The radiographic hallmarks of alveolar disease are listed in Table 14.1. They include air bronchograms and ill-defined borders. Structures in the lungs are usually well outlined because of air in the alveoli; when the alveoli are filled with abnormality, there is nothing left to provide sharp margination. Alveolar disease may also have a characteristic distribution. For example, focal alveolar disease generally respects the anatomic barriers in the lungs, such as the pleural fissures, therefore demonstrating a lobar or segmental distribution (Fig. 14.1). Other characteristic distributions of more diffuse disease include the “butterfly” or “batwing” appearance (Fig. 14.2), predominantly occupying the central perihilar regions, typical of pulmonary edema, and a reverse batwing appearance of eosinophilic pneumonia. Areas of alveolar disease often coalesce together to form larger areas of alveolar disease (Fig. 14.3). Temporally speaking, alveolar disease processes often change quickly from day to day or even hour to hour.

When alveolar disease is present, the key diagnostic question to ask is whether it is acute or chronic. The answer to that question may be supplied by history; a patient may be able to tell you that he or she was fine until yesterday (acute) or has been symptomatic for 4 months (chronic). Furthermore, a patient who is seriously ill (e.g., an intensive care unit patient) virtually always has acute disease; an asymptomatic patient with significant CXR evidence of alveolar disease usually has chronic abnormality. Old radiographs are another avenue for determining acuteness or chronicity. However, they are sometimes misleading (Fig. 14.4).

Acute alveolar disease has an easy differential diagnosis, as listed in Table 14.2. It is blood or pus or water. Some radiologists like to add cells so they can include adult respiratory distress syndrome in the differential diagnosis; because I think of adult respiratory
distress syndrome along the lines of noncardiogenic pulmonary edema, I get to the same end point by a different route. Cells from lymphoma or bronchoalveolar cell carcinoma also fall under that category. Blood or pus or water means that we should consider pulmonary hemorrhage, pneumonia, and edema. Pulmonary hemorrhage is a relatively uncommon cause and is often (although not always) associated with hemoptysis or anemia. Practically speaking, acute alveolar disease is usually pneumonia or pulmonary edema. Finally, some would add a category for protein to cover diseases such as alveolar proteinosis or silicoproteinosis.

It used to be that pneumonia was relatively focal most of the time, and edema was usually diffuse. In the era of acquired immunodeficiency syndrome (Chapter 6), pneumonia can also be diffuse. Edema is often a result of congestive heart failure, which may produce associated cardiomegaly, but there are a number of causes of noncardiogenic edema (Table 10.11) (Fig. 14.5). Furthermore, edema may have unusual distributions (Table 10.12); it may be unilateral (Fig. 14.6) in the upper lobes (with pulmonary hypertension caused by chronic pulmonary emboli) or irregularly irregular (especially in chronic obstructive pulmonary disease). The bottom line is I make my best guess, but I leave it to the clinician caring for the patient to apply my radiographic input to the rest of the available clinical data and thereby to come up with the best diagnosis.

As for chronic alveolar disease, allow me to introduce you to the first of my mnemonics, as listed in Table 14.3. A few words about using mnemonics are in order. I do not want someone to regurgitate such a list at me, demonstrating a complete inability to think and
to synthesize information. I want this to be a safety net that allows you to remember the key entities to consider with a given pattern. As you run down the checklist mentally, you should be considering the applicability of each diagnosis to the patient and the pattern at hand. Is the patient a 70-year-old woman? Alveolar proteinosis is generally seen in men from ages 20 to 50, but sarcoidosis is sufficiently widely distributed in the population that almost any age and either gender is perfectly compatible. Is the disease predominantly basilar? That is good for lipoid pneumonia (Fig. 14.7) and desquamative interstitial pneumonitis (DIP), whereas tuberculosis (TB) and fungus are often in the upper lobes, especially with reactivation of previously dormant disease. Are there associated lymph nodes or pleural effusions? If so, lymphoma is a good diagnosis; if not, bronchoalveolar cell carcinoma is a much better bet.

A few more words about mnemonics come to mind. DIP is pretty uncommon; it is really there because “Dallas” is better known than “Allas.” When I have visited other programs
I have been told about other mnemonics (as a way to include eosinophilic pneumonia, for instance; I will have more to say about that diagnosis later in this chapter). The best mnemonic is the one you can remember. If you want to include more diagnoses on your checklist, suit yourself. My goal is to include the most common diagnoses, the ones I really do not want to forget. If the answer is not on my list of common entities, I can always consult Gamuts in Radiology (1) for more ideas. A variant mnemonic was “STAPLE” for chronic alveolar disease, which some of our residents in the 1980s learned at the Armed Forces Institute of Pathology (sarcoid, TB/fungus, alveolar cell carcinoma, proteinosis, lymphoma, eosinophilic pneumonia, I believe). To the “TB, fungus, Dallas” mnemonic, one of our residents added “Wins Big” for Wegener granulomatosis and bronchiolitis obliterans organizing pneumonia (more correctly now known as cryptogenic organizing pneumonia [COP]).

Turning to the patterns generally thought of as “interstitial,” we start with miliary nodules. They are small nodules (under 5 mm in size) that are very uniform in size and sharply defined. Even when they are extremely numerous they do not tend to become confluent. As I often tell residents, the result is the sense that with a pair of tweezers and a lot of patience, you could pick nodules off the radiograph for hours on end. The mnemonic for miliary lung disease is listed in Table 14.4.

Most tumors that metastasize to the lungs do so a little at a time—one metastasis in January, three in February, two in April, five in May—so that the radiographic picture is nodules varying in size. At the time of detection it would be unusual to have a large number of lesions all under 5 mm in size. The exception would be a very vascular primary neoplasm that could release a shower of metastases all at once. Miliary metastases are mainly seen with thyroid carcinomas; renal cell carcinoma is the next most common source (Fig. 14.8). However, I like to teach that in a given patient, the most likely primary neoplasm to produce a given finding is that patient’s neoplasm. Miliary nodules in a patient with duodenal leiomyosarcoma could indicate opportunistic infection, but if they prove to be malignant, they will probably be metastatic duodenal leiomyosarcoma (Fig. 14.9).

Usual interstitial pneumonitis (UIP) is on the list not because it tends to cause miliary nodules, but because with diffuse lung disease it is sometimes difficult to decide if the abnormality is the opaque areas (miliary nodules) or the lucent zones in between (honeycombing, see below). On HRCT, which provides better anatomic resolution, UIP is definitely not miliary. Small miliary nodules may also be seen in secondary hemosiderosis due to longstanding mitral valvular disease.

Calcified miliary nodules are a special category. They usually result from histoplasmosis or chickenpox pneumonia. Less commonly, they are seen with silicosis and hemosiderosis. At the far end of the spectrum is alveolar microlithiasis (Fig. 14.10), where tiny calcifications become so numerous they may simulate very opaque alveolar disease; microlithiasis occurs in families, although the inheritance pattern is difficult to discern. Affected patients are often surprisingly asymptomatic.

This is a pattern of clustered cystic spaces, ranging from 1 to 10 mm in size. Honeycombing tends to occur at the periphery of the lungs, although that may not be easy to appreciate with the CXR. It should also be noted that diseases that do not result in honeycombing at HRCT may nevertheless have a CXR appearance that looks like honeycombing (e.g., lymphangiomyomatosis, with cysts rather than honeycombing at HRCT). The mnemonic for honeycombing is given in Table 14.5. In deference to me, some of our former residents used to make this HIPS ARDS, adding amyloidosis (I was previously obsessed with that disease); because it almost never causes honeycombing, and I am no longer obsessed with amyloidosis, I do not prefer that version.

As previously mentioned, differential diagnosis requires more than a simple recollection of the applicable mnemonic. Disease distribution can be a very helpful piece of ancillary information. A quick glance at a lateral radiograph will convince you that the lungs are somewhat triangular; they are much larger at the bases than at the apices. As a result, disease that is truly uniformly distributed from apices to bases will look somewhat worse at the bases on the CXR. However, some diseases are virtually limited to the lung bases, whereas others will more prominently affect the upper lungs. Examples of upper lung diseases are listed in Table 14.6 and include TB, fungus, sarcoid, ankylosing spondylitis, primary adenocarcinoma of lung, typical emphysema, respiratory bronchiolitis (RB), eosinophilic granuloma (the last four associated with cigarette smoking), and silicosis (and most other inhalational diseases). Lower lung diseases are listed in Table 14.7 and include anything related to blood flow (such as metastases, pulmonary emboli, and most miliary nodules), collagen vascular diseases other than ankylosing spondylitis (RDS in the listed mnemonic), aspiration and lipoid pneumonia, asbestosis, UIP, and emphysema secondary to α₁-antitrypsin deficiency. Please note that this list includes a number of conditions that do not cause honeycombing. Still, honeycombing with upper lung predominance usually boils down to Langerhans cell histiocytosis (a.k.a. eosinophilic granuloma), sarcoidosis, or silicosis (Fig. 14.11); honeycombing with lower lung predominance is often collagen vascular disease, UIP, or asbestosis (Fig. 14.12).

In my experience most drug-induced lung disease is diffuse or at least not lower lung predominant, although amiodarone-induced lung toxicity tends to be basilar. This sometimes allows me to sort out the disease (such as rheumatoid arthritis) from the effect of its treatment (sometimes treated with methotrexate, which can also result in honeycomb lung). Among the collagen vascular diseases, scleroderma in particular is sometimes limited to the extreme lung bases (Fig. 14.13).

Gender can also be helpful in sorting among differential diagnostic possibilities. Langerhans cell histiocytosis (a.k.a. eosinophilic granuloma) used to have a significant male predilection (9:1); because it is a cigarette-smoking related disease, it no longer has a pronounced gender preference, likely due to the profound increase in women smoking over the last few decades. Sarcoid has something of a female predominance, but it is so ubiquitous that this turns out not to be helpful in differential diagnosis (many males are also affected). Diseases with continuing significant male predominance include rheumatoid lung (even though rheumatoid arthritis itself has a strong female predominance), ankylosing spondylitis, alveolar proteinosis, silicosis, and asbestosis. Female gender is more typical of lymphangiomyomatosis, dermatomyositis, and scleroderma.

This pattern is also referred to as reticulonodular. It describes a complex combination of lines, dots, and spaces and is something of a wastebasket term for thousands of entities that involve the interstitium of the lungs. Although any “interstitial” disease can manifest this way, I particularly try to remember four common entities, as listed in Table 14.8: lymphangitic metastases (Fig. 14.14), sarcoidosis, collagen vascular disease, and pneumoconiosis. It is worth remembering that from a pattern recognition standpoint, this is a less helpful descriptor than miliary or honeycombing.

This is a pattern of abnormality of the bronchi, characterized by bronchial enlargement or bronchial wall thickening as manifestations of bronchiectasis, discussed in greater detail in Chapter 15. Bronchial abnormality can be mistaken for honeycombing because enlarged bronchi may result in apparent cystic spaces. However, such bronchi may contain fluid, and they tend not to cluster to quite the same extent as do that honeycomb cysts. Furthermore, careful review of frontal and lateral radiographs will generally demonstrate that abnormal bronchi have a different appearance depending on whether they are viewed in short axis (where they often resemble cystic spaces) or long axis (where their appearance often simulates railroad or tram tracks). As treatment for cystic fibrosis improves and patients survive further and further into adulthood, cystic fibrosis is a more and more common cause of bronchial abnormality (Fig. 14.15).
Bronchiectasis may instead be postinflammatory or congenital (Kartagener syndrome). Bronchial wall thickening is also sometimes seen in asthma and chronic bronchitis.

This is a pattern characterized by nodules of varying size. Unlike alveolar nodules, the nodules considered here are sharply outlined. Unlike miliary nodules, they are not uniform in size and they often range well above the 5 mm limit for miliary nodules. The important entities in the differential diagnosis are metastases, metastases, metastases, metastases, and granulomas. This reinforces the fact that metastatic disease is numerically the most common cause and also the most important cause (based on effect on patient outcome) to remember. There are numerous other causes, including Wegener granulomatosis, hamartomas, and arteriovenous malformations; if you have a special interest in the less common causes, this is a good time to go to the Gamut book (Fig. 14.16). Cavitary nodules have a separate mnemonic, as listed in Table 14.9. This mnemonic can be used for a single cavity or for multiple cavities.

In the context of multiple cavities, cancer usually means squamous cell carcinoma metastases. Sarcoma metastases actually have a greater tendency to cavitate, but they are less common (Fig. 14.17). Apart from Wegener granulomatosis, rheumatoid nodules may also cavitate. Vascular reminds us that both bland and septic emboli may cavitate. Infection refers to bacterial lung abscesses and to infections like TB and fungus with a predilection for cavitation. A variety of traumatic lesions presents with cavitations (including laceration,

contusion, and pneumatocele). Although congenital abnormalities that cavitate (such as bronchogenic cyst and sequestration) are usually single lesions, they may on occasion be multiple (Box 14.1).

Calcified nodules usually indicate benign disease. Several such nodules are often seen in old TB or histoplasmosis. Hamartomas also calcify and are sometimes multiple, as in Carney triad (Fig. 14.18). However, metastases may also calcify. This is particularly likely after chemotherapy has been administered. De novo calcification (or ossification) of metastases is usually a manifestation of sarcomas, especially osteosarcoma (Fig. 14.19). Calcification of lesser magnitude is occasionally seen in adenocarcinoma metastases from the breast or gastrointestinal tract (Fig. 14.20).

HRCT is a sampling tool for evaluating the lung parenchyma, during which images are obtained at thin collimation, typically 1 or 1.5 mm (2). HRCT differs from most thoracic CT examinations in two ways. First, during most thoracic CT examinations the entire thoracic volume is captured during the CT acquisition, such as when looking for lung metastases. In contrast, HRCT images are obtained at intervals spaced throughout the lungs. Techniques range from an image every 1 cm to clusters of images at predefined anatomic levels, such as the aortic arch, the carina, and just above the diaphragm. Second, HRCT images are reconstructed using a high spatial frequency reconstruction algorithm that enhances edges, thereby emphasizing the borders of fine lines and nodules, similar (if not identical to) a bone algorithm. In contrast, most thoracic CT examinations use an algorithm for reconstruction that creates a smoother image that is more pleasing to the eye. HRCT images are inherently noisy and suffer from quantum mottle; this usually has little effect on the images reviewed on lung window settings. However, this noise is readily apparent when reviewing on soft tissue window settings. Most HRCT examinations are
performed at inspiration. Images are often obtained at expiration to evaluate for air trapping that indicates small airway disease. Prone images are useful to exclude lung disease in areas of dependent opacity that is commonly seen in the subpleural posterior portion of the lower lobes, created by atelectasis. This finding is more common with increasing age and in current or former smokers than nonsmokers (3).