required. Generally speaking "palliative" doses designed to reduce the growth rate of a painful metastatic deposit are generally too low to produce a good-going pneumonitis. In general, anything less than 20 Gy will probably not produce a pneumonitis, and anything more than 40 Gy probably will. The only exception to this is a situation where the radiation is happening at the same time as some sort of chemotherapy, as cytotoxic agents can have a radiosensitising effect, reducing the dose that is necessary to cause this complication. From a list by Bledsoe et al (2017), drugs that increase the risk of radiation pneumonitis include:
This condition develops gradually in the weeks and months following radiotherapy and can progress slowly, taking months to resolve. It is therefore perfectly plausible that a doctor assessing the patient presenting with respiratory failure would not connect the radiotherapy (months ago) with the consolidation of today.
Once the history becomes available and raises your suspicion, there are a few other findings which can confirm the diagnosis. From Park et al (2000):
This weird thing comes in acute and chronic flavours, where the acute variety could easily act as a mimic for a community-acquired pneumonia and confound the intensivist by its failure to improve with antibiotics. Inflammatory markers (including even procalcitonin) could potentially be raised, and it may come with a fever and sputum production. Wheeze should also be expected: it is basically a hypersensitivity reaction, analogous to allergic urticaria, except localised to the lung. Well, to be frank, it does not need to be localised, and there are several variants of this condition which present as a part of some kind of horrible multisystem disease. From De Giacomi et al (2017):
Note how a few of these could be considered a completely understandable eosinophilic response to parasitic invasion. Some might say it is perfectly reasonable to eosinophilically inflame half a lung if you have a coven of Toxocara nesting in it. The resolution of the parasitic illness should then lead to the resolution of the eosinophilic pulmonary infiltrate, and this whole category should probably be grouped together with the parasitic pneumonias listed above. Then, there are also eosinophilic pneumonias which are without a clear infectious precipitant. Noninfectious causes of eosinophilic inflammation can include cigarette smoking, inhalation of flavoured vape or hydrocarbons, smoking meth or cocaine, fine airborne dust, and exposure to drugs (eg. DRESS).
All eosinophilic pneumonia-like processes tend to have some shared features. Unsurprisingly, the unexpected overabundance of eosinophils is one of the main characteristics.
Radiologically, both the systemic and the mainly pulmonary eosinophilic pneumonias tend to have
Obviously when your list of differentials includes parasite infestation and Churg-Strauss, you will have endless variability in the pattern of radiological appearance, and these loose observations are to be taken with a grain of salt. It would be fair to say that swarms of eosinophils in the BAL fluid would be worth a lot more than the CT, if you had to choose which investigation you were going to prioritise.
Sarcoidosis, a multi-system disease of cell-mediated immunity gone mad, owes its characteristic features to an out-of-control proliferation of macrophages, with granulomatous inflammation occurring in foci all around the body. It is thought to represent some sort of administrative overreach on the part of T-helper cells, though it would be better to leave the discussions of the pathophysiology to experts such as Baughman et al (2011). The pulmonary manifestations of sarcoidosis can be mistaken for a lung infection, considering especially that it often comes with a raised WCC and CRP, but there are a few features that help to identify it. From Criado et al (2010):
An autoimmune cause of lung disease is definitely something you need to contemplate in the case of a nonresolving pneumonia. There are realistically two main categories of this: autoimmune pneumonitis, and pulmonary vasculitis. It is the latter that we will discuss in more detail, as the diagnostic category of "autoimmune pneumonitis" is not completely established in the minds of experts, and one would not want to blurt it out baffoonishly in the middle of an ICU ward round. To be sure, most people would agree that there are clearly many autoimmune causes of inflammatory lung disease; but it is not clear where or how they fit into the existing (confusing) classification frameworks. For example, some scholars will hold that many of those "idiopathic" interstitial pneumonias like NSIP and AIP are autoimmune-mediated and we just haven't found the immunological mechanism which underlies their pathogenesis.
And the main reason we haven't found any of those mechanisms is because nobody has identified a convenient serological biomarker or immunofluorescence pattern which we could use to test for them. Not so for the pulmonary vasculitis category. We have an extensive selection of tests we could send to confirm their diagnosis, and a range of differentials which they help discriminate. Here is a representative list of lung-affecting connective tissue diseases and vasculites (vasculitides? Vasculituses?) from Lally & Spiera, 2015, with some tests added in parentheses to help the reader remember what tests to send:
Connective tissue diseases
This list is obviously far from complete (where's scleroderma, you cry), but these diseases were thrown into this unordered list not because this was some exhaustive attempt to represent each variation taxonomically, but because the point was to demonstrate that the range of possibilities is vast, and to prepare the reader for the knowledge that there is no unifying pattern or pathognomonic radiological finding they could poke at on the radiology monitor to say, "Aha! Vasculitis!". Each has very different features and each can present in a variety of ways, plus mostly the pneumonia-like disease due to a pulmonary vasculitis may look and act the same as the pneumonia-like disease due to another cause. Still, some very broad patterns can be described:
The reader interested in a deep dive into pulmonary vasculitis radiology is invited to review Castañer et al (2010), where an excellent collection of CT images serves to illustrate the diversity of presentations. One unifying feature is the tendency to produce haemoptysis and diffuse alveolar haemorrhage. Additionally, these are usually multisystem diseases and the pulmonary manifestations are usually coupled to some other findings (rash, mucosal ulcers, renal failure, arthritis and so forth), and those can serve as clues.
This becomes a serious differential wherever the unexplained non-resolving pneumonia is taking place alongside some kind of primary or metastatic pulmonary malignancy. The most common cancers of origin seem to be variations on adenocarcinoma, eg. breast, stomach and primary lung.
Apart from being associated with malignancy, the following features are often seen:
This is a group of several different lung diseases which the ERS and ATS have loosely classified according to their behaviour and the frequency of their appearance, rather than on any aetiological grounds.
UIP is not a synonym for IPF, but IPF usually presents as UIP. To be clear: idiopathic pulmonary fibrosis is an interstitial lung disease that presents initially as a clinical and radiological picture which is described as a "usual" interstitial pneumonia, a presentation which can also be associated with other conditions (eg. various connective tissue diseases or hypersensitivity pneumonitis). You might say that UIP causes lung damage which ends up degenerating into pulmonary fibrosis, and you'd have to call this fibrosis "idiopathic" if you don't have a clear reason for why it happened. And then we call it "usual" because it is the most common pattern of interstitial lung disease, all others being relatively more rare. The diagnosis is established when UIP pattern is identified on CT and no apparent cause presents itself.
Certain radiological features are common for UIP:
So if you thought that the multiple possible associated pathologies and lack of diagnosis make UIP sound very generic, just wait till you get a load of NSIP. It was originally designed as a wastebasket diagnosis for all the lung biopsies that did not return any identifying histological pattern. These days professional societies recognise idiopathic NSIP as a distinct diagnostic category; it is a pattern of interstitial pneumonia which occurs in a variety of scenarios, including connective tissue disease, drug-induced pneumonitis, hypersensitivity pneumonitis, and various others - but most commonly without any apparent cause.
The following radiological patterns help discriminate between NSIP and UIP:
Otherwise known as Hamman-Rich syndrome, this thing is a multilobar disease with bilateral involvement, characterised (like the rest of these) by the absence of an identifiable cause or predisposing condition. Mukhopadhyay et al (2012) also required DAD (diffuse alveolar damage) on biopsy, but let's face it - who's going to biopsy these people when their PEEP is 14 and their FiO2 is 70%. Ultimately, one will usually have only some history and radiological findings to work with, and AIP will often only be confirmed histologically upon postmortem.
Radiologically and clinically, it is basically ARDS, just with no obvious precipitant and no treatment options beyond "supportive care". Here's a particularly nasty one from Feuillet & Tazi (2011):
Characteristic features, which are all fairly nonspecific, consist of:
Apart from the lack of obvious cause and DAD on biopsy, AIP does not jump out and yell its diagnosis, nor is the history in any way helpful. It is just some sort of rapidly progressive disease occurring in people who are often previously well. In the event that a biopsy does somehow occur, the DAD you see could be histologically anything.; which is a convenient point to discuss this term.
Diffuse alveolar damage (DAD) is the final common pathway of severe lung injury, a nonspecific deadness of the alveoli that does not point to any particular aetiology, but which rather represents the end result of severe inflammation. The key histological feature of this thing is the appearance of hyaline membranes in the alveoli. These are proteinaceous deposits which appear inside the alveolar walls, and which are generally made up of proteinaceous debris and curdled surfactant. Kaarteenaho & Kinnula (2011) explain it better, and perhaps the details should be left to them; for the casual reader it will suffice to mention that this is a histological finding common to all sorts of idiopathic interstitial pneumonias as well as ARDS from infectious vasculitic or chemico-irritant causes.
"Organising pneumonia" is what you call a histopathological diagnosis of granulation tissue forming in the alveoli and distal airways, which begins as fibrin deposits but is slowly upgraded into proper collagen by the activity of fibroblasts. This is a fairly non-specific end stage phase of basically any severe inflammatory or infectious process, and so is not especially remarkable - basically anything that causes serious alveolar damage (ARDS, lobar pneumonia, pulmonary infarction, vasculitis) can ultimately go through an organising phase.
When this is happening in the lumens of bronchi, the granulation tissue can occlude the airway and obliterate the distal branches, hence the former name bronchiolitis obliterans organising pneumonia. And when it happens for no apparent reason, i.e. the aetiology is mysterious, you'd have to call it "cryptogenic" (meaning of obscure or uncertain origin). Why the Illuminati of respiratory nomenclature had settled on this term remains unclear, as they are clearly not shy about using the term "idiopathic" for essentially all of their interstitial lung diseases. Even more weirdly, they had discarded the older term "cryptogenic fibrosing alveolitis", and we now call that condition "idiopathic pulmonary fibrosis", not "cryptogenic pulmonary fibrosis", which raises the suspicion that all human activity is futile and meaningless.
Organising pneumonia (cryptogenic or otherwise) has specific radiological (CT) features that can point one to the diagnosis, although they are highly variable, and each alone is not especially diagnostic. It comes in so many different forms that ordinarily sober authors resort to words like "kaleidoscope" to describe its manifestations. This table was constructed using mainly the excellent papers by Tiralongo et al (2020) and Baque-Juston et al (2014):
|Peripheral consolidation with air bronchograms: this disease likes the pleura, and develops dense consolidation at the periphery of the lung. The large bronchi are usually not obliterated, and can still be seen as air bronchograms (images from Pinto-Lopes et al, 2016). This looks very similar to normal bacterial consolidation.|
|Perilobular pattern: the inflammation seems to encircle lobules, forming these "arcades" of inflammatory change that separate islands of relatively healthy-looking tissue. This image is from Roberton et al (2011)|
|Bronchocentric distribution: the consolidation seems to follow the bronchi (image from the ATS/ERS classification statement).|
|The "Reverse Halo" sign is, as the name suggests, the opposite of the normal "halo sign" you might see surrounding an aspergilloma. Instead of a nodule being surrounded in ground glass inflammatory change, you see a band of consolidation encircling a patch of lung with ground glass change. It is also occasionally referred to as the "atoll sign".|
|Band-like consolidation: linear consolidation is sometimes seen, which is obviously very distinct from the lobar or segmental consolidation of normal bacterial pneumonia (images from Tiralongo et al, 2020)|
This relatively rare interstitial inflammatory pneumonia is thought to be the earlier milder part of the same continuum as desquamative interstitial pneumonia (DIP), and possible also Langerhan's lymphohistiocytosis. It is always present in current smokers, and seems to be viewed as an adaptation to chronic smoke exposure, in that management consists of quitting. Ground glass opacity and centrilobular nodules are the non-specific radiological findings, and these are usually relatively mild, making this condition a much less likely candidate for causes of nonresolving pneumonia. If your chronically chainsmoking patient is not weaning from the ventilator, chances are RP-ILD is not the cause of their persisting x-ray infiltrates. Here, a CT from Attilli et al (2008) requires an arrow to point out the subtle centrilobular changes:
|Desquamative interstitial pneumonia (DIP) is basically the same condition, except you didn't stop smoking like you were politely asked to. It is just a more severe form, with more severe centrilobular ground glass change, and some cyst formation (as seen in this CT from the same paper). The arrow is pointing at a completely unrelated and irrelevant pneumothorax.|
Pulmonary Langerhans cell lymphohistiocytosis could theoretically be grouped with RP-ILD and DIP because it is almost always seen in smokers. Or, it could be lumped together with sarcoid, as it is also a disease characterised by granulomae and related to some mindless bureaucratic overactivity by a species of monocytes. However, it looks quite different radiologically, presenting as a reticular pattern on CT rather than nodules (and the reticulations, if you look at them closely, are a bunch of tiny little cysts). The disease evolves into ever larger cystic lesions which can cavitate and therefore resemble some kind of necrotising pneumonia, except there's usually no pleural effusion and - unlike with sarcoid - the mediastinal lymph nodes are usually not enlarged. To demonstrate how this could resemble a nasty infection, this CT from an excellent review by Tazi (2006) shows off some of the characteristic features.
This is a strange rare disease that is characterised by the lymphocyte-dominant alveolar septal infiltrate, considered by some to be on the same spectrum as MALT hyperplasia and lymphoma. There are a few "idiopathic" cases, but more often it appears in the presence of an autoimmune disease, most often Sjogren’s syndrome, SLE, rheumatoid arthritis or polymyositis. The most interesting radiological feature is the homogeneity of ground glass appearance and the presence of numerous (occasionally large) thin-walled cysts, which are not confined to the subpleural periphery as they would be in UIP. This scan is from Courtney, 2021 (p.1244):
Bonifazi et al (2017) describes this incredibly rare condition, and all the detail can be left in the body of that paper, as the normal CICM trainee does not need to know very much about it (other than it is incredibly rare and sometimes associated with chemotherapy). Fortunately, this condition does have memorable and distinctive CT findings, which should alert you to the possibility that the neutropenic patient you are looking at has something other than a predictable hospital-acquired pneumonia:
It is worth having here, mainly because there are few other differentials for a pneumonia-like disease preferentially affecting the upper lobes. A representative CT from the same paper illustrates the patent of involvement:
Atelectasis, the smart-sounding term to describe the incomplete inflation of a lung region, can look a little like pneumonia on x-ray or CT. It presents as a discrete opacity and - if produced by something tenacious like a tumour - can persist for a rather long period of time. Moreover because of the blockage the downstream bronchi and alveoli become a nidus for infection which may be refractory to a normal course of antibiotics (giving rise to persistently high fevers and raised inflammatory markers). Atelectasis can often present in a lobar distribution, giving rise to the impression that pneumonia has developed, but there are usually radiological features which can help discriminate between atelectasis and consolidation:
A brief scan of the literature might suggest that pneumonia is mistaken for malignancy more often than the other way around. Still, clearly one can present the impression of the other, and it works in both directions. Moreover malignancy can obstruct a proximal bronchus and cause a pneumonia by trapping infectious secretions, or by cavitating and turning into an abscess. The best article for this turned out to be "Neoplastic mimics of pneumonia" by Casey, from 1995.
The various discriminating radiological features may include:
It is not uncommon for ICU people to receive referrals for bilateral lower lobe pneumonia which resolve rapidly and dramatically under the influence of PEEP and diuresis. It is more uncommon for ICU patients (particularly the invasively ventilated ones) to have pulmonary oedema which remains mistaken for some kind of infectious process for a prolonged period. Itself a transient and dynamic phenomenon affecting the lungs symmetrically, pulmonary oedema is unlikely to give rise to a lasting pneumonia-like radiological picture. Scenarios which can fool even experts include the various causes of unilateral pulmonary oedema, such as:
Pleural fluid can occupy a discrete pocked of pleural space bounded by adhesions, or the potential space created by an interlobar fissure. Confusingly, this tends to happen after an actual bacterial pneumonia has settled down, and the exudative collection has failed to reabsorb. This differential is difficult to appreciate on plain films, but it is very clear from CT images: the collection has a characteristic lentiform appearance, sits astride a fissure, and is internally homogenous in density. Here, a couple of helpfully annotated images from Gowrinath et al (2018) demonstrate these features:
Pulmonary infarction can produce a wedge-shaped pleural-based opacity which is not obviously wedge-shaped or pleural-based on the chest Xray. The result is the mistaken impression of pneumonia (the patient will usually have had a PE, and would therefore be hypoxic and tachycardic, which is not unexpected in pneumonia). It can even cavitate, like an abscess. However, the unimpressive inflammatory markers and the failure of antibiotic response will eventually guide you to look for other causes. There are really just two main aetiologies: pulmonary emboli are the most common reason for a pulmonary infarction, and the other dominant cause is direct invasion by malignancy.
What does this look like? Balakrishnan et al (1989) commented on the most common CT appearance of this pathology, presumably using some sort of steampunk CT scanner with shiny rivets and brass valves. The following characteristics were noted in the majority of the cases they reviewed:
Here, some examples from Revel et al (2007) demonstrate the most common appearance of these lesions:
It is very hard to develop pulmonary contusions without having some sort of (usually severe) trauma, which means these CXR opacities usually don't get misinterpreted as pneumonia. However, it is not impossible. Scenarios present themselves to the imagination where the trauma is viewed as too trivial ("t'was only two cycles of CPR, surely that couldn't have" etc), or some sort of pneumonia seems more plausible (again following CPR, where it is almost mandatory to aspirate). So you start them on antibiotics, and the mimicry continues - as the contusions tend to get better over the course of a week or so, making you think that your antibiotics are working. On imaging, the lesions are often unhelpfully pneumonic-looking, evolving over hours following the injury from nothing, to ground glass, and to consolidation within a couple of days. Cohn & DuBose (2010) looked at some of the most common CT features, and concluded that the only real discriminating features were the appearance of subpleural sparing, and the presence of some kind of obvious trauma (overlying chest wall haematoma or fractured ribs)
Blood in the lung, whatever its origin, can produce an appearance of pneumonia, and could produce actual pneumonia with cavitation if there is enough clot to block large portions of a lung.
The possibilities are:
The main distinguishing feature which unites all causes of diffuse alveolar haemorrhage under one roof is the presence of alveolar microcirculatory injury. There are a lot of these causes, and a full list can be seen in Lara et al (2010), as well as here. Of the few different ways this could happen, most are autoimmune - variations on the theme of "capilliritis". These include:
And then, there are a group of diseases which are not related to the inflammation of the smallest vessels, which Lara et al group under the condescending term "bland pulmonary haemorrhage", implying these milquetoast pathologies are more timid and feeble than "proper" DAH. A selected few of these are:
Lastly, there's a category of causes that result from diffuse alveolar damage, i.e not just the capillaries but everything else is also being destroyed. From this, it follows that basically anything that causes DAD can cause diffuse alveolar haemorrhage. Lara et al list some choices from this broad range:
So: when do you think of alveolar haemorrhage, instead of pursuing any of the other hundred causes of diffuse pulmonary infiltrates? Well: one needs to be reminded that with most of the other possibilities, there should be no haemoptysis. Your various atypical and viral pneumonias, your NSIPs and your sarcoidosis - these are not diseases characterised by haemoptysis; whereas DAH presents with haemoptysis in 66% of cases. In the minority of cases who are not coughing out frank blood (too hypoxic to cough?) a bronchoalveolar lavage raises the suspicion by returning unusually bloodstained RBC-rich fluid.
Radiological findings are not especially specific. Cortese et al (2008) do a pretty good job of consolidating their surprisingly large experience of this syndrome. Here's a representative CT and a list of features:
The incidence of alveolar proteinosis being as low as 3.7 per million, the reader might wonder at the wisdom of including it in this list. Still, no list of differentials is complete without the zebras, because otherwise how would people be able to recognise your confident erudition. Pulmonary alveolar proteinosis is basically a disease of excess surfactant accumulation which leads to the buildup of surfactant protein in the alveoli, and it comes in congenital secondary and autoimmune flavours, which means theoretically there is a group of people out there who will develop this randomly in their adulthood, present to hospital, and end up in ICU, giving everyone ample opportunity to mistake their condition for something more common.
How do you end up with this problem, apart from being born with some sort of genetic abnormality? Well. You need to develop a deficit of GM-CSF, or an abnormal response to this cytokine by means of expressing a defective receptor. The former is most common, accounting for about 90% of all PAP cases, and results from GM-CSF antibodies. The latter is rarer and mainly known from case reports in patients with AML.
How exactly does this mimic pneumonia? It really doesn't, to be frank. It's more of a mimic of pulmonary oedema, as the chest Xray changes are perihilar and basal. Unlike APO, the presentation is nonspecific (shortness of breath) and insidious (weeks), with no cardiomegaly or pleural effusions. CT imaging is usually more meaningful, as there are some patterns that might make you think of PAP to exclusion of other diagnostic categories:
From the paper by Borie et al (2011), here is a CT slice to demonstrate that "geographic distribution":
No matter how characteristic this might be, the real diagnosis of PAP is achieved by bronchoalveolar lavage, which yields milky proteinaceous fluid, and by finding anti-GM-CSF antibodies. The BAL fluid can also be analysed to reveal characteristic findings (extraalveolar hyaline material forming large weird clumps, for example, or lamellar bodies).